diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30096 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.3125,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.1146,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.0472,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 1.0704,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.9849,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.0916,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 1.014,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 0.9877,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.0786,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-06,
+      "loss": 0.9832,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.0262,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.2443,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 1.2251,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.1242,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.1073,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-06,
+      "loss": 1.1928,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.1416,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 1.1521,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 1.13,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-06,
+      "loss": 1.096,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.1385,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.05e-05,
+      "loss": 1.2291,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.1128,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 1.2224,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.262,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-05,
+      "loss": 1.2659,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.1624,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.277,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.2417,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.45e-05,
+      "loss": 1.1581,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.0746,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.55e-05,
+      "loss": 1.0386,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.9746,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.65e-05,
+      "loss": 1.0132,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.0714,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.75e-05,
+      "loss": 1.0824,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 0.9959,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.85e-05,
+      "loss": 0.9433,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 0.9768,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 0.9967,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.8436,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.05e-05,
+      "loss": 0.8816,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 0.8731,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.15e-05,
+      "loss": 0.9255,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.8649,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.25e-05,
+      "loss": 0.9426,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.8175,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.35e-05,
+      "loss": 0.7347,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 0.8036,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.45e-05,
+      "loss": 0.9762,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 1.1561,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 1.0879,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.0087,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 1.1749,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.1362,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 1.0565,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.2068,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 1.1463,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9e-05,
+      "loss": 1.012,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.95e-05,
+      "loss": 1.2017,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 1.2827,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.05e-05,
+      "loss": 1.1984,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1e-05,
+      "loss": 1.2717,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.15e-05,
+      "loss": 1.1651,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.2319,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 1.2267,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3e-05,
+      "loss": 1.2394,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.35e-05,
+      "loss": 1.2165,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.2242,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.45e-05,
+      "loss": 1.2025,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 1.281,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.55e-05,
+      "loss": 1.2517,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 1.2373,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.65e-05,
+      "loss": 1.1754,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7e-05,
+      "loss": 1.2438,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.1875,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-05,
+      "loss": 1.164,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.85e-05,
+      "loss": 1.2704,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.2704,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 1.2702,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 1.2771,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.05e-05,
+      "loss": 1.2303,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-05,
+      "loss": 1.2977,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.15e-05,
+      "loss": 1.3385,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 1.2743,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.25e-05,
+      "loss": 1.3101,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3e-05,
+      "loss": 1.2947,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.35e-05,
+      "loss": 1.2981,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.4569,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 1.3231,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 1.3856,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.55e-05,
+      "loss": 1.3464,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.3768,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 1.3876,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7e-05,
+      "loss": 1.3031,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75e-05,
+      "loss": 1.3997,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 1.405,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.85e-05,
+      "loss": 1.3466,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9e-05,
+      "loss": 1.3712,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 1.3512,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 1.3658,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999999512004847e-05,
+      "loss": 1.2987,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999998048019406e-05,
+      "loss": 1.2244,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999995608043735e-05,
+      "loss": 1.4285,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999992192077935e-05,
+      "loss": 1.3133,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999998780012213e-05,
+      "loss": 1.272,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99999824321765e-05,
+      "loss": 1.3265,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999997608824124e-05,
+      "loss": 1.3047,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999996876831662e-05,
+      "loss": 1.3499,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99999604724029e-05,
+      "loss": 1.2014,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999951200500425e-05,
+      "loss": 1.3023,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999940952609544e-05,
+      "loss": 1.288,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999992972873067e-05,
+      "loss": 1.3201,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999991752886423e-05,
+      "loss": 1.2382,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999904353010704e-05,
+      "loss": 1.2993,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999890201170604e-05,
+      "loss": 1.1939,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999875073344495e-05,
+      "loss": 1.2277,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999985896953294e-05,
+      "loss": 1.1916,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999841889736595e-05,
+      "loss": 1.27,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999982383395612e-05,
+      "loss": 1.3169,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999980480219221e-05,
+      "loss": 1.2712,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999784794445616e-05,
+      "loss": 1.2955,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999763810717115e-05,
+      "loss": 1.3122,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999974185100754e-05,
+      "loss": 1.2541,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999718915317726e-05,
+      "loss": 1.3676,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999695003648586e-05,
+      "loss": 1.2701,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999967011600104e-05,
+      "loss": 1.2767,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999964425237607e-05,
+      "loss": 1.224,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999961741277469e-05,
+      "loss": 1.2942,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999958959719793e-05,
+      "loss": 1.1083,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999956080564688e-05,
+      "loss": 1.1076,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999953103812269e-05,
+      "loss": 1.1602,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999950029462649e-05,
+      "loss": 1.2749,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999946857515949e-05,
+      "loss": 1.1527,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999435879722934e-05,
+      "loss": 1.1516,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999402208318104e-05,
+      "loss": 1.0438,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999936756094629e-05,
+      "loss": 1.1487,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999331937608876e-05,
+      "loss": 1.1053,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999929533830723e-05,
+      "loss": 1.1995,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999925776304278e-05,
+      "loss": 1.1671,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999219211817007e-05,
+      "loss": 1.1868,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999917968463141e-05,
+      "loss": 1.1712,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999139181487536e-05,
+      "loss": 1.1735,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999909770238695e-05,
+      "loss": 1.1504,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99990552473313e-05,
+      "loss": 1.2488,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999901181632222e-05,
+      "loss": 1.2932,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999896740936141e-05,
+      "loss": 1.0362,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999892202645061e-05,
+      "loss": 1.0515,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9998875667591584e-05,
+      "loss": 1.0913,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999882833278615e-05,
+      "loss": 1.0558,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999878002203614e-05,
+      "loss": 1.0494,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999873073534347e-05,
+      "loss": 1.1887,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999868047271004e-05,
+      "loss": 1.1697,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999862923413781e-05,
+      "loss": 1.1046,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999857701962879e-05,
+      "loss": 1.3216,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9998523829185016e-05,
+      "loss": 1.1768,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999846966280857e-05,
+      "loss": 1.271,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999841452050156e-05,
+      "loss": 1.3155,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999835840226614e-05,
+      "loss": 1.2011,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99983013081045e-05,
+      "loss": 1.2125,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999824323801887e-05,
+      "loss": 1.3028,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999818419201153e-05,
+      "loss": 1.2961,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999812417008476e-05,
+      "loss": 1.2563,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9998063172240915e-05,
+      "loss": 1.2048,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999800119848238e-05,
+      "loss": 1.2396,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999793824881157e-05,
+      "loss": 1.1453,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999787432323095e-05,
+      "loss": 1.0687,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997809421743006e-05,
+      "loss": 1.0385,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999774354435027e-05,
+      "loss": 1.1103,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999767669105533e-05,
+      "loss": 1.0682,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999760886186078e-05,
+      "loss": 1.1515,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999754005676928e-05,
+      "loss": 1.0349,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99974702757835e-05,
+      "loss": 1.1206,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997399518906174e-05,
+      "loss": 1.0454,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999732778614007e-05,
+      "loss": 1.0928,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999725507748798e-05,
+      "loss": 1.0036,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997181392952744e-05,
+      "loss": 1.1049,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999710673253723e-05,
+      "loss": 1.1437,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999703109624437e-05,
+      "loss": 1.1608,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999695448407712e-05,
+      "loss": 1.2053,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999687689603845e-05,
+      "loss": 1.1976,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996798332131387e-05,
+      "loss": 1.2446,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996718792359035e-05,
+      "loss": 1.1968,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999663827672446e-05,
+      "loss": 1.1652,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996556785230816e-05,
+      "loss": 1.1945,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999647431788129e-05,
+      "loss": 1.1742,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99963908746791e-05,
+      "loss": 1.1395,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996306455627504e-05,
+      "loss": 1.2198,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99962210607298e-05,
+      "loss": 1.2284,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999613468998932e-05,
+      "loss": 1.2212,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996047343409426e-05,
+      "loss": 1.2161,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995959020993544e-05,
+      "loss": 1.2482,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999586972274511e-05,
+      "loss": 1.2679,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995779448667615e-05,
+      "loss": 1.1259,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999568819876458e-05,
+      "loss": 1.2925,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999559597303958e-05,
+      "loss": 1.2872,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995502771496194e-05,
+      "loss": 1.24,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999540859413808e-05,
+      "loss": 1.2745,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995313440968904e-05,
+      "loss": 1.2302,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999521731199238e-05,
+      "loss": 1.2712,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999512020721228e-05,
+      "loss": 1.2677,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995022126632366e-05,
+      "loss": 1.2125,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999492307025648e-05,
+      "loss": 1.3023,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999482303808849e-05,
+      "loss": 1.2498,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99947220301323e-05,
+      "loss": 1.3263,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999462004639186e-05,
+      "loss": 1.2984,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999451708687114e-05,
+      "loss": 1.3233,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994413151574167e-05,
+      "loss": 1.3471,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994308240504995e-05,
+      "loss": 1.2803,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994202353667726e-05,
+      "loss": 1.3067,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994095491066486e-05,
+      "loss": 1.3424,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999398765270544e-05,
+      "loss": 1.3686,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993878838588824e-05,
+      "loss": 1.3488,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993769048720865e-05,
+      "loss": 1.3197,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999365828310585e-05,
+      "loss": 1.4239,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999354654174811e-05,
+      "loss": 1.0858,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993433824652004e-05,
+      "loss": 0.9676,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999332013182193e-05,
+      "loss": 1.0083,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999320546326234e-05,
+      "loss": 0.9975,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999308981897769e-05,
+      "loss": 0.9651,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992973198972505e-05,
+      "loss": 1.0076,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999285560325134e-05,
+      "loss": 0.9894,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999273703181879e-05,
+      "loss": 0.9601,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999261748467947e-05,
+      "loss": 0.9618,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999249696183805e-05,
+      "loss": 0.9448,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999237546329925e-05,
+      "loss": 0.9899,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992252989067804e-05,
+      "loss": 1.1369,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999212953914848e-05,
+      "loss": 1.2827,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999200511354613e-05,
+      "loss": 1.1033,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991879712265574e-05,
+      "loss": 1.19,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999175333531173e-05,
+      "loss": 1.1627,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999162598268953e-05,
+      "loss": 1.2774,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991497654403936e-05,
+      "loss": 1.1138,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991368350459966e-05,
+      "loss": 1.1933,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999123807086267e-05,
+      "loss": 1.1432,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991106815617125e-05,
+      "loss": 1.2339,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990974584728465e-05,
+      "loss": 1.3457,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990841378201845e-05,
+      "loss": 1.3103,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999070719604248e-05,
+      "loss": 1.2708,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990572038255585e-05,
+      "loss": 1.1865,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990435904846447e-05,
+      "loss": 1.2488,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999029879582039e-05,
+      "loss": 1.2955,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9990160711182746e-05,
+      "loss": 1.2862,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999002165093893e-05,
+      "loss": 1.3345,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9989881615094356e-05,
+      "loss": 1.3306,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998974060365449e-05,
+      "loss": 1.2078,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998959861662484e-05,
+      "loss": 1.1694,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998945565401096e-05,
+      "loss": 1.1119,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998931171581841e-05,
+      "loss": 1.0752,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998916680205283e-05,
+      "loss": 1.1073,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9989020912719864e-05,
+      "loss": 1.1205,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998887404782521e-05,
+      "loss": 1.0997,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99887262073746e-05,
+      "loss": 1.1395,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998857739137381e-05,
+      "loss": 1.1473,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9988427599828656e-05,
+      "loss": 1.0821,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998827683274498e-05,
+      "loss": 1.213,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9988125090128654e-05,
+      "loss": 1.3067,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998797237198561e-05,
+      "loss": 1.4007,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998781867832183e-05,
+      "loss": 1.3861,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998766400914329e-05,
+      "loss": 1.4187,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9987508364456035e-05,
+      "loss": 1.3189,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9987351744266154e-05,
+      "loss": 1.301,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998719414857974e-05,
+      "loss": 1.3027,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998703557740295e-05,
+      "loss": 1.3597,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998687603074199e-05,
+      "loss": 1.3523,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9986715508603076e-05,
+      "loss": 1.4035,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998655401099248e-05,
+      "loss": 1.3845,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99863915379165e-05,
+      "loss": 1.3405,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998622808938149e-05,
+      "loss": 1.4893,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998606366539382e-05,
+      "loss": 1.4129,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998589826595991e-05,
+      "loss": 1.3122,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9985731891086226e-05,
+      "loss": 1.3535,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9985564540779254e-05,
+      "loss": 1.331,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998539621504553e-05,
+      "loss": 1.3503,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9985226913891624e-05,
+      "loss": 1.3604,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9985056637324154e-05,
+      "loss": 1.3168,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998488538534976e-05,
+      "loss": 1.3559,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9984713157975126e-05,
+      "loss": 1.3224,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9984539955206985e-05,
+      "loss": 1.3087,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998436577705209e-05,
+      "loss": 1.2445,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998419062351724e-05,
+      "loss": 1.2709,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998401449460927e-05,
+      "loss": 1.2329,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998383739033507e-05,
+      "loss": 1.2168,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9983659310701546e-05,
+      "loss": 1.2919,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998348025571565e-05,
+      "loss": 1.206,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9983300225384374e-05,
+      "loss": 1.317,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9983119219714734e-05,
+      "loss": 1.2121,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998293723871382e-05,
+      "loss": 1.2488,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998275428238872e-05,
+      "loss": 1.2819,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998257035074658e-05,
+      "loss": 1.2697,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998238544379458e-05,
+      "loss": 1.3164,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9982199561539945e-05,
+      "loss": 1.4072,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998201270398992e-05,
+      "loss": 1.2982,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9981824871151806e-05,
+      "loss": 1.2541,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998163606303294e-05,
+      "loss": 1.2653,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9981446279640685e-05,
+      "loss": 1.2688,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998125552098245e-05,
+      "loss": 1.2476,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99810637870657e-05,
+      "loss": 1.2691,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99808710778979e-05,
+      "loss": 1.3853,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998067739348657e-05,
+      "loss": 1.3542,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99804827338393e-05,
+      "loss": 1.292,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998028709896366e-05,
+      "loss": 1.3684,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99800904888673e-05,
+      "loss": 1.3418,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997989290355789e-05,
+      "loss": 1.3581,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997969434304315e-05,
+      "loss": 1.3272,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997949480733083e-05,
+      "loss": 1.3546,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997929429642872e-05,
+      "loss": 1.2012,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997909281034464e-05,
+      "loss": 1.2455,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997889034908647e-05,
+      "loss": 1.1294,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997868691266211e-05,
+      "loss": 1.1316,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997848250107949e-05,
+      "loss": 1.163,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997827711434661e-05,
+      "loss": 1.1664,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997807075247146e-05,
+      "loss": 1.1742,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9977863415462126e-05,
+      "loss": 1.1787,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997765510332668e-05,
+      "loss": 1.1992,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997744581607327e-05,
+      "loss": 1.1749,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997723555371007e-05,
+      "loss": 1.1981,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997702431624527e-05,
+      "loss": 1.19,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9976812103687124e-05,
+      "loss": 1.2141,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9976598916043926e-05,
+      "loss": 1.1981,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997638475332398e-05,
+      "loss": 1.1517,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9976169615535664e-05,
+      "loss": 1.2519,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997595350268738e-05,
+      "loss": 1.228,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997573641478754e-05,
+      "loss": 1.2044,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9975518351844644e-05,
+      "loss": 1.2637,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99752993138672e-05,
+      "loss": 1.1967,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997507930086375e-05,
+      "loss": 1.2091,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9974858312842884e-05,
+      "loss": 1.2068,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997463634981324e-05,
+      "loss": 1.0723,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9974413411783484e-05,
+      "loss": 1.1731,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99741894987623e-05,
+      "loss": 1.227,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997396461075845e-05,
+      "loss": 1.3079,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99737387477807e-05,
+      "loss": 1.1506,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997351190983788e-05,
+      "loss": 1.1488,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9973284096938836e-05,
+      "loss": 1.3112,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997305530909247e-05,
+      "loss": 1.1584,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9972825546307695e-05,
+      "loss": 1.2159,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99725948085935e-05,
+      "loss": 1.2753,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997236309595889e-05,
+      "loss": 1.1689,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997213040841291e-05,
+      "loss": 1.2298,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997189674596463e-05,
+      "loss": 1.1743,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99716621086232e-05,
+      "loss": 1.1873,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9971426496397755e-05,
+      "loss": 1.1721,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99711899092975e-05,
+      "loss": 1.2022,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997095234733168e-05,
+      "loss": 1.1572,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9970713810509566e-05,
+      "loss": 1.3747,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997047429884046e-05,
+      "loss": 1.3706,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997023381233373e-05,
+      "loss": 1.3134,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996999235099875e-05,
+      "loss": 1.3611,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9969749914844946e-05,
+      "loss": 1.3974,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9969506503881794e-05,
+      "loss": 1.3455,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9969262118118784e-05,
+      "loss": 1.3604,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996901675756547e-05,
+      "loss": 1.3646,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996877042223142e-05,
+      "loss": 1.3591,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9968523112126255e-05,
+      "loss": 1.2651,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996827482725963e-05,
+      "loss": 1.1524,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996802556764124e-05,
+      "loss": 1.1345,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996777533328081e-05,
+      "loss": 1.043,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996752412418811e-05,
+      "loss": 1.018,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996727194037296e-05,
+      "loss": 1.1952,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996701878184519e-05,
+      "loss": 1.1249,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9966764648614685e-05,
+      "loss": 1.0894,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996650954069137e-05,
+      "loss": 1.0692,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9966253458085204e-05,
+      "loss": 1.1392,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9965996400806183e-05,
+      "loss": 1.1962,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996573836886435e-05,
+      "loss": 1.1207,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996547936226976e-05,
+      "loss": 1.187,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9965219381032546e-05,
+      "loss": 1.2472,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996495842516285e-05,
+      "loss": 1.2274,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9964696494670855e-05,
+      "loss": 1.1627,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996443358956678e-05,
+      "loss": 1.1605,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996416970986091e-05,
+      "loss": 1.1567,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996390485556354e-05,
+      "loss": 1.2063,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996363902668499e-05,
+      "loss": 1.3295,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9963372223235655e-05,
+      "loss": 1.3295,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996310444522595e-05,
+      "loss": 1.2671,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9962835692666334e-05,
+      "loss": 1.3367,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996256596556728e-05,
+      "loss": 1.4021,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996229526393934e-05,
+      "loss": 1.2853,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996202358779307e-05,
+      "loss": 1.3226,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9961750937139075e-05,
+      "loss": 1.3567,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9961477311988e-05,
+      "loss": 1.2795,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996120271235054e-05,
+      "loss": 1.3075,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99609271382374e-05,
+      "loss": 1.339,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9960650589659344e-05,
+      "loss": 1.2649,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996037306662717e-05,
+      "loss": 1.2425,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996009456915171e-05,
+      "loss": 1.2826,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995981509724383e-05,
+      "loss": 1.2828,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995953465091445e-05,
+      "loss": 1.2563,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995925323017452e-05,
+      "loss": 1.2285,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995897083503502e-05,
+      "loss": 1.3101,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9958687465506984e-05,
+      "loss": 1.2908,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995840312160146e-05,
+      "loss": 1.2883,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995811780332955e-05,
+      "loss": 1.3615,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995783151070241e-05,
+      "loss": 1.3429,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99575442437312e-05,
+      "loss": 1.3586,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9957256002427146e-05,
+      "loss": 1.3052,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995696678680148e-05,
+      "loss": 1.3236,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9956676596865524e-05,
+      "loss": 1.3391,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995638543263059e-05,
+      "loss": 1.4221,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995609329410804e-05,
+      "loss": 1.3161,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9955800181309287e-05,
+      "loss": 1.2438,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995550609424577e-05,
+      "loss": 1.0806,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995521103292897e-05,
+      "loss": 1.1398,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9954914997370416e-05,
+      "loss": 1.0395,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995461798758165e-05,
+      "loss": 1.1382,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9954320003574274e-05,
+      "loss": 1.1282,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995402104535993e-05,
+      "loss": 1.0381,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995372111295027e-05,
+      "loss": 1.0777,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995342020635702e-05,
+      "loss": 1.0258,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995311832559192e-05,
+      "loss": 1.02,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995281547066676e-05,
+      "loss": 1.0299,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9952511641593356e-05,
+      "loss": 1.1105,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995220683838358e-05,
+      "loss": 1.1672,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995190106104931e-05,
+      "loss": 1.1997,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995159430960251e-05,
+      "loss": 1.2854,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9951286584055135e-05,
+      "loss": 1.2168,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995097788441921e-05,
+      "loss": 1.1905,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995066821070679e-05,
+      "loss": 1.1791,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9950357562929956e-05,
+      "loss": 1.1921,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995004594110083e-05,
+      "loss": 1.2163,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99497333452316e-05,
+      "loss": 1.1762,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994941977533444e-05,
+      "loss": 1.0356,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994910523142161e-05,
+      "loss": 0.8181,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994878971350539e-05,
+      "loss": 0.8708,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9948473221598094e-05,
+      "loss": 0.8966,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9948155755712076e-05,
+      "loss": 0.9546,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994783731585973e-05,
+      "loss": 0.9116,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9947517902053486e-05,
+      "loss": 1.0574,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994719751430582e-05,
+      "loss": 0.9325,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994687615262924e-05,
+      "loss": 0.9567,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994655381703629e-05,
+      "loss": 0.9559,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994623050753955e-05,
+      "loss": 0.913,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994590622415164e-05,
+      "loss": 0.9698,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9945580966885226e-05,
+      "loss": 0.9541,
+      "step": 434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9945254735753e-05,
+      "loss": 1.2418,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9944927530767715e-05,
+      "loss": 1.2641,
+      "step": 436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994459935194212e-05,
+      "loss": 1.2016,
+      "step": 437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9944270199289045e-05,
+      "loss": 1.2326,
+      "step": 438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994394007282133e-05,
+      "loss": 1.2474,
+      "step": 439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994360897255187e-05,
+      "loss": 1.1549,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994327689849359e-05,
+      "loss": 1.2052,
+      "step": 441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9942943850659444e-05,
+      "loss": 1.1919,
+      "step": 442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994260982906245e-05,
+      "loss": 1.1885,
+      "step": 443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994227483371564e-05,
+      "loss": 1.23,
+      "step": 444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994193886463209e-05,
+      "loss": 1.416,
+      "step": 445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994160192182491e-05,
+      "loss": 1.3013,
+      "step": 446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9941264005307275e-05,
+      "loss": 1.3403,
+      "step": 447
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994092511509236e-05,
+      "loss": 1.3278,
+      "step": 448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9940585251193394e-05,
+      "loss": 1.2798,
+      "step": 449
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.994024441362366e-05,
+      "loss": 1.2665,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993990260239645e-05,
+      "loss": 1.2912,
+      "step": 451
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993955981752512e-05,
+      "loss": 1.3984,
+      "step": 452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993921605902304e-05,
+      "loss": 1.2812,
+      "step": 453
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9938871326903636e-05,
+      "loss": 1.3073,
+      "step": 454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9938525621180365e-05,
+      "loss": 1.2601,
+      "step": 455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993817894186672e-05,
+      "loss": 1.3222,
+      "step": 456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993783128897625e-05,
+      "loss": 1.2266,
+      "step": 457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9937482662522505e-05,
+      "loss": 1.2113,
+      "step": 458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993713306251911e-05,
+      "loss": 1.1768,
+      "step": 459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993678248897972e-05,
+      "loss": 1.2558,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9936430941918e-05,
+      "loss": 1.2039,
+      "step": 461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993607842134769e-05,
+      "loss": 1.2739,
+      "step": 462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9935724927282545e-05,
+      "loss": 1.1909,
+      "step": 463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9935370459736376e-05,
+      "loss": 1.1711,
+      "step": 464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9935015018723e-05,
+      "loss": 1.1973,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993465860425631e-05,
+      "loss": 1.2701,
+      "step": 466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993430121635023e-05,
+      "loss": 1.3617,
+      "step": 467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993394285501869e-05,
+      "loss": 1.283,
+      "step": 468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993358352027568e-05,
+      "loss": 1.3724,
+      "step": 469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9933223212135256e-05,
+      "loss": 1.2952,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9932861930611454e-05,
+      "loss": 1.2521,
+      "step": 471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993249967571839e-05,
+      "loss": 1.2625,
+      "step": 472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9932136447470214e-05,
+      "loss": 1.2594,
+      "step": 473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993177224588109e-05,
+      "loss": 1.2996,
+      "step": 474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9931407070965254e-05,
+      "loss": 1.3454,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993104092273695e-05,
+      "loss": 1.1706,
+      "step": 476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993067380121048e-05,
+      "loss": 1.2501,
+      "step": 477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.993030570640017e-05,
+      "loss": 1.2399,
+      "step": 478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9929936638320386e-05,
+      "loss": 1.1735,
+      "step": 479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9929566596985546e-05,
+      "loss": 1.1707,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992919558241009e-05,
+      "loss": 1.2245,
+      "step": 481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992882359460851e-05,
+      "loss": 1.2995,
+      "step": 482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992845063359533e-05,
+      "loss": 1.2049,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992807669938509e-05,
+      "loss": 1.2618,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992770179199241e-05,
+      "loss": 1.1373,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992732591143191e-05,
+      "loss": 1.2549,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992694905771828e-05,
+      "loss": 1.1871,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992657123086623e-05,
+      "loss": 1.1477,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9926192430890486e-05,
+      "loss": 1.2716,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992581265780587e-05,
+      "loss": 1.1871,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992543191162719e-05,
+      "loss": 1.2347,
+      "step": 491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992505019236932e-05,
+      "loss": 1.2102,
+      "step": 492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992466750004714e-05,
+      "loss": 1.1942,
+      "step": 493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992428383467562e-05,
+      "loss": 1.2707,
+      "step": 494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992389919626972e-05,
+      "loss": 1.3444,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992351358484445e-05,
+      "loss": 1.3297,
+      "step": 496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992312700041488e-05,
+      "loss": 1.336,
+      "step": 497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9922739442996106e-05,
+      "loss": 1.3677,
+      "step": 498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9922350912603236e-05,
+      "loss": 1.2832,
+      "step": 499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9921961409251464e-05,
+      "loss": 1.344,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 1.1721256971359253,
+      "eval_runtime": 111.9125,
+      "eval_samples_per_second": 34.143,
+      "eval_steps_per_second": 0.357,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992157093295597e-05,
+      "loss": 1.3127,
+      "step": 501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9921179483732014e-05,
+      "loss": 1.3083,
+      "step": 502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9920787061594877e-05,
+      "loss": 1.2754,
+      "step": 503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992039366655987e-05,
+      "loss": 1.0631,
+      "step": 504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9919999298642364e-05,
+      "loss": 1.1014,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991960395785774e-05,
+      "loss": 1.1502,
+      "step": 506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991920764422145e-05,
+      "loss": 1.1538,
+      "step": 507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991881035774896e-05,
+      "loss": 1.2661,
+      "step": 508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991841209845577e-05,
+      "loss": 1.1714,
+      "step": 509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9918012866357426e-05,
+      "loss": 1.113,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9917612661469534e-05,
+      "loss": 1.1675,
+      "step": 511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99172114838077e-05,
+      "loss": 1.2225,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9916809333387595e-05,
+      "loss": 1.0616,
+      "step": 513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991640621022491e-05,
+      "loss": 1.1297,
+      "step": 514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9916002114335394e-05,
+      "loss": 1.0662,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991559704573482e-05,
+      "loss": 1.0387,
+      "step": 516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991519100443899e-05,
+      "loss": 0.9992,
+      "step": 517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9914783990463766e-05,
+      "loss": 1.1527,
+      "step": 518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991437600382504e-05,
+      "loss": 1.0726,
+      "step": 519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9913967044538734e-05,
+      "loss": 1.0565,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9913557112620815e-05,
+      "loss": 1.0678,
+      "step": 521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991314620808729e-05,
+      "loss": 1.0333,
+      "step": 522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991273433095419e-05,
+      "loss": 1.1838,
+      "step": 523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991232148123761e-05,
+      "loss": 1.0426,
+      "step": 524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9911907658953664e-05,
+      "loss": 1.1439,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991149286411849e-05,
+      "loss": 1.1514,
+      "step": 526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99110770967483e-05,
+      "loss": 1.1852,
+      "step": 527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991066035685932e-05,
+      "loss": 1.1706,
+      "step": 528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991024264446782e-05,
+      "loss": 1.162,
+      "step": 529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9909823959590105e-05,
+      "loss": 1.2691,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990940430224252e-05,
+      "loss": 1.207,
+      "step": 531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990898367244146e-05,
+      "loss": 1.1649,
+      "step": 532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9908562070203325e-05,
+      "loss": 1.1615,
+      "step": 533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990813949554458e-05,
+      "loss": 1.0595,
+      "step": 534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990771594848174e-05,
+      "loss": 1.1037,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990729142903132e-05,
+      "loss": 1.0993,
+      "step": 536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990686593720991e-05,
+      "loss": 1.1023,
+      "step": 537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990643947303411e-05,
+      "loss": 1.1287,
+      "step": 538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990601203652056e-05,
+      "loss": 1.1519,
+      "step": 539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990558362768597e-05,
+      "loss": 1.1285,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990515424654705e-05,
+      "loss": 1.1583,
+      "step": 541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9904723893120565e-05,
+      "loss": 1.1438,
+      "step": 542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990429256742331e-05,
+      "loss": 1.115,
+      "step": 543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9903860269472133e-05,
+      "loss": 0.9055,
+      "step": 544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990342699928392e-05,
+      "loss": 1.0113,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990299275687556e-05,
+      "loss": 1.009,
+      "step": 546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990255754226403e-05,
+      "loss": 1.1286,
+      "step": 547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9902121355466296e-05,
+      "loss": 1.0532,
+      "step": 548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990168419649941e-05,
+      "loss": 1.0614,
+      "step": 549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990124606538042e-05,
+      "loss": 1.0379,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9900806962126454e-05,
+      "loss": 1.0158,
+      "step": 551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9900366886754626e-05,
+      "loss": 0.9375,
+      "step": 552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989992583928214e-05,
+      "loss": 1.1555,
+      "step": 553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98994838197262e-05,
+      "loss": 1.2452,
+      "step": 554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989904082810407e-05,
+      "loss": 1.1713,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9898596864433036e-05,
+      "loss": 1.2498,
+      "step": 556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989815192873044e-05,
+      "loss": 1.1766,
+      "step": 557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9897706021013645e-05,
+      "loss": 1.1754,
+      "step": 558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989725914130006e-05,
+      "loss": 1.2669,
+      "step": 559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9896811289607134e-05,
+      "loss": 1.2549,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989636246595235e-05,
+      "loss": 1.1969,
+      "step": 561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989591267035323e-05,
+      "loss": 1.2214,
+      "step": 562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9895461902827336e-05,
+      "loss": 1.1445,
+      "step": 563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989501016339225e-05,
+      "loss": 1.204,
+      "step": 564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9894557452065635e-05,
+      "loss": 1.2169,
+      "step": 565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9894103768865144e-05,
+      "loss": 1.2044,
+      "step": 566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989364911380849e-05,
+      "loss": 1.2872,
+      "step": 567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989319348691343e-05,
+      "loss": 1.15,
+      "step": 568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989273688819776e-05,
+      "loss": 1.1521,
+      "step": 569
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989227931767929e-05,
+      "loss": 1.1215,
+      "step": 570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989182077537589e-05,
+      "loss": 1.1808,
+      "step": 571
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9891361261305444e-05,
+      "loss": 1.2878,
+      "step": 572
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989090077548592e-05,
+      "loss": 1.2484,
+      "step": 573
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989043931793528e-05,
+      "loss": 1.276,
+      "step": 574
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988997688867153e-05,
+      "loss": 1.3036,
+      "step": 575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988951348771275e-05,
+      "loss": 1.3421,
+      "step": 576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9889049115077005e-05,
+      "loss": 1.2319,
+      "step": 577
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9888583770782434e-05,
+      "loss": 1.1819,
+      "step": 578
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988811745484721e-05,
+      "loss": 1.2912,
+      "step": 579
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9887650167289525e-05,
+      "loss": 1.214,
+      "step": 580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9887181908127625e-05,
+      "loss": 1.2634,
+      "step": 581
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98867126773798e-05,
+      "loss": 1.2557,
+      "step": 582
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988624247506436e-05,
+      "loss": 1.2878,
+      "step": 583
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988577130119966e-05,
+      "loss": 1.1936,
+      "step": 584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9885299155804107e-05,
+      "loss": 1.2709,
+      "step": 585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988482603889612e-05,
+      "loss": 1.2348,
+      "step": 586
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9884351950494165e-05,
+      "loss": 1.2414,
+      "step": 587
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9883876890616776e-05,
+      "loss": 1.234,
+      "step": 588
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988340085928247e-05,
+      "loss": 1.2341,
+      "step": 589
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988292385650985e-05,
+      "loss": 1.2722,
+      "step": 590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988244588231753e-05,
+      "loss": 1.2007,
+      "step": 591
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988196693672417e-05,
+      "loss": 1.2504,
+      "step": 592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988148701974847e-05,
+      "loss": 1.3906,
+      "step": 593
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988100613140916e-05,
+      "loss": 1.2899,
+      "step": 594
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988052427172503e-05,
+      "loss": 1.2606,
+      "step": 595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988004144071487e-05,
+      "loss": 1.2915,
+      "step": 596
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987955763839754e-05,
+      "loss": 1.3118,
+      "step": 597
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987907286479193e-05,
+      "loss": 1.2797,
+      "step": 598
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9878587119916967e-05,
+      "loss": 1.2317,
+      "step": 599
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98781004037916e-05,
+      "loss": 1.2843,
+      "step": 600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987761271643485e-05,
+      "loss": 1.2987,
+      "step": 601
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987712405786574e-05,
+      "loss": 1.3934,
+      "step": 602
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9876634428103356e-05,
+      "loss": 1.4066,
+      "step": 603
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987614382716681e-05,
+      "loss": 1.3496,
+      "step": 604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9875652255075254e-05,
+      "loss": 1.3747,
+      "step": 605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9875159711847886e-05,
+      "loss": 1.3767,
+      "step": 606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9874666197503925e-05,
+      "loss": 1.3404,
+      "step": 607
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9874171712062636e-05,
+      "loss": 1.2873,
+      "step": 608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987367625554333e-05,
+      "loss": 1.3467,
+      "step": 609
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987317982796535e-05,
+      "loss": 1.3669,
+      "step": 610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987268242934808e-05,
+      "loss": 1.1034,
+      "step": 611
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9872184059710926e-05,
+      "loss": 1.2153,
+      "step": 612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9871684719073355e-05,
+      "loss": 1.1338,
+      "step": 613
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987118440745485e-05,
+      "loss": 1.1794,
+      "step": 614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9870683124874954e-05,
+      "loss": 1.1175,
+      "step": 615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9870180871353236e-05,
+      "loss": 1.0324,
+      "step": 616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9869677646909296e-05,
+      "loss": 1.1284,
+      "step": 617
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9869173451562784e-05,
+      "loss": 1.103,
+      "step": 618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986866828533339e-05,
+      "loss": 1.1583,
+      "step": 619
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986816214824082e-05,
+      "loss": 1.1096,
+      "step": 620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986765504030485e-05,
+      "loss": 1.175,
+      "step": 621
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986714696154527e-05,
+      "loss": 0.9785,
+      "step": 622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986663791198191e-05,
+      "loss": 0.9405,
+      "step": 623
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9866127891634654e-05,
+      "loss": 0.8736,
+      "step": 624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9865616900523406e-05,
+      "loss": 1.0566,
+      "step": 625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986510493866812e-05,
+      "loss": 0.9558,
+      "step": 626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9864592006088766e-05,
+      "loss": 0.9553,
+      "step": 627
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986407810280539e-05,
+      "loss": 0.9916,
+      "step": 628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986356322883805e-05,
+      "loss": 1.027,
+      "step": 629
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9863047384206835e-05,
+      "loss": 0.8968,
+      "step": 630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98625305689319e-05,
+      "loss": 0.9742,
+      "step": 631
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9862012783033405e-05,
+      "loss": 0.9514,
+      "step": 632
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986149402653157e-05,
+      "loss": 0.9766,
+      "step": 633
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986097429944666e-05,
+      "loss": 1.0189,
+      "step": 634
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986045360179894e-05,
+      "loss": 1.1323,
+      "step": 635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985993193360876e-05,
+      "loss": 1.1229,
+      "step": 636
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9859409294896476e-05,
+      "loss": 1.1254,
+      "step": 637
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9858885685682485e-05,
+      "loss": 1.1691,
+      "step": 638
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985836110598725e-05,
+      "loss": 1.136,
+      "step": 639
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985783555583123e-05,
+      "loss": 1.2289,
+      "step": 640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9857309035234943e-05,
+      "loss": 1.0775,
+      "step": 641
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9856781544218955e-05,
+      "loss": 1.2211,
+      "step": 642
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985625308280386e-05,
+      "loss": 1.3116,
+      "step": 643
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985572365101028e-05,
+      "loss": 1.3251,
+      "step": 644
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985519324885888e-05,
+      "loss": 1.2784,
+      "step": 645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9854661876370376e-05,
+      "loss": 1.3032,
+      "step": 646
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9854129533565515e-05,
+      "loss": 1.2261,
+      "step": 647
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9853596220465074e-05,
+      "loss": 1.222,
+      "step": 648
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985306193708987e-05,
+      "loss": 1.3061,
+      "step": 649
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985252668346076e-05,
+      "loss": 1.3138,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9851990459598657e-05,
+      "loss": 1.21,
+      "step": 651
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985145326552448e-05,
+      "loss": 1.0456,
+      "step": 652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98509151012592e-05,
+      "loss": 1.0107,
+      "step": 653
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9850375966823834e-05,
+      "loss": 1.0979,
+      "step": 654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9849835862239426e-05,
+      "loss": 1.0976,
+      "step": 655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984929478752706e-05,
+      "loss": 1.0143,
+      "step": 656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984875274270786e-05,
+      "loss": 1.062,
+      "step": 657
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9848209727803e-05,
+      "loss": 1.0234,
+      "step": 658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984766574283365e-05,
+      "loss": 1.0672,
+      "step": 659
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9847120787821076e-05,
+      "loss": 0.9884,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984657486278654e-05,
+      "loss": 1.0423,
+      "step": 661
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9846027967751355e-05,
+      "loss": 1.2186,
+      "step": 662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984548010273687e-05,
+      "loss": 1.278,
+      "step": 663
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9844931267764475e-05,
+      "loss": 1.3037,
+      "step": 664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9844381462855606e-05,
+      "loss": 1.1926,
+      "step": 665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984383068803171e-05,
+      "loss": 1.2286,
+      "step": 666
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98432789433143e-05,
+      "loss": 1.2221,
+      "step": 667
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984272622872491e-05,
+      "loss": 1.308,
+      "step": 668
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984217254428513e-05,
+      "loss": 1.2072,
+      "step": 669
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9841617890016556e-05,
+      "loss": 1.3355,
+      "step": 670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984106226594086e-05,
+      "loss": 1.2729,
+      "step": 671
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984050567207972e-05,
+      "loss": 1.2619,
+      "step": 672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983994810845488e-05,
+      "loss": 1.361,
+      "step": 673
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983938957508809e-05,
+      "loss": 1.3972,
+      "step": 674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9838830072001165e-05,
+      "loss": 1.3368,
+      "step": 675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983826959921595e-05,
+      "loss": 1.4566,
+      "step": 676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983770815675432e-05,
+      "loss": 1.446,
+      "step": 677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9837145744638184e-05,
+      "loss": 1.3634,
+      "step": 678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9836582362889515e-05,
+      "loss": 1.3482,
+      "step": 679
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9836018011530305e-05,
+      "loss": 1.352,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9835452690582583e-05,
+      "loss": 1.4117,
+      "step": 681
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9834886400068414e-05,
+      "loss": 1.423,
+      "step": 682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983431914000991e-05,
+      "loss": 1.3161,
+      "step": 683
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9833750910429223e-05,
+      "loss": 1.2074,
+      "step": 684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9833181711348526e-05,
+      "loss": 1.0688,
+      "step": 685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983261154279004e-05,
+      "loss": 1.1633,
+      "step": 686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9832040404776035e-05,
+      "loss": 1.2382,
+      "step": 687
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9831468297328796e-05,
+      "loss": 1.1465,
+      "step": 688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983089522047067e-05,
+      "loss": 1.1146,
+      "step": 689
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983032117422402e-05,
+      "loss": 1.1879,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982974615861126e-05,
+      "loss": 1.1504,
+      "step": 691
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982917017365484e-05,
+      "loss": 1.0364,
+      "step": 692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982859321937724e-05,
+      "loss": 1.2228,
+      "step": 693
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982801529580099e-05,
+      "loss": 1.3019,
+      "step": 694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982743640294865e-05,
+      "loss": 1.2126,
+      "step": 695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9826856540842824e-05,
+      "loss": 1.2954,
+      "step": 696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982627570950614e-05,
+      "loss": 1.2774,
+      "step": 697
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982569390896128e-05,
+      "loss": 1.1203,
+      "step": 698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982511113923096e-05,
+      "loss": 1.2005,
+      "step": 699
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982452740033793e-05,
+      "loss": 1.2388,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982394269230497e-05,
+      "loss": 1.2732,
+      "step": 701
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9823357015154914e-05,
+      "loss": 1.193,
+      "step": 702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982277036891063e-05,
+      "loss": 1.0853,
+      "step": 703
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982218275359501e-05,
+      "loss": 1.1101,
+      "step": 704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9821594169231004e-05,
+      "loss": 1.1426,
+      "step": 705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982100461584158e-05,
+      "loss": 1.1519,
+      "step": 706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982041409344978e-05,
+      "loss": 1.1067,
+      "step": 707
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9819822602078616e-05,
+      "loss": 1.1429,
+      "step": 708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981923014175121e-05,
+      "loss": 1.1014,
+      "step": 709
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9818636712490684e-05,
+      "loss": 1.174,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98180423143202e-05,
+      "loss": 1.0283,
+      "step": 711
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981744694726298e-05,
+      "loss": 1.0881,
+      "step": 712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9816850611342246e-05,
+      "loss": 1.0288,
+      "step": 713
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981625330658128e-05,
+      "loss": 1.1138,
+      "step": 714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981565503300342e-05,
+      "loss": 1.1533,
+      "step": 715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981505579063201e-05,
+      "loss": 1.0123,
+      "step": 716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981445557949044e-05,
+      "loss": 1.0093,
+      "step": 717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981385439960215e-05,
+      "loss": 1.165,
+      "step": 718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9813252250990596e-05,
+      "loss": 0.9753,
+      "step": 719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98126491336793e-05,
+      "loss": 1.1847,
+      "step": 720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981204504769181e-05,
+      "loss": 1.2521,
+      "step": 721
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98114399930517e-05,
+      "loss": 1.2794,
+      "step": 722
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981083396978259e-05,
+      "loss": 1.3514,
+      "step": 723
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981022697790815e-05,
+      "loss": 1.3864,
+      "step": 724
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980961901745206e-05,
+      "loss": 1.3823,
+      "step": 725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980901008843807e-05,
+      "loss": 1.2835,
+      "step": 726
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980840019088995e-05,
+      "loss": 1.3294,
+      "step": 727
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9807789324831495e-05,
+      "loss": 1.3524,
+      "step": 728
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980717749028657e-05,
+      "loss": 1.3154,
+      "step": 729
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980656468727906e-05,
+      "loss": 1.2787,
+      "step": 730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980595091583288e-05,
+      "loss": 1.1504,
+      "step": 731
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980533617597199e-05,
+      "loss": 1.1791,
+      "step": 732
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98047204677204e-05,
+      "loss": 1.1392,
+      "step": 733
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980410379110214e-05,
+      "loss": 1.1773,
+      "step": 734
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9803486146141285e-05,
+      "loss": 1.1734,
+      "step": 735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980286753286195e-05,
+      "loss": 1.1868,
+      "step": 736
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980224795128828e-05,
+      "loss": 1.1676,
+      "step": 737
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980162740144447e-05,
+      "loss": 1.1484,
+      "step": 738
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980100588335474e-05,
+      "loss": 1.1575,
+      "step": 739
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980038339704337e-05,
+      "loss": 1.2216,
+      "step": 740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979975994253464e-05,
+      "loss": 1.2199,
+      "step": 741
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97991355198529e-05,
+      "loss": 1.3068,
+      "step": 742
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979851012902253e-05,
+      "loss": 1.1384,
+      "step": 743
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9797883770067935e-05,
+      "loss": 1.2658,
+      "step": 744
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979725644301357e-05,
+      "loss": 1.2765,
+      "step": 745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9796628147883936e-05,
+      "loss": 1.2296,
+      "step": 746
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9795998884703555e-05,
+      "loss": 1.2646,
+      "step": 747
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9795368653496985e-05,
+      "loss": 1.18,
+      "step": 748
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979473745428884e-05,
+      "loss": 1.1435,
+      "step": 749
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979410528710377e-05,
+      "loss": 1.1679,
+      "step": 750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9793472151966425e-05,
+      "loss": 1.1012,
+      "step": 751
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979283804890155e-05,
+      "loss": 1.123,
+      "step": 752
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979220297793389e-05,
+      "loss": 1.2368,
+      "step": 753
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9791566939088244e-05,
+      "loss": 1.1709,
+      "step": 754
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9790929932389444e-05,
+      "loss": 1.1504,
+      "step": 755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979029195786234e-05,
+      "loss": 1.1397,
+      "step": 756
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9789653015531854e-05,
+      "loss": 1.1946,
+      "step": 757
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978901310542293e-05,
+      "loss": 1.1017,
+      "step": 758
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978837222756054e-05,
+      "loss": 1.161,
+      "step": 759
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978773038196972e-05,
+      "loss": 1.2861,
+      "step": 760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9787087568675506e-05,
+      "loss": 1.1425,
+      "step": 761
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978644378770302e-05,
+      "loss": 1.2034,
+      "step": 762
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978579903907736e-05,
+      "loss": 1.2676,
+      "step": 763
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9785153322823735e-05,
+      "loss": 1.2554,
+      "step": 764
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978450663896733e-05,
+      "loss": 1.2238,
+      "step": 765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9783858987533393e-05,
+      "loss": 1.1909,
+      "step": 766
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978321036854721e-05,
+      "loss": 1.1433,
+      "step": 767
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978256078203411e-05,
+      "loss": 1.2448,
+      "step": 768
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9781910228019446e-05,
+      "loss": 1.23,
+      "step": 769
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978125870652861e-05,
+      "loss": 1.2702,
+      "step": 770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978060621758705e-05,
+      "loss": 1.2634,
+      "step": 771
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9779952761220226e-05,
+      "loss": 1.2714,
+      "step": 772
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9779298337453665e-05,
+      "loss": 1.1929,
+      "step": 773
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97786429463129e-05,
+      "loss": 1.2672,
+      "step": 774
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9777986587823523e-05,
+      "loss": 1.2087,
+      "step": 775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977732926201115e-05,
+      "loss": 1.2018,
+      "step": 776
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9776670968901465e-05,
+      "loss": 1.2975,
+      "step": 777
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9776011708520146e-05,
+      "loss": 1.2462,
+      "step": 778
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977535148089293e-05,
+      "loss": 1.2269,
+      "step": 779
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9774690286045604e-05,
+      "loss": 1.1789,
+      "step": 780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977402812400398e-05,
+      "loss": 1.2116,
+      "step": 781
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97733649947939e-05,
+      "loss": 1.2722,
+      "step": 782
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977270089844126e-05,
+      "loss": 1.2059,
+      "step": 783
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9772035834971984e-05,
+      "loss": 1.2391,
+      "step": 784
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977136980441203e-05,
+      "loss": 1.2624,
+      "step": 785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97707028067874e-05,
+      "loss": 1.3578,
+      "step": 786
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977003484212416e-05,
+      "loss": 1.2974,
+      "step": 787
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976936591044834e-05,
+      "loss": 1.1776,
+      "step": 788
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976869601178609e-05,
+      "loss": 1.1647,
+      "step": 789
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9768025146163555e-05,
+      "loss": 1.212,
+      "step": 790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976735331360692e-05,
+      "loss": 1.2328,
+      "step": 791
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976668051414241e-05,
+      "loss": 1.196,
+      "step": 792
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97660067477963e-05,
+      "loss": 1.2265,
+      "step": 793
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976533201459489e-05,
+      "loss": 1.2088,
+      "step": 794
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976465631456453e-05,
+      "loss": 1.2763,
+      "step": 795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976397964773157e-05,
+      "loss": 1.1665,
+      "step": 796
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976330201412247e-05,
+      "loss": 1.2484,
+      "step": 797
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976262341376365e-05,
+      "loss": 1.1988,
+      "step": 798
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.976194384668161e-05,
+      "loss": 1.2053,
+      "step": 799
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9761263312902895e-05,
+      "loss": 1.3713,
+      "step": 800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9760581812454054e-05,
+      "loss": 1.2155,
+      "step": 801
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975989934536171e-05,
+      "loss": 1.2034,
+      "step": 802
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975921591165248e-05,
+      "loss": 1.1831,
+      "step": 803
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975853151135307e-05,
+      "loss": 1.1886,
+      "step": 804
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975784614449019e-05,
+      "loss": 1.2132,
+      "step": 805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9757159811090595e-05,
+      "loss": 1.2373,
+      "step": 806
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9756472511181084e-05,
+      "loss": 1.0769,
+      "step": 807
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975578424478849e-05,
+      "loss": 1.0045,
+      "step": 808
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9755095011939675e-05,
+      "loss": 0.9587,
+      "step": 809
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9754404812661546e-05,
+      "loss": 0.7925,
+      "step": 810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975371364698106e-05,
+      "loss": 0.8373,
+      "step": 811
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975302151492518e-05,
+      "loss": 0.9687,
+      "step": 812
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975232841652094e-05,
+      "loss": 0.9906,
+      "step": 813
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975163435179541e-05,
+      "loss": 1.0229,
+      "step": 814
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9750939320775665e-05,
+      "loss": 0.9737,
+      "step": 815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975024332348884e-05,
+      "loss": 0.9824,
+      "step": 816
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974954635996212e-05,
+      "loss": 0.9935,
+      "step": 817
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974884843022271e-05,
+      "loss": 1.1692,
+      "step": 818
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974814953429785e-05,
+      "loss": 1.2785,
+      "step": 819
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974744967221483e-05,
+      "loss": 1.24,
+      "step": 820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974674884400097e-05,
+      "loss": 1.2114,
+      "step": 821
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974604704968363e-05,
+      "loss": 1.1236,
+      "step": 822
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9745344289290205e-05,
+      "loss": 1.1234,
+      "step": 823
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974464056284814e-05,
+      "loss": 1.1904,
+      "step": 824
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9743935870384906e-05,
+      "loss": 1.2335,
+      "step": 825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9743230211928006e-05,
+      "loss": 1.219,
+      "step": 826
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9742523587504996e-05,
+      "loss": 1.0903,
+      "step": 827
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9741815997143455e-05,
+      "loss": 1.0118,
+      "step": 828
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9741107440871016e-05,
+      "loss": 0.9704,
+      "step": 829
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9740397918715334e-05,
+      "loss": 1.1186,
+      "step": 830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973968743070411e-05,
+      "loss": 1.0891,
+      "step": 831
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973897597686508e-05,
+      "loss": 1.0633,
+      "step": 832
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9738263557226015e-05,
+      "loss": 0.9631,
+      "step": 833
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973755017181475e-05,
+      "loss": 1.0695,
+      "step": 834
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973683582065911e-05,
+      "loss": 1.0834,
+      "step": 835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973612050378699e-05,
+      "loss": 1.0717,
+      "step": 836
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973540422122632e-05,
+      "loss": 1.0682,
+      "step": 837
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9734686973005054e-05,
+      "loss": 1.1907,
+      "step": 838
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9733968759151206e-05,
+      "loss": 1.1217,
+      "step": 839
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973324957969281e-05,
+      "loss": 1.0715,
+      "step": 840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9732529434657936e-05,
+      "loss": 1.1177,
+      "step": 841
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9731808324074717e-05,
+      "loss": 1.1679,
+      "step": 842
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973108624797127e-05,
+      "loss": 1.0835,
+      "step": 843
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9730363206375816e-05,
+      "loss": 1.1515,
+      "step": 844
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972963919931658e-05,
+      "loss": 1.16,
+      "step": 845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972891422682181e-05,
+      "loss": 1.2386,
+      "step": 846
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972818828891982e-05,
+      "loss": 1.2859,
+      "step": 847
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9727461385638944e-05,
+      "loss": 1.3257,
+      "step": 848
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972673351700757e-05,
+      "loss": 1.2819,
+      "step": 849
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97260046830541e-05,
+      "loss": 1.312,
+      "step": 850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9725274883807e-05,
+      "loss": 1.2956,
+      "step": 851
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972454411929476e-05,
+      "loss": 1.3138,
+      "step": 852
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97238123895459e-05,
+      "loss": 1.3544,
+      "step": 853
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972307969458899e-05,
+      "loss": 1.3193,
+      "step": 854
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972234603445264e-05,
+      "loss": 1.3177,
+      "step": 855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972161140916548e-05,
+      "loss": 0.9616,
+      "step": 856
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972087581875621e-05,
+      "loss": 1.1026,
+      "step": 857
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9720139263253515e-05,
+      "loss": 1.0328,
+      "step": 858
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9719401742686186e-05,
+      "loss": 0.952,
+      "step": 859
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9718663257082984e-05,
+      "loss": 1.0373,
+      "step": 860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9717923806472764e-05,
+      "loss": 0.9947,
+      "step": 861
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971718339088438e-05,
+      "loss": 0.9999,
+      "step": 862
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971644201034674e-05,
+      "loss": 1.0643,
+      "step": 863
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971569966488879e-05,
+      "loss": 1.0826,
+      "step": 864
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971495635453951e-05,
+      "loss": 0.9796,
+      "step": 865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9714212079327914e-05,
+      "loss": 0.9355,
+      "step": 866
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971346683928307e-05,
+      "loss": 0.9061,
+      "step": 867
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971272063443406e-05,
+      "loss": 0.9106,
+      "step": 868
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9711973464810015e-05,
+      "loss": 1.0008,
+      "step": 869
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9711225330440103e-05,
+      "loss": 0.9137,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971047623135355e-05,
+      "loss": 1.0685,
+      "step": 871
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9709726167579574e-05,
+      "loss": 0.9475,
+      "step": 872
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970897513914748e-05,
+      "loss": 0.9823,
+      "step": 873
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9708223146086576e-05,
+      "loss": 1.0531,
+      "step": 874
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970747018842622e-05,
+      "loss": 1.0285,
+      "step": 875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9706716266195804e-05,
+      "loss": 1.1665,
+      "step": 876
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970596137942477e-05,
+      "loss": 1.24,
+      "step": 877
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9705205528142586e-05,
+      "loss": 1.2752,
+      "step": 878
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970444871237875e-05,
+      "loss": 1.2195,
+      "step": 879
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9703690932162824e-05,
+      "loss": 1.2541,
+      "step": 880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9702932187524376e-05,
+      "loss": 1.2929,
+      "step": 881
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9702172478493035e-05,
+      "loss": 1.2581,
+      "step": 882
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.970141180509847e-05,
+      "loss": 1.3096,
+      "step": 883
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9700650167370356e-05,
+      "loss": 1.216,
+      "step": 884
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9699887565338445e-05,
+      "loss": 1.3822,
+      "step": 885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969912399903249e-05,
+      "loss": 1.2677,
+      "step": 886
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969835946848232e-05,
+      "loss": 1.3419,
+      "step": 887
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9697593973717774e-05,
+      "loss": 1.0412,
+      "step": 888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969682751476874e-05,
+      "loss": 1.105,
+      "step": 889
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969606009166512e-05,
+      "loss": 1.0662,
+      "step": 890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969529170443691e-05,
+      "loss": 1.1378,
+      "step": 891
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969452235311407e-05,
+      "loss": 1.1629,
+      "step": 892
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969375203772667e-05,
+      "loss": 1.1435,
+      "step": 893
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9692980758304754e-05,
+      "loss": 1.0964,
+      "step": 894
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9692208514878444e-05,
+      "loss": 1.0756,
+      "step": 895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96914353074779e-05,
+      "loss": 1.1059,
+      "step": 896
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969066113613328e-05,
+      "loss": 1.1039,
+      "step": 897
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968988600087483e-05,
+      "loss": 1.3718,
+      "step": 898
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968910990173281e-05,
+      "loss": 1.3796,
+      "step": 899
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9688332838737504e-05,
+      "loss": 1.4275,
+      "step": 900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968755481191927e-05,
+      "loss": 1.4453,
+      "step": 901
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968677582130845e-05,
+      "loss": 1.4731,
+      "step": 902
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9685995866935495e-05,
+      "loss": 1.5,
+      "step": 903
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9685214948830825e-05,
+      "loss": 1.509,
+      "step": 904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968443306702493e-05,
+      "loss": 1.3768,
+      "step": 905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968365022154835e-05,
+      "loss": 1.3944,
+      "step": 906
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968286641243163e-05,
+      "loss": 1.4264,
+      "step": 907
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968208163970538e-05,
+      "loss": 1.3507,
+      "step": 908
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9681295903400226e-05,
+      "loss": 1.4508,
+      "step": 909
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968050920354686e-05,
+      "loss": 1.1814,
+      "step": 910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9679721540175986e-05,
+      "loss": 1.2808,
+      "step": 911
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.967893291331834e-05,
+      "loss": 1.1409,
+      "step": 912
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.967814332300473e-05,
+      "loss": 1.2567,
+      "step": 913
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.967735276926599e-05,
+      "loss": 1.077,
+      "step": 914
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.967656125213295e-05,
+      "loss": 1.1244,
+      "step": 915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.967576877163652e-05,
+      "loss": 1.182,
+      "step": 916
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.967497532780766e-05,
+      "loss": 1.117,
+      "step": 917
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9674180920677324e-05,
+      "loss": 1.2845,
+      "step": 918
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9673385550276537e-05,
+      "loss": 1.0927,
+      "step": 919
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9672589216636346e-05,
+      "loss": 1.0784,
+      "step": 920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9671791919787836e-05,
+      "loss": 1.2001,
+      "step": 921
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9670993659762135e-05,
+      "loss": 1.1338,
+      "step": 922
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9670194436590415e-05,
+      "loss": 1.1094,
+      "step": 923
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966939425030386e-05,
+      "loss": 1.005,
+      "step": 924
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966859310093372e-05,
+      "loss": 1.1496,
+      "step": 925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966779098851128e-05,
+      "loss": 1.0371,
+      "step": 926
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966698791306784e-05,
+      "loss": 1.0439,
+      "step": 927
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966618387463476e-05,
+      "loss": 1.1512,
+      "step": 928
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966537887324342e-05,
+      "loss": 1.0223,
+      "step": 929
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966457290892526e-05,
+      "loss": 1.0277,
+      "step": 930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9663765981711736e-05,
+      "loss": 1.0032,
+      "step": 931
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966295809163435e-05,
+      "loss": 1.0605,
+      "step": 932
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.966214923872464e-05,
+      "loss": 1.1108,
+      "step": 933
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9661339423014195e-05,
+      "loss": 1.0878,
+      "step": 934
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9660528644534624e-05,
+      "loss": 1.107,
+      "step": 935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965971690331757e-05,
+      "loss": 1.0272,
+      "step": 936
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965890419939473e-05,
+      "loss": 1.1114,
+      "step": 937
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965809053279783e-05,
+      "loss": 1.0745,
+      "step": 938
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965727590355864e-05,
+      "loss": 1.1923,
+      "step": 939
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9656460311708963e-05,
+      "loss": 1.2801,
+      "step": 940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9655643757280635e-05,
+      "loss": 1.3708,
+      "step": 941
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965482624030553e-05,
+      "loss": 1.3589,
+      "step": 942
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965400776081558e-05,
+      "loss": 1.3548,
+      "step": 943
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965318831884271e-05,
+      "loss": 1.4131,
+      "step": 944
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965236791441895e-05,
+      "loss": 1.3425,
+      "step": 945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965154654757629e-05,
+      "loss": 1.3467,
+      "step": 946
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.965072421834681e-05,
+      "loss": 1.3494,
+      "step": 947
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964990092676263e-05,
+      "loss": 1.3055,
+      "step": 948
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964907667285587e-05,
+      "loss": 1.3896,
+      "step": 949
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96482514566587e-05,
+      "loss": 1.344,
+      "step": 950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964742527820336e-05,
+      "loss": 1.2676,
+      "step": 951
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96465981375221e-05,
+      "loss": 1.2232,
+      "step": 952
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964577003464721e-05,
+      "loss": 1.2875,
+      "step": 953
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964494096961101e-05,
+      "loss": 1.304,
+      "step": 954
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9644110942445864e-05,
+      "loss": 1.2794,
+      "step": 955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964327995318418e-05,
+      "loss": 1.3844,
+      "step": 956
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964244800185842e-05,
+      "loss": 1.4483,
+      "step": 957
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9641615088501035e-05,
+      "loss": 1.2625,
+      "step": 958
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.964078121314455e-05,
+      "loss": 1.2248,
+      "step": 959
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9639946375821524e-05,
+      "loss": 1.2729,
+      "step": 960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9639110576564544e-05,
+      "loss": 1.3325,
+      "step": 961
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963827381540624e-05,
+      "loss": 1.2411,
+      "step": 962
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963743609237929e-05,
+      "loss": 1.3711,
+      "step": 963
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9636597407516384e-05,
+      "loss": 1.2873,
+      "step": 964
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9635757760850264e-05,
+      "loss": 1.3033,
+      "step": 965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963491715241372e-05,
+      "loss": 1.3212,
+      "step": 966
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9634075582239556e-05,
+      "loss": 1.2728,
+      "step": 967
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963323305036064e-05,
+      "loss": 1.2568,
+      "step": 968
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963238955680985e-05,
+      "loss": 1.2759,
+      "step": 969
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9631545101620125e-05,
+      "loss": 1.3063,
+      "step": 970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.963069968482443e-05,
+      "loss": 1.3038,
+      "step": 971
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9629853306455784e-05,
+      "loss": 1.23,
+      "step": 972
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9629005966547205e-05,
+      "loss": 1.1983,
+      "step": 973
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962815766513178e-05,
+      "loss": 1.2259,
+      "step": 974
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962730840224263e-05,
+      "loss": 1.2731,
+      "step": 975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9626458177912916e-05,
+      "loss": 1.3922,
+      "step": 976
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962560699217582e-05,
+      "loss": 1.227,
+      "step": 977
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962475484506458e-05,
+      "loss": 1.2587,
+      "step": 978
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962390173661245e-05,
+      "loss": 1.2849,
+      "step": 979
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9623047666852754e-05,
+      "loss": 1.2249,
+      "step": 980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962219263581882e-05,
+      "loss": 1.176,
+      "step": 981
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962133664354404e-05,
+      "loss": 1.2063,
+      "step": 982
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962047969006182e-05,
+      "loss": 1.2435,
+      "step": 983
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961962177540561e-05,
+      "loss": 1.2803,
+      "step": 984
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961876289960893e-05,
+      "loss": 1.1724,
+      "step": 985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9617903062705286e-05,
+      "loss": 1.0651,
+      "step": 986
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9617042264728246e-05,
+      "loss": 1.194,
+      "step": 987
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961618050571143e-05,
+      "loss": 1.0922,
+      "step": 988
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961531778568846e-05,
+      "loss": 1.142,
+      "step": 989
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961445410469304e-05,
+      "loss": 1.2091,
+      "step": 990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961358946275888e-05,
+      "loss": 1.3047,
+      "step": 991
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961272385991972e-05,
+      "loss": 1.2433,
+      "step": 992
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9611857296209374e-05,
+      "loss": 1.1844,
+      "step": 993
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961098977166166e-05,
+      "loss": 1.266,
+      "step": 994
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9610121286310443e-05,
+      "loss": 1.2414,
+      "step": 995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960925184018964e-05,
+      "loss": 1.2751,
+      "step": 996
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9608381433333185e-05,
+      "loss": 1.2184,
+      "step": 997
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9607510065775066e-05,
+      "loss": 1.1467,
+      "step": 998
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960663773754929e-05,
+      "loss": 1.1738,
+      "step": 999
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960576444868992e-05,
+      "loss": 1.3737,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 1.170454978942871,
+      "eval_runtime": 111.2756,
+      "eval_samples_per_second": 34.338,
+      "eval_steps_per_second": 0.359,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960489019923105e-05,
+      "loss": 1.3183,
+      "step": 1001
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9604014989206804e-05,
+      "loss": 1.4039,
+      "step": 1002
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960313881865136e-05,
+      "loss": 1.4363,
+      "step": 1003
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960226168759891e-05,
+      "loss": 1.3837,
+      "step": 1004
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9601383596083705e-05,
+      "loss": 1.5267,
+      "step": 1005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960050454414003e-05,
+      "loss": 1.4183,
+      "step": 1006
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9599624531802186e-05,
+      "loss": 1.3809,
+      "step": 1007
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959874355910454e-05,
+      "loss": 1.4917,
+      "step": 1008
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959786162608149e-05,
+      "loss": 1.2856,
+      "step": 1009
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9596978732767464e-05,
+      "loss": 1.234,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959609487919692e-05,
+      "loss": 1.2581,
+      "step": 1011
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9595210065404364e-05,
+      "loss": 1.2289,
+      "step": 1012
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9594324291424355e-05,
+      "loss": 1.264,
+      "step": 1013
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9593437557291456e-05,
+      "loss": 1.2829,
+      "step": 1014
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9592549863040295e-05,
+      "loss": 1.2376,
+      "step": 1015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959166120870552e-05,
+      "loss": 1.2891,
+      "step": 1016
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9590771594321826e-05,
+      "loss": 1.1728,
+      "step": 1017
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958988101992395e-05,
+      "loss": 1.306,
+      "step": 1018
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9588989485546654e-05,
+      "loss": 1.2243,
+      "step": 1019
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9588096991224744e-05,
+      "loss": 1.2392,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958720353699306e-05,
+      "loss": 1.2241,
+      "step": 1021
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958630912288648e-05,
+      "loss": 1.2283,
+      "step": 1022
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958541374893992e-05,
+      "loss": 1.2258,
+      "step": 1023
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958451741518835e-05,
+      "loss": 1.1513,
+      "step": 1024
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9583620121666754e-05,
+      "loss": 1.2862,
+      "step": 1025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958272186841016e-05,
+      "loss": 1.2584,
+      "step": 1026
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958182265545364e-05,
+      "loss": 1.179,
+      "step": 1027
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958092248283228e-05,
+      "loss": 1.2031,
+      "step": 1028
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958002135058125e-05,
+      "loss": 1.2383,
+      "step": 1029
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957911925873572e-05,
+      "loss": 1.2841,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95782162073309e-05,
+      "loss": 1.1836,
+      "step": 1031
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9577312196402055e-05,
+      "loss": 1.2921,
+      "step": 1032
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957640722598447e-05,
+      "loss": 1.2201,
+      "step": 1033
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957550129611347e-05,
+      "loss": 1.1936,
+      "step": 1034
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957459440682444e-05,
+      "loss": 1.172,
+      "step": 1035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9573686558152765e-05,
+      "loss": 1.2739,
+      "step": 1036
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95727777501339e-05,
+      "loss": 1.215,
+      "step": 1037
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9571867982803325e-05,
+      "loss": 1.1172,
+      "step": 1038
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957095725619654e-05,
+      "loss": 1.2053,
+      "step": 1039
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.957004557034912e-05,
+      "loss": 1.1464,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9569132925296644e-05,
+      "loss": 1.2274,
+      "step": 1041
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9568219321074747e-05,
+      "loss": 1.2141,
+      "step": 1042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956730475771909e-05,
+      "loss": 1.1756,
+      "step": 1043
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956638923526539e-05,
+      "loss": 1.2518,
+      "step": 1044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956547275374937e-05,
+      "loss": 1.1901,
+      "step": 1045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956455531320683e-05,
+      "loss": 1.1954,
+      "step": 1046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9563636913673575e-05,
+      "loss": 1.1101,
+      "step": 1047
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9562717555185454e-05,
+      "loss": 1.2452,
+      "step": 1048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9561797237778366e-05,
+      "loss": 1.2886,
+      "step": 1049
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.956087596148824e-05,
+      "loss": 1.2657,
+      "step": 1050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955995372635104e-05,
+      "loss": 1.219,
+      "step": 1051
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9559030532402776e-05,
+      "loss": 1.2417,
+      "step": 1052
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955810637967947e-05,
+      "loss": 1.294,
+      "step": 1053
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9557181268217227e-05,
+      "loss": 1.2169,
+      "step": 1054
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9556255198052135e-05,
+      "loss": 1.252,
+      "step": 1055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955532816922037e-05,
+      "loss": 1.3136,
+      "step": 1056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955440018175811e-05,
+      "loss": 1.1891,
+      "step": 1057
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95534712357016e-05,
+      "loss": 1.1691,
+      "step": 1058
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955254133108709e-05,
+      "loss": 1.2036,
+      "step": 1059
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955161046795088e-05,
+      "loss": 1.1725,
+      "step": 1060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.955067864632932e-05,
+      "loss": 1.2037,
+      "step": 1061
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954974586625879e-05,
+      "loss": 1.1669,
+      "step": 1062
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95488121277757e-05,
+      "loss": 1.1914,
+      "step": 1063
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95478774309165e-05,
+      "loss": 1.2138,
+      "step": 1064
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954694177571768e-05,
+      "loss": 1.1255,
+      "step": 1065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954600516221578e-05,
+      "loss": 1.1804,
+      "step": 1066
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954506759044736e-05,
+      "loss": 1.1702,
+      "step": 1067
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9544129060449006e-05,
+      "loss": 1.1664,
+      "step": 1068
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9543189572257384e-05,
+      "loss": 1.1295,
+      "step": 1069
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9542249125909146e-05,
+      "loss": 1.2006,
+      "step": 1070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9541307721441026e-05,
+      "loss": 1.1663,
+      "step": 1071
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.954036535888977e-05,
+      "loss": 1.2252,
+      "step": 1072
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.953942203829216e-05,
+      "loss": 1.2064,
+      "step": 1073
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.953847775968504e-05,
+      "loss": 1.1829,
+      "step": 1074
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.953753252310526e-05,
+      "loss": 1.2763,
+      "step": 1075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.953658632858972e-05,
+      "loss": 1.2439,
+      "step": 1076
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9535639176175366e-05,
+      "loss": 1.2392,
+      "step": 1077
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9534691065899164e-05,
+      "loss": 1.1519,
+      "step": 1078
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9533741997798146e-05,
+      "loss": 1.1159,
+      "step": 1079
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.953279197190935e-05,
+      "loss": 1.1864,
+      "step": 1080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.953184098826987e-05,
+      "loss": 1.1436,
+      "step": 1081
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.953088904691684e-05,
+      "loss": 1.191,
+      "step": 1082
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95299361478874e-05,
+      "loss": 1.1289,
+      "step": 1083
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9528982291218774e-05,
+      "loss": 1.1851,
+      "step": 1084
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9528027476948184e-05,
+      "loss": 1.1396,
+      "step": 1085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.952707170511291e-05,
+      "loss": 1.2331,
+      "step": 1086
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.952611497575028e-05,
+      "loss": 1.2644,
+      "step": 1087
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.952515728889763e-05,
+      "loss": 1.1928,
+      "step": 1088
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9524198644592334e-05,
+      "loss": 1.2582,
+      "step": 1089
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.952323904287185e-05,
+      "loss": 1.2547,
+      "step": 1090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.952227848377361e-05,
+      "loss": 1.3429,
+      "step": 1091
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.952131696733513e-05,
+      "loss": 1.2841,
+      "step": 1092
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.952035449359395e-05,
+      "loss": 1.2413,
+      "step": 1093
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.951939106258764e-05,
+      "loss": 1.3201,
+      "step": 1094
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95184266743538e-05,
+      "loss": 1.2783,
+      "step": 1095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9517461328930095e-05,
+      "loss": 1.3698,
+      "step": 1096
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.951649502635421e-05,
+      "loss": 1.2638,
+      "step": 1097
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9515527766663864e-05,
+      "loss": 1.2725,
+      "step": 1098
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.951455954989682e-05,
+      "loss": 1.3046,
+      "step": 1099
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.951359037609088e-05,
+      "loss": 1.3059,
+      "step": 1100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.951262024528387e-05,
+      "loss": 1.3098,
+      "step": 1101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9511649157513674e-05,
+      "loss": 1.3227,
+      "step": 1102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9510677112818206e-05,
+      "loss": 1.1668,
+      "step": 1103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95097041112354e-05,
+      "loss": 1.3205,
+      "step": 1104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.950873015280325e-05,
+      "loss": 1.2346,
+      "step": 1105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9507755237559786e-05,
+      "loss": 1.2599,
+      "step": 1106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9506779365543046e-05,
+      "loss": 1.2147,
+      "step": 1107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.950580253679116e-05,
+      "loss": 1.3804,
+      "step": 1108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9504824751342225e-05,
+      "loss": 1.3014,
+      "step": 1109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.950384600923445e-05,
+      "loss": 1.1809,
+      "step": 1110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9502866310506026e-05,
+      "loss": 1.1928,
+      "step": 1111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9501885655195204e-05,
+      "loss": 1.0993,
+      "step": 1112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.950090404334026e-05,
+      "loss": 1.1464,
+      "step": 1113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.949992147497953e-05,
+      "loss": 1.0764,
+      "step": 1114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.949893795015136e-05,
+      "loss": 1.2097,
+      "step": 1115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9497953468894154e-05,
+      "loss": 1.1463,
+      "step": 1116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.949696803124635e-05,
+      "loss": 1.0771,
+      "step": 1117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94959816372464e-05,
+      "loss": 1.0566,
+      "step": 1118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.949499428693284e-05,
+      "loss": 1.1125,
+      "step": 1119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9494005980344194e-05,
+      "loss": 1.1147,
+      "step": 1120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9493016717519054e-05,
+      "loss": 1.181,
+      "step": 1121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.949202649849604e-05,
+      "loss": 1.1501,
+      "step": 1122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9491035323313804e-05,
+      "loss": 1.1131,
+      "step": 1123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9490043192011046e-05,
+      "loss": 1.1997,
+      "step": 1124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94890501046265e-05,
+      "loss": 1.1624,
+      "step": 1125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.948805606119894e-05,
+      "loss": 1.2082,
+      "step": 1126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9487061061767156e-05,
+      "loss": 1.1301,
+      "step": 1127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9486065106370016e-05,
+      "loss": 1.1372,
+      "step": 1128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.948506819504638e-05,
+      "loss": 1.1271,
+      "step": 1129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.948407032783518e-05,
+      "loss": 1.2458,
+      "step": 1130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.948307150477537e-05,
+      "loss": 1.3895,
+      "step": 1131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.948207172590594e-05,
+      "loss": 1.2239,
+      "step": 1132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.948107099126592e-05,
+      "loss": 1.2357,
+      "step": 1133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.948006930089438e-05,
+      "loss": 1.3228,
+      "step": 1134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947906665483043e-05,
+      "loss": 1.3329,
+      "step": 1135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947806305311321e-05,
+      "loss": 1.3459,
+      "step": 1136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94770584957819e-05,
+      "loss": 1.3372,
+      "step": 1137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947605298287572e-05,
+      "loss": 1.3362,
+      "step": 1138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947504651443392e-05,
+      "loss": 1.2431,
+      "step": 1139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94740390904958e-05,
+      "loss": 1.2714,
+      "step": 1140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9473030711100674e-05,
+      "loss": 1.3117,
+      "step": 1141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947202137628793e-05,
+      "loss": 1.3663,
+      "step": 1142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947101108609695e-05,
+      "loss": 1.2818,
+      "step": 1143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9469999840567195e-05,
+      "loss": 1.282,
+      "step": 1144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9468987639738135e-05,
+      "loss": 1.2858,
+      "step": 1145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9467974483649284e-05,
+      "loss": 1.3741,
+      "step": 1146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9466960372340196e-05,
+      "loss": 1.3467,
+      "step": 1147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946594530585046e-05,
+      "loss": 1.3133,
+      "step": 1148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946492928421971e-05,
+      "loss": 1.3471,
+      "step": 1149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94639123074876e-05,
+      "loss": 1.2766,
+      "step": 1150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946289437569385e-05,
+      "loss": 1.3504,
+      "step": 1151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9461875488878184e-05,
+      "loss": 1.3554,
+      "step": 1152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946085564708038e-05,
+      "loss": 1.177,
+      "step": 1153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945983485034027e-05,
+      "loss": 1.1868,
+      "step": 1154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945881309869768e-05,
+      "loss": 1.1384,
+      "step": 1155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9457790392192516e-05,
+      "loss": 1.1481,
+      "step": 1156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9456766730864694e-05,
+      "loss": 1.17,
+      "step": 1157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945574211475419e-05,
+      "loss": 1.2655,
+      "step": 1158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9454716543900994e-05,
+      "loss": 1.209,
+      "step": 1159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9453690018345144e-05,
+      "loss": 1.3122,
+      "step": 1160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945266253812672e-05,
+      "loss": 1.1191,
+      "step": 1161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9451634103285835e-05,
+      "loss": 1.1471,
+      "step": 1162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945060471386263e-05,
+      "loss": 1.2887,
+      "step": 1163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94495743698973e-05,
+      "loss": 1.2648,
+      "step": 1164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944854307143007e-05,
+      "loss": 1.303,
+      "step": 1165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9447510818501196e-05,
+      "loss": 1.2583,
+      "step": 1166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944647761115098e-05,
+      "loss": 1.2808,
+      "step": 1167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944544344941976e-05,
+      "loss": 1.2169,
+      "step": 1168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944440833334791e-05,
+      "loss": 1.2581,
+      "step": 1169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944337226297583e-05,
+      "loss": 1.2905,
+      "step": 1170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944233523834398e-05,
+      "loss": 1.2885,
+      "step": 1171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9441297259492835e-05,
+      "loss": 1.178,
+      "step": 1172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944025832646293e-05,
+      "loss": 1.1186,
+      "step": 1173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943921843929481e-05,
+      "loss": 1.2009,
+      "step": 1174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943817759802908e-05,
+      "loss": 1.2821,
+      "step": 1175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9437135802706375e-05,
+      "loss": 1.1832,
+      "step": 1176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943609305336736e-05,
+      "loss": 1.149,
+      "step": 1177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9435049350052755e-05,
+      "loss": 1.1956,
+      "step": 1178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943400469280329e-05,
+      "loss": 1.1367,
+      "step": 1179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943295908165977e-05,
+      "loss": 1.1842,
+      "step": 1180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9431912516662996e-05,
+      "loss": 1.2655,
+      "step": 1181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943086499785382e-05,
+      "loss": 1.3412,
+      "step": 1182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942981652527315e-05,
+      "loss": 1.2799,
+      "step": 1183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942876709896193e-05,
+      "loss": 1.2919,
+      "step": 1184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94277167189611e-05,
+      "loss": 1.3622,
+      "step": 1185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942666538531169e-05,
+      "loss": 1.3146,
+      "step": 1186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9425613098054735e-05,
+      "loss": 1.3116,
+      "step": 1187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9424559857231314e-05,
+      "loss": 1.2908,
+      "step": 1188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942350566288255e-05,
+      "loss": 1.2167,
+      "step": 1189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94224505150496e-05,
+      "loss": 1.3134,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942139441377364e-05,
+      "loss": 1.2011,
+      "step": 1191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942033735909592e-05,
+      "loss": 1.0621,
+      "step": 1192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9419279351057696e-05,
+      "loss": 1.0676,
+      "step": 1193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941822038970028e-05,
+      "loss": 0.9982,
+      "step": 1194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9417160475065e-05,
+      "loss": 1.0232,
+      "step": 1195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9416099607193244e-05,
+      "loss": 1.0727,
+      "step": 1196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941503778612643e-05,
+      "loss": 1.2344,
+      "step": 1197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9413975011906014e-05,
+      "loss": 1.0099,
+      "step": 1198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941291128457347e-05,
+      "loss": 1.0168,
+      "step": 1199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9411846604170345e-05,
+      "loss": 0.9844,
+      "step": 1200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941078097073818e-05,
+      "loss": 1.1445,
+      "step": 1201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940971438431861e-05,
+      "loss": 1.234,
+      "step": 1202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940864684495324e-05,
+      "loss": 1.3169,
+      "step": 1203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9407578352683775e-05,
+      "loss": 1.3441,
+      "step": 1204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940650890755191e-05,
+      "loss": 1.3016,
+      "step": 1205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94054385095994e-05,
+      "loss": 1.2392,
+      "step": 1206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9404367158868036e-05,
+      "loss": 1.3073,
+      "step": 1207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940329485539964e-05,
+      "loss": 1.299,
+      "step": 1208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940222159923608e-05,
+      "loss": 1.2426,
+      "step": 1209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940114739041924e-05,
+      "loss": 1.276,
+      "step": 1210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940007222899108e-05,
+      "loss": 1.1647,
+      "step": 1211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9398996114993554e-05,
+      "loss": 1.2247,
+      "step": 1212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.939791904846869e-05,
+      "loss": 1.101,
+      "step": 1213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.939684102945852e-05,
+      "loss": 1.1373,
+      "step": 1214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9395762058005136e-05,
+      "loss": 1.2256,
+      "step": 1215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.939468213415066e-05,
+      "loss": 1.1889,
+      "step": 1216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9393601257937264e-05,
+      "loss": 1.1327,
+      "step": 1217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.939251942940712e-05,
+      "loss": 1.142,
+      "step": 1218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.939143664860248e-05,
+      "loss": 1.1131,
+      "step": 1219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.939035291556561e-05,
+      "loss": 1.1629,
+      "step": 1220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9389268230338823e-05,
+      "loss": 1.1555,
+      "step": 1221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.938818259296446e-05,
+      "loss": 1.2047,
+      "step": 1222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.938709600348491e-05,
+      "loss": 1.2713,
+      "step": 1223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9386008461942584e-05,
+      "loss": 1.2095,
+      "step": 1224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9384919968379945e-05,
+      "loss": 1.2715,
+      "step": 1225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9383830522839484e-05,
+      "loss": 1.1678,
+      "step": 1226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9382740125363736e-05,
+      "loss": 1.2526,
+      "step": 1227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.938164877599527e-05,
+      "loss": 1.2606,
+      "step": 1228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.938055647477669e-05,
+      "loss": 1.1673,
+      "step": 1229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.937946322175064e-05,
+      "loss": 1.0923,
+      "step": 1230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.937836901695979e-05,
+      "loss": 0.9472,
+      "step": 1231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9377273860446885e-05,
+      "loss": 0.9842,
+      "step": 1232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.937617775225464e-05,
+      "loss": 1.0232,
+      "step": 1233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.937508069242588e-05,
+      "loss": 1.0323,
+      "step": 1234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.937398268100342e-05,
+      "loss": 1.0779,
+      "step": 1235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.937288371803013e-05,
+      "loss": 0.9662,
+      "step": 1236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.937178380354891e-05,
+      "loss": 1.0621,
+      "step": 1237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93706829376027e-05,
+      "loss": 1.0734,
+      "step": 1238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9369581120234474e-05,
+      "loss": 1.1481,
+      "step": 1239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.936847835148725e-05,
+      "loss": 1.3177,
+      "step": 1240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.936737463140408e-05,
+      "loss": 1.3943,
+      "step": 1241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.936626996002807e-05,
+      "loss": 1.3517,
+      "step": 1242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.936516433740231e-05,
+      "loss": 1.289,
+      "step": 1243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9364057763569986e-05,
+      "loss": 1.3647,
+      "step": 1244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93629502385743e-05,
+      "loss": 1.2651,
+      "step": 1245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.936184176245848e-05,
+      "loss": 1.3288,
+      "step": 1246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93607323352658e-05,
+      "loss": 1.2475,
+      "step": 1247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.935962195703959e-05,
+      "loss": 1.315,
+      "step": 1248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.935851062782317e-05,
+      "loss": 1.2641,
+      "step": 1249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.935739834765995e-05,
+      "loss": 1.2444,
+      "step": 1250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.935628511659334e-05,
+      "loss": 1.2292,
+      "step": 1251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93551709346668e-05,
+      "loss": 1.2474,
+      "step": 1252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9354055801923834e-05,
+      "loss": 1.286,
+      "step": 1253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.935293971840798e-05,
+      "loss": 1.1661,
+      "step": 1254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9351822684162787e-05,
+      "loss": 1.2863,
+      "step": 1255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9350704699231884e-05,
+      "loss": 1.1881,
+      "step": 1256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.934958576365892e-05,
+      "loss": 1.3293,
+      "step": 1257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.934846587748757e-05,
+      "loss": 1.2528,
+      "step": 1258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.934734504076155e-05,
+      "loss": 1.3277,
+      "step": 1259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.934622325352461e-05,
+      "loss": 1.2711,
+      "step": 1260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.934510051582058e-05,
+      "loss": 1.2315,
+      "step": 1261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.934397682769324e-05,
+      "loss": 1.3081,
+      "step": 1262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9342852189186496e-05,
+      "loss": 1.2664,
+      "step": 1263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.934172660034424e-05,
+      "loss": 1.2596,
+      "step": 1264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9340600061210416e-05,
+      "loss": 1.2801,
+      "step": 1265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.933947257182901e-05,
+      "loss": 1.2583,
+      "step": 1266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.933834413224403e-05,
+      "loss": 1.1779,
+      "step": 1267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9337214742499524e-05,
+      "loss": 1.2517,
+      "step": 1268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9336084402639605e-05,
+      "loss": 1.2514,
+      "step": 1269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9334953112708385e-05,
+      "loss": 1.1848,
+      "step": 1270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9333820872750026e-05,
+      "loss": 1.2551,
+      "step": 1271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.933268768280873e-05,
+      "loss": 1.1804,
+      "step": 1272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.933155354292875e-05,
+      "loss": 1.1884,
+      "step": 1273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.933041845315436e-05,
+      "loss": 1.1803,
+      "step": 1274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932928241352987e-05,
+      "loss": 1.2088,
+      "step": 1275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9328145424099614e-05,
+      "loss": 1.2674,
+      "step": 1276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9327007484908e-05,
+      "loss": 1.1548,
+      "step": 1277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932586859599945e-05,
+      "loss": 1.2939,
+      "step": 1278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932472875741842e-05,
+      "loss": 1.2448,
+      "step": 1279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9323587969209414e-05,
+      "loss": 1.3097,
+      "step": 1280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932244623141696e-05,
+      "loss": 1.2237,
+      "step": 1281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9321303544085644e-05,
+      "loss": 1.2974,
+      "step": 1282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932015990726006e-05,
+      "loss": 1.2611,
+      "step": 1283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.931901532098487e-05,
+      "loss": 1.2806,
+      "step": 1284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.931786978530476e-05,
+      "loss": 1.1943,
+      "step": 1285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9316723300264436e-05,
+      "loss": 1.3575,
+      "step": 1286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.931557586590866e-05,
+      "loss": 1.2537,
+      "step": 1287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9314427482282234e-05,
+      "loss": 1.2478,
+      "step": 1288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.931327814942999e-05,
+      "loss": 1.2569,
+      "step": 1289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9312127867396784e-05,
+      "loss": 1.183,
+      "step": 1290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.931097663622754e-05,
+      "loss": 1.1644,
+      "step": 1291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93098244559672e-05,
+      "loss": 1.3239,
+      "step": 1292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9308671326660735e-05,
+      "loss": 1.3161,
+      "step": 1293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9307517248353165e-05,
+      "loss": 1.2482,
+      "step": 1294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9306362221089554e-05,
+      "loss": 1.2077,
+      "step": 1295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930520624491498e-05,
+      "loss": 1.1943,
+      "step": 1296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930404931987458e-05,
+      "loss": 1.2321,
+      "step": 1297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930289144601352e-05,
+      "loss": 1.2424,
+      "step": 1298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9301732623377005e-05,
+      "loss": 1.2734,
+      "step": 1299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930057285201027e-05,
+      "loss": 1.0976,
+      "step": 1300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9299412131958596e-05,
+      "loss": 1.2578,
+      "step": 1301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9298250463267295e-05,
+      "loss": 1.1427,
+      "step": 1302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9297087845981715e-05,
+      "loss": 1.2139,
+      "step": 1303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9295924280147246e-05,
+      "loss": 1.1468,
+      "step": 1304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929475976580932e-05,
+      "loss": 1.2481,
+      "step": 1305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929359430301338e-05,
+      "loss": 1.3442,
+      "step": 1306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929242789180495e-05,
+      "loss": 1.2661,
+      "step": 1307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929126053222956e-05,
+      "loss": 1.2728,
+      "step": 1308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929009222433276e-05,
+      "loss": 1.213,
+      "step": 1309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928892296816019e-05,
+      "loss": 1.3305,
+      "step": 1310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928775276375749e-05,
+      "loss": 1.3451,
+      "step": 1311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928658161117033e-05,
+      "loss": 1.3304,
+      "step": 1312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928540951044445e-05,
+      "loss": 1.2415,
+      "step": 1313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9284236461625594e-05,
+      "loss": 1.2424,
+      "step": 1314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9283062464759563e-05,
+      "loss": 1.1993,
+      "step": 1315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928188751989219e-05,
+      "loss": 1.1766,
+      "step": 1316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928071162706935e-05,
+      "loss": 1.1966,
+      "step": 1317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927953478633694e-05,
+      "loss": 1.1514,
+      "step": 1318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9278356997740904e-05,
+      "loss": 1.2172,
+      "step": 1319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927717826132723e-05,
+      "loss": 1.2514,
+      "step": 1320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9275998577141926e-05,
+      "loss": 1.2945,
+      "step": 1321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927481794523105e-05,
+      "loss": 1.2186,
+      "step": 1322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92736363656407e-05,
+      "loss": 1.1914,
+      "step": 1323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927245383841699e-05,
+      "loss": 1.3017,
+      "step": 1324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9271270363606106e-05,
+      "loss": 1.2129,
+      "step": 1325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927008594125423e-05,
+      "loss": 1.2632,
+      "step": 1326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9268900571407605e-05,
+      "loss": 1.244,
+      "step": 1327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926771425411252e-05,
+      "loss": 1.272,
+      "step": 1328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926652698941529e-05,
+      "loss": 1.305,
+      "step": 1329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926533877736223e-05,
+      "loss": 1.3177,
+      "step": 1330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926414961799978e-05,
+      "loss": 1.2506,
+      "step": 1331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926295951137433e-05,
+      "loss": 1.2157,
+      "step": 1332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926176845753234e-05,
+      "loss": 1.1942,
+      "step": 1333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926057645652032e-05,
+      "loss": 1.2289,
+      "step": 1334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9259383508384816e-05,
+      "loss": 1.1379,
+      "step": 1335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9258189613172374e-05,
+      "loss": 1.2231,
+      "step": 1336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925699477092962e-05,
+      "loss": 1.17,
+      "step": 1337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9255798981703195e-05,
+      "loss": 1.2035,
+      "step": 1338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925460224553978e-05,
+      "loss": 1.2058,
+      "step": 1339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92534045624861e-05,
+      "loss": 1.2025,
+      "step": 1340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9252205932588916e-05,
+      "loss": 1.2435,
+      "step": 1341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925100635589501e-05,
+      "loss": 1.2503,
+      "step": 1342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9249805832451225e-05,
+      "loss": 1.2873,
+      "step": 1343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924860436230442e-05,
+      "loss": 1.0901,
+      "step": 1344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9247401945501505e-05,
+      "loss": 1.2333,
+      "step": 1345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924619858208942e-05,
+      "loss": 1.2367,
+      "step": 1346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924499427211514e-05,
+      "loss": 1.333,
+      "step": 1347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924378901562569e-05,
+      "loss": 1.224,
+      "step": 1348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9242582812668115e-05,
+      "loss": 1.1973,
+      "step": 1349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9241375663289516e-05,
+      "loss": 1.171,
+      "step": 1350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9240167567537e-05,
+      "loss": 1.2501,
+      "step": 1351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923895852545775e-05,
+      "loss": 1.2229,
+      "step": 1352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9237748537098946e-05,
+      "loss": 1.2908,
+      "step": 1353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923653760250785e-05,
+      "loss": 1.2397,
+      "step": 1354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923532572173172e-05,
+      "loss": 1.2588,
+      "step": 1355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9234112894817874e-05,
+      "loss": 1.3233,
+      "step": 1356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923289912181365e-05,
+      "loss": 1.1808,
+      "step": 1357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923168440276645e-05,
+      "loss": 1.2872,
+      "step": 1358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923046873772368e-05,
+      "loss": 1.1398,
+      "step": 1359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9229252126732814e-05,
+      "loss": 1.1883,
+      "step": 1360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922803456984133e-05,
+      "loss": 1.1644,
+      "step": 1361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9226816067096785e-05,
+      "loss": 1.2501,
+      "step": 1362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922559661854672e-05,
+      "loss": 1.1588,
+      "step": 1363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922437622423877e-05,
+      "loss": 1.1465,
+      "step": 1364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9223154884220554e-05,
+      "loss": 0.9822,
+      "step": 1365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9221932598539764e-05,
+      "loss": 1.128,
+      "step": 1366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922070936724413e-05,
+      "loss": 1.1624,
+      "step": 1367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9219485190381386e-05,
+      "loss": 1.1126,
+      "step": 1368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.921826006799933e-05,
+      "loss": 1.1408,
+      "step": 1369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92170340001458e-05,
+      "loss": 1.0021,
+      "step": 1370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.921580698686865e-05,
+      "loss": 1.1122,
+      "step": 1371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9214579028215776e-05,
+      "loss": 1.3026,
+      "step": 1372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.921335012423514e-05,
+      "loss": 1.2785,
+      "step": 1373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9212120274974695e-05,
+      "loss": 1.3064,
+      "step": 1374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.921088948048247e-05,
+      "loss": 1.313,
+      "step": 1375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9209657740806505e-05,
+      "loss": 1.1851,
+      "step": 1376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.920842505599489e-05,
+      "loss": 1.3118,
+      "step": 1377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9207191426095744e-05,
+      "loss": 1.2644,
+      "step": 1378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.920595685115723e-05,
+      "loss": 1.2661,
+      "step": 1379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9204721331227556e-05,
+      "loss": 1.3191,
+      "step": 1380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9203484866354935e-05,
+      "loss": 1.2343,
+      "step": 1381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.920224745658766e-05,
+      "loss": 1.3829,
+      "step": 1382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.920100910197403e-05,
+      "loss": 1.2474,
+      "step": 1383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.919976980256238e-05,
+      "loss": 1.1951,
+      "step": 1384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.919852955840111e-05,
+      "loss": 1.3436,
+      "step": 1385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.919728836953862e-05,
+      "loss": 1.2464,
+      "step": 1386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9196046236023387e-05,
+      "loss": 1.2892,
+      "step": 1387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.919480315790388e-05,
+      "loss": 1.3456,
+      "step": 1388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9193559135228645e-05,
+      "loss": 1.3114,
+      "step": 1389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.919231416804624e-05,
+      "loss": 1.3202,
+      "step": 1390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.919106825640527e-05,
+      "loss": 1.2688,
+      "step": 1391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9189821400354383e-05,
+      "loss": 1.2396,
+      "step": 1392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.918857359994225e-05,
+      "loss": 1.217,
+      "step": 1393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.918732485521757e-05,
+      "loss": 1.1801,
+      "step": 1394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.918607516622911e-05,
+      "loss": 1.1494,
+      "step": 1395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9184824533025666e-05,
+      "loss": 1.2264,
+      "step": 1396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.918357295565604e-05,
+      "loss": 1.1621,
+      "step": 1397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9182320434169105e-05,
+      "loss": 1.1776,
+      "step": 1398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9181066968613755e-05,
+      "loss": 1.2694,
+      "step": 1399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.917981255903893e-05,
+      "loss": 1.1947,
+      "step": 1400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9178557205493594e-05,
+      "loss": 1.1192,
+      "step": 1401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.917730090802677e-05,
+      "loss": 1.0616,
+      "step": 1402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9176043666687485e-05,
+      "loss": 1.0094,
+      "step": 1403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9174785481524835e-05,
+      "loss": 1.0794,
+      "step": 1404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.917352635258794e-05,
+      "loss": 0.9856,
+      "step": 1405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.917226627992594e-05,
+      "loss": 1.2136,
+      "step": 1406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9171005263588035e-05,
+      "loss": 1.1766,
+      "step": 1407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.916974330362346e-05,
+      "loss": 1.1753,
+      "step": 1408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9168480400081485e-05,
+      "loss": 1.0305,
+      "step": 1409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.91672165530114e-05,
+      "loss": 1.1506,
+      "step": 1410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.916595176246256e-05,
+      "loss": 1.0601,
+      "step": 1411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9164686028484324e-05,
+      "loss": 1.253,
+      "step": 1412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9163419351126114e-05,
+      "loss": 1.3774,
+      "step": 1413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.916215173043739e-05,
+      "loss": 1.3162,
+      "step": 1414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9160883166467635e-05,
+      "loss": 1.3632,
+      "step": 1415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.915961365926637e-05,
+      "loss": 1.374,
+      "step": 1416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.915834320888315e-05,
+      "loss": 1.3285,
+      "step": 1417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9157071815367574e-05,
+      "loss": 1.3165,
+      "step": 1418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.915579947876929e-05,
+      "loss": 1.3378,
+      "step": 1419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9154526199137964e-05,
+      "loss": 1.4049,
+      "step": 1420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.915325197652329e-05,
+      "loss": 1.2609,
+      "step": 1421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.915197681097503e-05,
+      "loss": 1.1322,
+      "step": 1422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.915070070254296e-05,
+      "loss": 1.2135,
+      "step": 1423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.91494236512769e-05,
+      "loss": 1.1478,
+      "step": 1424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.914814565722671e-05,
+      "loss": 1.1874,
+      "step": 1425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9146866720442276e-05,
+      "loss": 1.1531,
+      "step": 1426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9145586840973525e-05,
+      "loss": 1.1152,
+      "step": 1427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.914430601887042e-05,
+      "loss": 1.246,
+      "step": 1428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.914302425418298e-05,
+      "loss": 1.1735,
+      "step": 1429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9141741546961236e-05,
+      "loss": 1.2824,
+      "step": 1430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.914045789725526e-05,
+      "loss": 1.0751,
+      "step": 1431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.913917330511517e-05,
+      "loss": 1.1294,
+      "step": 1432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9137887770591104e-05,
+      "loss": 1.2499,
+      "step": 1433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9136601293733275e-05,
+      "loss": 1.1623,
+      "step": 1434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.913531387459188e-05,
+      "loss": 1.2605,
+      "step": 1435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.91340255132172e-05,
+      "loss": 1.1409,
+      "step": 1436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9132736209659516e-05,
+      "loss": 1.1215,
+      "step": 1437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.913144596396917e-05,
+      "loss": 1.1671,
+      "step": 1438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9130154776196524e-05,
+      "loss": 1.2541,
+      "step": 1439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9128862646392004e-05,
+      "loss": 1.2914,
+      "step": 1440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.912756957460604e-05,
+      "loss": 1.2894,
+      "step": 1441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9126275560889106e-05,
+      "loss": 1.2595,
+      "step": 1442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9124980605291736e-05,
+      "loss": 1.2633,
+      "step": 1443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.912368470786448e-05,
+      "loss": 1.0935,
+      "step": 1444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.912238786865793e-05,
+      "loss": 1.2191,
+      "step": 1445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.91210900877227e-05,
+      "loss": 1.1615,
+      "step": 1446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911979136510947e-05,
+      "loss": 1.1474,
+      "step": 1447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911849170086894e-05,
+      "loss": 1.3068,
+      "step": 1448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911719109505185e-05,
+      "loss": 1.2937,
+      "step": 1449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911588954770897e-05,
+      "loss": 1.1332,
+      "step": 1450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9114587058891114e-05,
+      "loss": 1.207,
+      "step": 1451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911328362864913e-05,
+      "loss": 1.2358,
+      "step": 1452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.91119792570339e-05,
+      "loss": 1.2301,
+      "step": 1453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911067394409635e-05,
+      "loss": 1.2419,
+      "step": 1454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9109367689887435e-05,
+      "loss": 1.2488,
+      "step": 1455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910806049445816e-05,
+      "loss": 1.215,
+      "step": 1456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910675235785955e-05,
+      "loss": 1.2784,
+      "step": 1457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910544328014268e-05,
+      "loss": 1.234,
+      "step": 1458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910413326135864e-05,
+      "loss": 1.2671,
+      "step": 1459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910282230155859e-05,
+      "loss": 1.3491,
+      "step": 1460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9101510400793715e-05,
+      "loss": 1.2935,
+      "step": 1461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9100197559115206e-05,
+      "loss": 1.2709,
+      "step": 1462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909888377657433e-05,
+      "loss": 1.3402,
+      "step": 1463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909756905322238e-05,
+      "loss": 1.2506,
+      "step": 1464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909625338911068e-05,
+      "loss": 1.259,
+      "step": 1465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909493678429059e-05,
+      "loss": 1.3385,
+      "step": 1466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909361923881352e-05,
+      "loss": 1.3064,
+      "step": 1467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909230075273088e-05,
+      "loss": 1.2107,
+      "step": 1468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909098132609417e-05,
+      "loss": 1.252,
+      "step": 1469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908966095895489e-05,
+      "loss": 1.2567,
+      "step": 1470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908833965136459e-05,
+      "loss": 1.2979,
+      "step": 1471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908701740337485e-05,
+      "loss": 1.2238,
+      "step": 1472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908569421503728e-05,
+      "loss": 1.3298,
+      "step": 1473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908437008640356e-05,
+      "loss": 1.2363,
+      "step": 1474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908304501752536e-05,
+      "loss": 1.3238,
+      "step": 1475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908171900845443e-05,
+      "loss": 1.2867,
+      "step": 1476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908039205924253e-05,
+      "loss": 1.3509,
+      "step": 1477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907906416994146e-05,
+      "loss": 1.2887,
+      "step": 1478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9077735340603056e-05,
+      "loss": 1.1188,
+      "step": 1479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9076405571279207e-05,
+      "loss": 0.9362,
+      "step": 1480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9075074862021815e-05,
+      "loss": 0.8526,
+      "step": 1481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907374321288284e-05,
+      "loss": 0.997,
+      "step": 1482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907241062391426e-05,
+      "loss": 1.0302,
+      "step": 1483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9071077095168104e-05,
+      "loss": 0.9577,
+      "step": 1484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906974262669644e-05,
+      "loss": 0.9737,
+      "step": 1485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906840721855135e-05,
+      "loss": 1.0704,
+      "step": 1486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906707087078498e-05,
+      "loss": 0.9918,
+      "step": 1487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906573358344949e-05,
+      "loss": 1.0278,
+      "step": 1488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90643953565971e-05,
+      "loss": 1.102,
+      "step": 1489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9063056190280044e-05,
+      "loss": 1.2036,
+      "step": 1490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90617160845506e-05,
+      "loss": 1.2879,
+      "step": 1491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906037503946109e-05,
+      "loss": 1.2084,
+      "step": 1492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905903305506387e-05,
+      "loss": 1.2288,
+      "step": 1493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9057690131411336e-05,
+      "loss": 1.1346,
+      "step": 1494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90563462685559e-05,
+      "loss": 1.1958,
+      "step": 1495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905500146655004e-05,
+      "loss": 1.1578,
+      "step": 1496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9053655725446244e-05,
+      "loss": 1.162,
+      "step": 1497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905230904529706e-05,
+      "loss": 1.2521,
+      "step": 1498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9050961426155064e-05,
+      "loss": 1.0964,
+      "step": 1499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9049612868072844e-05,
+      "loss": 0.9638,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 1.1715214252471924,
+      "eval_runtime": 115.6326,
+      "eval_samples_per_second": 33.044,
+      "eval_steps_per_second": 0.346,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904826337110308e-05,
+      "loss": 1.0842,
+      "step": 1501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904691293529843e-05,
+      "loss": 1.1033,
+      "step": 1502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904556156071162e-05,
+      "loss": 1.0647,
+      "step": 1503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904420924739542e-05,
+      "loss": 1.1434,
+      "step": 1504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90428559954026e-05,
+      "loss": 1.0757,
+      "step": 1505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9041501804786024e-05,
+      "loss": 1.0751,
+      "step": 1506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904014667559853e-05,
+      "loss": 1.0335,
+      "step": 1507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903879060789303e-05,
+      "loss": 1.0577,
+      "step": 1508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903743360172247e-05,
+      "loss": 1.1671,
+      "step": 1509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9036075657139824e-05,
+      "loss": 1.2525,
+      "step": 1510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90347167741981e-05,
+      "loss": 1.2416,
+      "step": 1511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9033356952950355e-05,
+      "loss": 1.2026,
+      "step": 1512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903199619344967e-05,
+      "loss": 1.1764,
+      "step": 1513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903063449574918e-05,
+      "loss": 1.3234,
+      "step": 1514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9029271859902036e-05,
+      "loss": 1.3579,
+      "step": 1515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9027908285961434e-05,
+      "loss": 1.2467,
+      "step": 1516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9026543773980604e-05,
+      "loss": 1.192,
+      "step": 1517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9025178324012836e-05,
+      "loss": 1.2823,
+      "step": 1518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.902381193611141e-05,
+      "loss": 1.2668,
+      "step": 1519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9022444610329685e-05,
+      "loss": 1.261,
+      "step": 1520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.902107634672104e-05,
+      "loss": 1.2737,
+      "step": 1521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.901970714533889e-05,
+      "loss": 1.2195,
+      "step": 1522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.901833700623668e-05,
+      "loss": 1.1519,
+      "step": 1523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9016965929467914e-05,
+      "loss": 1.1959,
+      "step": 1524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.901559391508611e-05,
+      "loss": 1.2937,
+      "step": 1525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9014220963144824e-05,
+      "loss": 1.2833,
+      "step": 1526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.901284707369767e-05,
+      "loss": 1.2492,
+      "step": 1527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.901147224679828e-05,
+      "loss": 1.1784,
+      "step": 1528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.901009648250032e-05,
+      "loss": 1.2427,
+      "step": 1529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9008719780857505e-05,
+      "loss": 1.3008,
+      "step": 1530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.900734214192358e-05,
+      "loss": 1.2477,
+      "step": 1531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.900596356575232e-05,
+      "loss": 1.229,
+      "step": 1532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9004584052397564e-05,
+      "loss": 1.2684,
+      "step": 1533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.900320360191315e-05,
+      "loss": 1.2061,
+      "step": 1534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.900182221435297e-05,
+      "loss": 1.2489,
+      "step": 1535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.900043988977096e-05,
+      "loss": 1.2603,
+      "step": 1536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8999056628221084e-05,
+      "loss": 1.196,
+      "step": 1537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.899767242975735e-05,
+      "loss": 1.1772,
+      "step": 1538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.899628729443378e-05,
+      "loss": 1.2004,
+      "step": 1539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8994901222304465e-05,
+      "loss": 1.184,
+      "step": 1540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8993514213423506e-05,
+      "loss": 1.2697,
+      "step": 1541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8992126267845065e-05,
+      "loss": 1.1733,
+      "step": 1542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.89907373856233e-05,
+      "loss": 1.2691,
+      "step": 1543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.898934756681246e-05,
+      "loss": 1.0829,
+      "step": 1544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.89879568114668e-05,
+      "loss": 1.3141,
+      "step": 1545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8986565119640606e-05,
+      "loss": 1.1888,
+      "step": 1546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8985172491388204e-05,
+      "loss": 1.2051,
+      "step": 1547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8983778926763965e-05,
+      "loss": 1.1756,
+      "step": 1548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.898238442582231e-05,
+      "loss": 1.2003,
+      "step": 1549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.898098898861766e-05,
+      "loss": 1.2201,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.89795926152045e-05,
+      "loss": 1.2224,
+      "step": 1551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.897819530563734e-05,
+      "loss": 1.17,
+      "step": 1552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.897679705997074e-05,
+      "loss": 1.2789,
+      "step": 1553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8975397878259275e-05,
+      "loss": 1.2477,
+      "step": 1554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8973997760557566e-05,
+      "loss": 1.421,
+      "step": 1555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8972596706920296e-05,
+      "loss": 0.9957,
+      "step": 1556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.897119471740214e-05,
+      "loss": 0.9926,
+      "step": 1557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.896979179205783e-05,
+      "loss": 0.9621,
+      "step": 1558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.896838793094215e-05,
+      "loss": 0.9626,
+      "step": 1559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.89669831341099e-05,
+      "loss": 0.9803,
+      "step": 1560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8965577401615916e-05,
+      "loss": 1.0466,
+      "step": 1561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.896417073351509e-05,
+      "loss": 0.9983,
+      "step": 1562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.896276312986233e-05,
+      "loss": 1.0607,
+      "step": 1563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.896135459071259e-05,
+      "loss": 0.9273,
+      "step": 1564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.895994511612085e-05,
+      "loss": 1.0456,
+      "step": 1565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.895853470614216e-05,
+      "loss": 1.012,
+      "step": 1566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8957123360831545e-05,
+      "loss": 1.2833,
+      "step": 1567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.895571108024414e-05,
+      "loss": 1.2758,
+      "step": 1568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8954297864435054e-05,
+      "loss": 1.2579,
+      "step": 1569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8952883713459474e-05,
+      "loss": 1.1925,
+      "step": 1570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.89514686273726e-05,
+      "loss": 1.284,
+      "step": 1571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.895005260622968e-05,
+      "loss": 1.2709,
+      "step": 1572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8948635650085986e-05,
+      "loss": 1.2054,
+      "step": 1573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.894721775899685e-05,
+      "loss": 1.3113,
+      "step": 1574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8945798933017616e-05,
+      "loss": 1.2874,
+      "step": 1575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.894437917220368e-05,
+      "loss": 1.057,
+      "step": 1576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8942958476610464e-05,
+      "loss": 1.0438,
+      "step": 1577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.894153684629344e-05,
+      "loss": 1.1217,
+      "step": 1578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8940114281308086e-05,
+      "loss": 1.1827,
+      "step": 1579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.893869078170996e-05,
+      "loss": 1.0401,
+      "step": 1580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.893726634755463e-05,
+      "loss": 1.0213,
+      "step": 1581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8935840978897704e-05,
+      "loss": 1.1649,
+      "step": 1582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8934414675794824e-05,
+      "loss": 1.0705,
+      "step": 1583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.893298743830168e-05,
+      "loss": 1.0515,
+      "step": 1584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.893155926647399e-05,
+      "loss": 1.0352,
+      "step": 1585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8930130160367504e-05,
+      "loss": 1.1603,
+      "step": 1586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.892870012003801e-05,
+      "loss": 1.3337,
+      "step": 1587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.892726914554135e-05,
+      "loss": 1.2521,
+      "step": 1588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.892583723693338e-05,
+      "loss": 1.2358,
+      "step": 1589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.892440439427e-05,
+      "loss": 1.308,
+      "step": 1590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8922970617607145e-05,
+      "loss": 1.3269,
+      "step": 1591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.892153590700079e-05,
+      "loss": 1.2729,
+      "step": 1592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.892010026250696e-05,
+      "loss": 1.2303,
+      "step": 1593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.891866368418169e-05,
+      "loss": 1.3066,
+      "step": 1594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.891722617208106e-05,
+      "loss": 1.3778,
+      "step": 1595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8915787726261195e-05,
+      "loss": 1.2405,
+      "step": 1596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8914348346778255e-05,
+      "loss": 1.181,
+      "step": 1597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8912908033688424e-05,
+      "loss": 1.153,
+      "step": 1598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8911466787047945e-05,
+      "loss": 1.1821,
+      "step": 1599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.891002460691306e-05,
+      "loss": 1.2579,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.89085814933401e-05,
+      "loss": 1.2063,
+      "step": 1601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.890713744638537e-05,
+      "loss": 1.232,
+      "step": 1602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.890569246610529e-05,
+      "loss": 1.2103,
+      "step": 1603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8904246552556235e-05,
+      "loss": 1.1799,
+      "step": 1604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8902799705794655e-05,
+      "loss": 1.2779,
+      "step": 1605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8901351925877056e-05,
+      "loss": 1.2502,
+      "step": 1606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.889990321285994e-05,
+      "loss": 1.3259,
+      "step": 1607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8898453566799867e-05,
+      "loss": 1.3388,
+      "step": 1608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.889700298775344e-05,
+      "loss": 1.3332,
+      "step": 1609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.889555147577728e-05,
+      "loss": 1.3291,
+      "step": 1610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8894099030928053e-05,
+      "loss": 1.2344,
+      "step": 1611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.889264565326247e-05,
+      "loss": 1.229,
+      "step": 1612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.889119134283726e-05,
+      "loss": 1.3173,
+      "step": 1613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888973609970921e-05,
+      "loss": 1.2383,
+      "step": 1614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888827992393513e-05,
+      "loss": 1.2201,
+      "step": 1615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8886822815571854e-05,
+      "loss": 1.2951,
+      "step": 1616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888536477467628e-05,
+      "loss": 1.2755,
+      "step": 1617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8883905801305326e-05,
+      "loss": 1.3328,
+      "step": 1618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888244589551596e-05,
+      "loss": 1.2268,
+      "step": 1619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8880985057365165e-05,
+      "loss": 1.3352,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887952328690997e-05,
+      "loss": 1.1639,
+      "step": 1621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887806058420744e-05,
+      "loss": 1.2972,
+      "step": 1622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.88765969493147e-05,
+      "loss": 1.2246,
+      "step": 1623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8875132382288856e-05,
+      "loss": 1.1766,
+      "step": 1624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8873666883187105e-05,
+      "loss": 1.1821,
+      "step": 1625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8872200452066664e-05,
+      "loss": 1.2863,
+      "step": 1626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887073308898478e-05,
+      "loss": 1.1408,
+      "step": 1627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8869264793998715e-05,
+      "loss": 1.2523,
+      "step": 1628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.886779556716582e-05,
+      "loss": 1.1252,
+      "step": 1629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.886632540854344e-05,
+      "loss": 1.2117,
+      "step": 1630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8864854318188965e-05,
+      "loss": 1.3393,
+      "step": 1631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.886338229615983e-05,
+      "loss": 1.2669,
+      "step": 1632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.886190934251351e-05,
+      "loss": 1.2676,
+      "step": 1633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8860435457307504e-05,
+      "loss": 1.262,
+      "step": 1634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885896064059934e-05,
+      "loss": 1.2422,
+      "step": 1635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885748489244661e-05,
+      "loss": 1.2623,
+      "step": 1636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8856008212906925e-05,
+      "loss": 1.2701,
+      "step": 1637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885453060203793e-05,
+      "loss": 1.2117,
+      "step": 1638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.88530520598973e-05,
+      "loss": 1.2544,
+      "step": 1639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885157258654277e-05,
+      "loss": 1.3229,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.88500921820321e-05,
+      "loss": 1.2551,
+      "step": 1641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884861084642308e-05,
+      "loss": 1.2507,
+      "step": 1642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884712857977354e-05,
+      "loss": 1.2652,
+      "step": 1643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884564538214135e-05,
+      "loss": 1.178,
+      "step": 1644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8844161253584394e-05,
+      "loss": 1.1962,
+      "step": 1645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8842676194160646e-05,
+      "loss": 1.1371,
+      "step": 1646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884119020392806e-05,
+      "loss": 1.2008,
+      "step": 1647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8839703282944655e-05,
+      "loss": 1.2191,
+      "step": 1648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8838215431268476e-05,
+      "loss": 1.206,
+      "step": 1649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883672664895761e-05,
+      "loss": 1.0941,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883523693607018e-05,
+      "loss": 1.3185,
+      "step": 1651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883374629266434e-05,
+      "loss": 1.2537,
+      "step": 1652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883225471879829e-05,
+      "loss": 1.2873,
+      "step": 1653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883076221453025e-05,
+      "loss": 1.1497,
+      "step": 1654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8829268779918504e-05,
+      "loss": 1.3056,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882777441502134e-05,
+      "loss": 1.2549,
+      "step": 1656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.88262791198971e-05,
+      "loss": 1.261,
+      "step": 1657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8824782894604165e-05,
+      "loss": 1.1967,
+      "step": 1658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882328573920094e-05,
+      "loss": 1.1967,
+      "step": 1659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882178765374589e-05,
+      "loss": 1.128,
+      "step": 1660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882028863829747e-05,
+      "loss": 1.3015,
+      "step": 1661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881878869291423e-05,
+      "loss": 1.3,
+      "step": 1662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881728781765471e-05,
+      "loss": 1.3389,
+      "step": 1663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8815786012577514e-05,
+      "loss": 1.2989,
+      "step": 1664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881428327774126e-05,
+      "loss": 1.259,
+      "step": 1665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881277961320463e-05,
+      "loss": 1.2549,
+      "step": 1666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8811275019026315e-05,
+      "loss": 1.2748,
+      "step": 1667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880976949526505e-05,
+      "loss": 1.2486,
+      "step": 1668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880826304197962e-05,
+      "loss": 1.2342,
+      "step": 1669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880675565922884e-05,
+      "loss": 1.2524,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880524734707154e-05,
+      "loss": 1.0758,
+      "step": 1671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880373810556662e-05,
+      "loss": 1.0869,
+      "step": 1672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880222793477299e-05,
+      "loss": 0.9034,
+      "step": 1673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880071683474962e-05,
+      "loss": 1.0015,
+      "step": 1674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.879920480555549e-05,
+      "loss": 1.0531,
+      "step": 1675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.879769184724963e-05,
+      "loss": 1.0182,
+      "step": 1676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.879617795989111e-05,
+      "loss": 1.0617,
+      "step": 1677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.879466314353903e-05,
+      "loss": 1.0378,
+      "step": 1678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8793147398252524e-05,
+      "loss": 0.9652,
+      "step": 1679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8791630724090775e-05,
+      "loss": 1.0268,
+      "step": 1680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8790113121112994e-05,
+      "loss": 1.2387,
+      "step": 1681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.878859458937841e-05,
+      "loss": 1.1884,
+      "step": 1682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.878707512894632e-05,
+      "loss": 1.2141,
+      "step": 1683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.878555473987604e-05,
+      "loss": 1.1491,
+      "step": 1684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8784033422226926e-05,
+      "loss": 1.1629,
+      "step": 1685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.878251117605838e-05,
+      "loss": 1.149,
+      "step": 1686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8780988001429805e-05,
+      "loss": 1.1894,
+      "step": 1687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.877946389840069e-05,
+      "loss": 1.2635,
+      "step": 1688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8777938867030524e-05,
+      "loss": 1.2657,
+      "step": 1689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.877641290737884e-05,
+      "loss": 1.1807,
+      "step": 1690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.877488601950522e-05,
+      "loss": 1.2265,
+      "step": 1691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.877335820346927e-05,
+      "loss": 1.2823,
+      "step": 1692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.877182945933063e-05,
+      "loss": 1.1975,
+      "step": 1693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8770299787148985e-05,
+      "loss": 1.3901,
+      "step": 1694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8768769186984056e-05,
+      "loss": 1.2473,
+      "step": 1695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.87672376588956e-05,
+      "loss": 1.2979,
+      "step": 1696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8765705202943395e-05,
+      "loss": 1.3198,
+      "step": 1697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.876417181918728e-05,
+      "loss": 1.2524,
+      "step": 1698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.87626375076871e-05,
+      "loss": 1.3212,
+      "step": 1699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.876110226850278e-05,
+      "loss": 1.2352,
+      "step": 1700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8759566101694234e-05,
+      "loss": 1.2125,
+      "step": 1701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8758029007321444e-05,
+      "loss": 1.1581,
+      "step": 1702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.875649098544441e-05,
+      "loss": 1.242,
+      "step": 1703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.875495203612318e-05,
+      "loss": 1.2744,
+      "step": 1704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.875341215941784e-05,
+      "loss": 1.2236,
+      "step": 1705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.875187135538849e-05,
+      "loss": 1.2988,
+      "step": 1706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8750329624095304e-05,
+      "loss": 1.1699,
+      "step": 1707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8748786965598456e-05,
+      "loss": 1.1472,
+      "step": 1708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.874724337995817e-05,
+      "loss": 1.1935,
+      "step": 1709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8745698867234715e-05,
+      "loss": 1.1503,
+      "step": 1710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.874415342748838e-05,
+      "loss": 1.1327,
+      "step": 1711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.87426070607795e-05,
+      "loss": 1.0933,
+      "step": 1712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.874105976716845e-05,
+      "loss": 1.1283,
+      "step": 1713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8739511546715634e-05,
+      "loss": 1.0912,
+      "step": 1714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.873796239948149e-05,
+      "loss": 1.1353,
+      "step": 1715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.87364123255265e-05,
+      "loss": 1.0992,
+      "step": 1716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.873486132491118e-05,
+      "loss": 1.157,
+      "step": 1717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.873330939769608e-05,
+      "loss": 1.0334,
+      "step": 1718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8731756543941776e-05,
+      "loss": 1.0705,
+      "step": 1719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.87302027637089e-05,
+      "loss": 1.0397,
+      "step": 1720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8728648057058116e-05,
+      "loss": 0.9878,
+      "step": 1721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8727092424050106e-05,
+      "loss": 1.0462,
+      "step": 1722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.872553586474561e-05,
+      "loss": 1.1208,
+      "step": 1723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.872397837920539e-05,
+      "loss": 1.2065,
+      "step": 1724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.872241996749026e-05,
+      "loss": 0.9399,
+      "step": 1725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.872086062966105e-05,
+      "loss": 1.1194,
+      "step": 1726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.871930036577864e-05,
+      "loss": 1.1132,
+      "step": 1727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8717739175903945e-05,
+      "loss": 1.1171,
+      "step": 1728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8716177060097906e-05,
+      "loss": 1.1162,
+      "step": 1729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.871461401842151e-05,
+      "loss": 1.1944,
+      "step": 1730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.871305005093578e-05,
+      "loss": 1.1883,
+      "step": 1731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.871148515770177e-05,
+      "loss": 1.0654,
+      "step": 1732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8709919338780576e-05,
+      "loss": 1.1296,
+      "step": 1733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.870835259423332e-05,
+      "loss": 1.2284,
+      "step": 1734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8706784924121174e-05,
+      "loss": 1.0452,
+      "step": 1735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.870521632850534e-05,
+      "loss": 1.2769,
+      "step": 1736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.870364680744706e-05,
+      "loss": 1.1743,
+      "step": 1737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.870207636100759e-05,
+      "loss": 1.2343,
+      "step": 1738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8700504989248244e-05,
+      "loss": 1.2852,
+      "step": 1739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.869893269223038e-05,
+      "loss": 1.2464,
+      "step": 1740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.869735947001538e-05,
+      "loss": 1.2541,
+      "step": 1741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8695785322664646e-05,
+      "loss": 1.3374,
+      "step": 1742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.869421025023965e-05,
+      "loss": 1.2213,
+      "step": 1743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8692634252801864e-05,
+      "loss": 1.2456,
+      "step": 1744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.869105733041283e-05,
+      "loss": 1.2359,
+      "step": 1745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8689479483134106e-05,
+      "loss": 1.255,
+      "step": 1746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.868790071102728e-05,
+      "loss": 1.2414,
+      "step": 1747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8686321014154e-05,
+      "loss": 1.1873,
+      "step": 1748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.868474039257593e-05,
+      "loss": 1.1814,
+      "step": 1749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8683158846354786e-05,
+      "loss": 1.2318,
+      "step": 1750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8681576375552304e-05,
+      "loss": 1.1932,
+      "step": 1751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.867999298023026e-05,
+      "loss": 1.2826,
+      "step": 1752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.867840866045047e-05,
+      "loss": 1.2009,
+      "step": 1753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8676823416274787e-05,
+      "loss": 1.1578,
+      "step": 1754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8675237247765105e-05,
+      "loss": 1.2124,
+      "step": 1755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8673650154983344e-05,
+      "loss": 1.2527,
+      "step": 1756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8672062137991446e-05,
+      "loss": 1.2974,
+      "step": 1757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.867047319685143e-05,
+      "loss": 1.236,
+      "step": 1758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.866888333162532e-05,
+      "loss": 1.3,
+      "step": 1759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8667292542375185e-05,
+      "loss": 1.2681,
+      "step": 1760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8665700829163124e-05,
+      "loss": 1.2185,
+      "step": 1761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.866410819205128e-05,
+      "loss": 1.2216,
+      "step": 1762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.866251463110182e-05,
+      "loss": 1.3234,
+      "step": 1763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.866092014637698e-05,
+      "loss": 1.2458,
+      "step": 1764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.865932473793898e-05,
+      "loss": 1.1759,
+      "step": 1765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.865772840585012e-05,
+      "loss": 1.2892,
+      "step": 1766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8656131150172716e-05,
+      "loss": 1.2014,
+      "step": 1767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.865453297096912e-05,
+      "loss": 1.2041,
+      "step": 1768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.865293386830174e-05,
+      "loss": 1.2327,
+      "step": 1769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.865133384223299e-05,
+      "loss": 1.2385,
+      "step": 1770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864973289282534e-05,
+      "loss": 1.2244,
+      "step": 1771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864813102014128e-05,
+      "loss": 1.2436,
+      "step": 1772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8646528224243366e-05,
+      "loss": 1.2269,
+      "step": 1773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864492450519415e-05,
+      "loss": 1.2024,
+      "step": 1774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864331986305626e-05,
+      "loss": 1.2105,
+      "step": 1775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8641714297892316e-05,
+      "loss": 1.108,
+      "step": 1776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8640107809765023e-05,
+      "loss": 1.0916,
+      "step": 1777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8638500398737086e-05,
+      "loss": 1.1452,
+      "step": 1778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8636892064871265e-05,
+      "loss": 0.9775,
+      "step": 1779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.863528280823033e-05,
+      "loss": 0.9804,
+      "step": 1780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.863367262887714e-05,
+      "loss": 0.9866,
+      "step": 1781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8632061526874515e-05,
+      "loss": 1.1388,
+      "step": 1782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.863044950228538e-05,
+      "loss": 1.1706,
+      "step": 1783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862883655517266e-05,
+      "loss": 1.0752,
+      "step": 1784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862722268559932e-05,
+      "loss": 1.203,
+      "step": 1785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862560789362837e-05,
+      "loss": 0.9733,
+      "step": 1786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862399217932285e-05,
+      "loss": 1.0756,
+      "step": 1787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8622375542745836e-05,
+      "loss": 1.1052,
+      "step": 1788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862075798396044e-05,
+      "loss": 1.1027,
+      "step": 1789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861913950302981e-05,
+      "loss": 1.1352,
+      "step": 1790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861752010001713e-05,
+      "loss": 1.0056,
+      "step": 1791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8615899774985626e-05,
+      "loss": 1.1306,
+      "step": 1792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8614278527998556e-05,
+      "loss": 1.0495,
+      "step": 1793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861265635911921e-05,
+      "loss": 1.1506,
+      "step": 1794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8611033268410914e-05,
+      "loss": 1.1284,
+      "step": 1795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.860940925593703e-05,
+      "loss": 1.0309,
+      "step": 1796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8607784321760964e-05,
+      "loss": 1.0541,
+      "step": 1797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.860615846594616e-05,
+      "loss": 1.071,
+      "step": 1798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8604531688556075e-05,
+      "loss": 1.1155,
+      "step": 1799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8602903989654224e-05,
+      "loss": 1.0935,
+      "step": 1800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8601275369304166e-05,
+      "loss": 1.1616,
+      "step": 1801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.859964582756946e-05,
+      "loss": 0.96,
+      "step": 1802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.859801536451373e-05,
+      "loss": 0.9915,
+      "step": 1803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8596383980200636e-05,
+      "loss": 1.1724,
+      "step": 1804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.859475167469386e-05,
+      "loss": 0.9781,
+      "step": 1805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.859311844805713e-05,
+      "loss": 1.1854,
+      "step": 1806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8591484300354204e-05,
+      "loss": 1.0782,
+      "step": 1807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8589849231648875e-05,
+      "loss": 1.1341,
+      "step": 1808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858821324200498e-05,
+      "loss": 1.1739,
+      "step": 1809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8586576331486386e-05,
+      "loss": 1.2126,
+      "step": 1810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8584938500157004e-05,
+      "loss": 1.2214,
+      "step": 1811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8583299748080766e-05,
+      "loss": 1.2116,
+      "step": 1812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858166007532165e-05,
+      "loss": 1.1915,
+      "step": 1813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8580019481943666e-05,
+      "loss": 1.1324,
+      "step": 1814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857837796801087e-05,
+      "loss": 1.2157,
+      "step": 1815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8576735533587336e-05,
+      "loss": 1.133,
+      "step": 1816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857509217873719e-05,
+      "loss": 1.188,
+      "step": 1817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.85734479035246e-05,
+      "loss": 1.2286,
+      "step": 1818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8571802708013726e-05,
+      "loss": 1.0658,
+      "step": 1819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857015659226882e-05,
+      "loss": 1.181,
+      "step": 1820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8568509556354136e-05,
+      "loss": 1.1794,
+      "step": 1821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856686160033399e-05,
+      "loss": 1.1831,
+      "step": 1822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.85652127242727e-05,
+      "loss": 1.1976,
+      "step": 1823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856356292823464e-05,
+      "loss": 1.1502,
+      "step": 1824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8561912212284223e-05,
+      "loss": 1.1762,
+      "step": 1825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856026057648589e-05,
+      "loss": 1.2068,
+      "step": 1826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855860802090412e-05,
+      "loss": 1.104,
+      "step": 1827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855695454560343e-05,
+      "loss": 1.2113,
+      "step": 1828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8555300150648366e-05,
+      "loss": 1.1505,
+      "step": 1829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855364483610352e-05,
+      "loss": 1.2429,
+      "step": 1830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8551988602033505e-05,
+      "loss": 1.2502,
+      "step": 1831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8550331448503e-05,
+      "loss": 1.2517,
+      "step": 1832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.854867337557668e-05,
+      "loss": 1.2179,
+      "step": 1833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.854701438331928e-05,
+      "loss": 1.1893,
+      "step": 1834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.854535447179558e-05,
+      "loss": 1.2568,
+      "step": 1835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.854369364107036e-05,
+      "loss": 1.2826,
+      "step": 1836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8542031891208475e-05,
+      "loss": 1.2631,
+      "step": 1837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.85403692222748e-05,
+      "loss": 1.23,
+      "step": 1838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.853870563433422e-05,
+      "loss": 1.2898,
+      "step": 1839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.853704112745172e-05,
+      "loss": 1.1406,
+      "step": 1840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.853537570169225e-05,
+      "loss": 1.2298,
+      "step": 1841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8533709357120844e-05,
+      "loss": 1.3206,
+      "step": 1842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8532042093802547e-05,
+      "loss": 1.269,
+      "step": 1843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.853037391180245e-05,
+      "loss": 1.2133,
+      "step": 1844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8528704811185686e-05,
+      "loss": 1.2873,
+      "step": 1845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.85270347920174e-05,
+      "loss": 1.2273,
+      "step": 1846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.852536385436281e-05,
+      "loss": 1.1662,
+      "step": 1847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.852369199828713e-05,
+      "loss": 1.2832,
+      "step": 1848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.852201922385564e-05,
+      "loss": 1.2362,
+      "step": 1849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8520345531133636e-05,
+      "loss": 1.2221,
+      "step": 1850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.851867092018647e-05,
+      "loss": 1.366,
+      "step": 1851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.85169953910795e-05,
+      "loss": 1.3632,
+      "step": 1852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.851531894387816e-05,
+      "loss": 1.4274,
+      "step": 1853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.851364157864788e-05,
+      "loss": 1.4226,
+      "step": 1854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8511963295454156e-05,
+      "loss": 1.3457,
+      "step": 1855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8510284094362504e-05,
+      "loss": 1.3742,
+      "step": 1856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.850860397543847e-05,
+      "loss": 1.4133,
+      "step": 1857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.850692293874766e-05,
+      "loss": 1.4019,
+      "step": 1858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8505240984355684e-05,
+      "loss": 1.3445,
+      "step": 1859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8503558112328217e-05,
+      "loss": 1.3643,
+      "step": 1860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.850187432273096e-05,
+      "loss": 1.3674,
+      "step": 1861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8500189615629635e-05,
+      "loss": 1.2783,
+      "step": 1862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.849850399109003e-05,
+      "loss": 1.1438,
+      "step": 1863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.849681744917793e-05,
+      "loss": 1.1495,
+      "step": 1864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8495129989959195e-05,
+      "loss": 1.2421,
+      "step": 1865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.84934416134997e-05,
+      "loss": 1.2499,
+      "step": 1866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.849175231986535e-05,
+      "loss": 1.3076,
+      "step": 1867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.84900621091221e-05,
+      "loss": 1.2315,
+      "step": 1868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.848837098133593e-05,
+      "loss": 1.2664,
+      "step": 1869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8486678936572865e-05,
+      "loss": 1.3676,
+      "step": 1870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.848498597489897e-05,
+      "loss": 1.3724,
+      "step": 1871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.848329209638032e-05,
+      "loss": 1.3836,
+      "step": 1872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.848159730108306e-05,
+      "loss": 1.3702,
+      "step": 1873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.847990158907333e-05,
+      "loss": 1.4071,
+      "step": 1874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8478204960417364e-05,
+      "loss": 1.4315,
+      "step": 1875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.847650741518137e-05,
+      "loss": 1.3625,
+      "step": 1876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8474808953431636e-05,
+      "loss": 1.3044,
+      "step": 1877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.847310957523447e-05,
+      "loss": 1.2712,
+      "step": 1878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.84714092806562e-05,
+      "loss": 1.3813,
+      "step": 1879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.846970806976321e-05,
+      "loss": 1.1322,
+      "step": 1880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.846800594262192e-05,
+      "loss": 0.8889,
+      "step": 1881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.846630289929878e-05,
+      "loss": 0.9454,
+      "step": 1882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8464598939860275e-05,
+      "loss": 0.992,
+      "step": 1883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.846289406437292e-05,
+      "loss": 1.0962,
+      "step": 1884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.846118827290329e-05,
+      "loss": 0.8882,
+      "step": 1885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.845948156551796e-05,
+      "loss": 0.9792,
+      "step": 1886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8457773942283566e-05,
+      "loss": 0.9584,
+      "step": 1887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.845606540326678e-05,
+      "loss": 0.9973,
+      "step": 1888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8454355948534286e-05,
+      "loss": 1.0522,
+      "step": 1889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8452645578152835e-05,
+      "loss": 1.0142,
+      "step": 1890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.84509342921892e-05,
+      "loss": 1.0207,
+      "step": 1891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.844922209071018e-05,
+      "loss": 1.1764,
+      "step": 1892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.844750897378262e-05,
+      "loss": 1.1003,
+      "step": 1893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.84457949414734e-05,
+      "loss": 1.098,
+      "step": 1894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.844407999384944e-05,
+      "loss": 1.0831,
+      "step": 1895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8442364130977684e-05,
+      "loss": 1.0958,
+      "step": 1896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.844064735292513e-05,
+      "loss": 1.1431,
+      "step": 1897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8438929659758795e-05,
+      "loss": 1.1244,
+      "step": 1898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8437211051545724e-05,
+      "loss": 1.0423,
+      "step": 1899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8435491528353026e-05,
+      "loss": 1.1765,
+      "step": 1900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.843377109024783e-05,
+      "loss": 1.1546,
+      "step": 1901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.843204973729729e-05,
+      "loss": 1.2501,
+      "step": 1902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8430327469568615e-05,
+      "loss": 1.2324,
+      "step": 1903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.842860428712905e-05,
+      "loss": 1.2647,
+      "step": 1904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8426880190045856e-05,
+      "loss": 1.26,
+      "step": 1905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.842515517838634e-05,
+      "loss": 1.1963,
+      "step": 1906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8423429252217844e-05,
+      "loss": 1.1815,
+      "step": 1907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.842170241160776e-05,
+      "loss": 1.314,
+      "step": 1908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8419974656623494e-05,
+      "loss": 1.1925,
+      "step": 1909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.84182459873325e-05,
+      "loss": 0.9704,
+      "step": 1910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8416516403802266e-05,
+      "loss": 1.075,
+      "step": 1911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.841478590610031e-05,
+      "loss": 1.1766,
+      "step": 1912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.841305449429419e-05,
+      "loss": 1.1207,
+      "step": 1913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.84113221684515e-05,
+      "loss": 1.0511,
+      "step": 1914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.840958892863987e-05,
+      "loss": 1.0919,
+      "step": 1915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.840785477492697e-05,
+      "loss": 1.0296,
+      "step": 1916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.840611970738049e-05,
+      "loss": 1.0672,
+      "step": 1917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.840438372606818e-05,
+      "loss": 1.1473,
+      "step": 1918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8402646831057806e-05,
+      "loss": 1.0963,
+      "step": 1919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.840090902241717e-05,
+      "loss": 1.2139,
+      "step": 1920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.839917030021412e-05,
+      "loss": 1.28,
+      "step": 1921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.839743066451654e-05,
+      "loss": 1.3231,
+      "step": 1922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8395690115392336e-05,
+      "loss": 1.1099,
+      "step": 1923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.839394865290946e-05,
+      "loss": 1.2864,
+      "step": 1924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8392206277135896e-05,
+      "loss": 1.3296,
+      "step": 1925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.839046298813967e-05,
+      "loss": 1.1511,
+      "step": 1926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.838871878598885e-05,
+      "loss": 1.1511,
+      "step": 1927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8386973670751515e-05,
+      "loss": 1.1653,
+      "step": 1928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.838522764249579e-05,
+      "loss": 1.1678,
+      "step": 1929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.838348070128984e-05,
+      "loss": 1.2306,
+      "step": 1930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8381732847201885e-05,
+      "loss": 1.2765,
+      "step": 1931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837998408030014e-05,
+      "loss": 1.1731,
+      "step": 1932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837823440065288e-05,
+      "loss": 1.2915,
+      "step": 1933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837648380832842e-05,
+      "loss": 1.241,
+      "step": 1934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837473230339509e-05,
+      "loss": 1.2029,
+      "step": 1935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837297988592128e-05,
+      "loss": 1.2018,
+      "step": 1936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8371226555975404e-05,
+      "loss": 1.2909,
+      "step": 1937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83694723136259e-05,
+      "loss": 1.2045,
+      "step": 1938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836771715894125e-05,
+      "loss": 1.1914,
+      "step": 1939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836596109199e-05,
+      "loss": 1.1127,
+      "step": 1940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836420411284067e-05,
+      "loss": 1.201,
+      "step": 1941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8362446221561886e-05,
+      "loss": 1.1138,
+      "step": 1942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836068741822226e-05,
+      "loss": 1.2107,
+      "step": 1943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835892770289045e-05,
+      "loss": 1.1924,
+      "step": 1944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8357167075635165e-05,
+      "loss": 1.1441,
+      "step": 1945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835540553652513e-05,
+      "loss": 1.2279,
+      "step": 1946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835364308562912e-05,
+      "loss": 1.1087,
+      "step": 1947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835187972301594e-05,
+      "loss": 1.1369,
+      "step": 1948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835011544875443e-05,
+      "loss": 1.2925,
+      "step": 1949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.834835026291347e-05,
+      "loss": 1.3217,
+      "step": 1950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.834658416556197e-05,
+      "loss": 1.247,
+      "step": 1951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8344817156768876e-05,
+      "loss": 1.2211,
+      "step": 1952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.834304923660317e-05,
+      "loss": 1.3197,
+      "step": 1953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8341280405133876e-05,
+      "loss": 1.3154,
+      "step": 1954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8339510662430046e-05,
+      "loss": 1.2311,
+      "step": 1955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833774000856077e-05,
+      "loss": 1.3632,
+      "step": 1956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833596844359517e-05,
+      "loss": 1.3526,
+      "step": 1957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833419596760241e-05,
+      "loss": 1.1611,
+      "step": 1958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8332422580651695e-05,
+      "loss": 1.0915,
+      "step": 1959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833064828281225e-05,
+      "loss": 1.0128,
+      "step": 1960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8328873074153334e-05,
+      "loss": 1.071,
+      "step": 1961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.832709695474427e-05,
+      "loss": 1.1287,
+      "step": 1962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.832531992465438e-05,
+      "loss": 1.169,
+      "step": 1963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.832354198395305e-05,
+      "loss": 1.0622,
+      "step": 1964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8321763132709674e-05,
+      "loss": 1.1294,
+      "step": 1965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8319983370993714e-05,
+      "loss": 1.0601,
+      "step": 1966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.831820269887465e-05,
+      "loss": 1.0905,
+      "step": 1967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8316421116421996e-05,
+      "loss": 1.083,
+      "step": 1968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83146386237053e-05,
+      "loss": 1.2126,
+      "step": 1969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8312855220794154e-05,
+      "loss": 1.3116,
+      "step": 1970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.831107090775817e-05,
+      "loss": 1.2355,
+      "step": 1971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8309285684667025e-05,
+      "loss": 1.2153,
+      "step": 1972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8307499551590405e-05,
+      "loss": 1.2901,
+      "step": 1973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.830571250859805e-05,
+      "loss": 1.2557,
+      "step": 1974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83039245557597e-05,
+      "loss": 1.2233,
+      "step": 1975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8302135693145176e-05,
+      "loss": 1.1863,
+      "step": 1976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8300345920824315e-05,
+      "loss": 1.2745,
+      "step": 1977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.829855523886698e-05,
+      "loss": 1.1712,
+      "step": 1978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.829676364734309e-05,
+      "loss": 1.1753,
+      "step": 1979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.829497114632257e-05,
+      "loss": 1.0971,
+      "step": 1980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8293177735875415e-05,
+      "loss": 1.1747,
+      "step": 1981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.829138341607163e-05,
+      "loss": 1.0897,
+      "step": 1982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8289588186981276e-05,
+      "loss": 1.0126,
+      "step": 1983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828779204867443e-05,
+      "loss": 1.1473,
+      "step": 1984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828599500122121e-05,
+      "loss": 1.1464,
+      "step": 1985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828419704469177e-05,
+      "loss": 1.0941,
+      "step": 1986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828239817915631e-05,
+      "loss": 1.1378,
+      "step": 1987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828059840468505e-05,
+      "loss": 1.1911,
+      "step": 1988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8278797721348255e-05,
+      "loss": 1.1234,
+      "step": 1989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8276996129216226e-05,
+      "loss": 1.155,
+      "step": 1990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827519362835929e-05,
+      "loss": 1.1089,
+      "step": 1991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827339021884783e-05,
+      "loss": 1.1452,
+      "step": 1992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827158590075224e-05,
+      "loss": 1.1072,
+      "step": 1993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8269780674142954e-05,
+      "loss": 1.1202,
+      "step": 1994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8267974539090455e-05,
+      "loss": 1.0126,
+      "step": 1995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.826616749566525e-05,
+      "loss": 1.1834,
+      "step": 1996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8264359543937894e-05,
+      "loss": 1.1084,
+      "step": 1997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.826255068397896e-05,
+      "loss": 1.2276,
+      "step": 1998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.826074091585906e-05,
+      "loss": 1.2824,
+      "step": 1999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8258930239648865e-05,
+      "loss": 1.2694,
+      "step": 2000
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 1.1724560260772705,
+      "eval_runtime": 113.3453,
+      "eval_samples_per_second": 33.711,
+      "eval_steps_per_second": 0.353,
+      "step": 2000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.825711865541905e-05,
+      "loss": 1.3017,
+      "step": 2001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.825530616324034e-05,
+      "loss": 1.2483,
+      "step": 2002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8253492763183495e-05,
+      "loss": 1.3119,
+      "step": 2003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8251678455319305e-05,
+      "loss": 1.209,
+      "step": 2004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8249863239718604e-05,
+      "loss": 1.1403,
+      "step": 2005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.824804711645227e-05,
+      "loss": 1.3011,
+      "step": 2006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.824623008559118e-05,
+      "loss": 1.3073,
+      "step": 2007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8244412147206284e-05,
+      "loss": 1.2412,
+      "step": 2008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8242593301368555e-05,
+      "loss": 1.0806,
+      "step": 2009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.824077354814899e-05,
+      "loss": 1.1144,
+      "step": 2010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8238952887618645e-05,
+      "loss": 1.0707,
+      "step": 2011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.823713131984858e-05,
+      "loss": 1.1564,
+      "step": 2012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8235308844909934e-05,
+      "loss": 1.0573,
+      "step": 2013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.823348546287383e-05,
+      "loss": 1.0029,
+      "step": 2014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.823166117381147e-05,
+      "loss": 1.0928,
+      "step": 2015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.822983597779406e-05,
+      "loss": 0.9539,
+      "step": 2016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8228009874892855e-05,
+      "loss": 1.0455,
+      "step": 2017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.822618286517916e-05,
+      "loss": 1.0456,
+      "step": 2018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8224354948724294e-05,
+      "loss": 1.1875,
+      "step": 2019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.822252612559961e-05,
+      "loss": 1.2651,
+      "step": 2020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8220696395876514e-05,
+      "loss": 1.1974,
+      "step": 2021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8218865759626436e-05,
+      "loss": 1.2287,
+      "step": 2022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8217034216920844e-05,
+      "loss": 1.1989,
+      "step": 2023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.821520176783123e-05,
+      "loss": 1.2128,
+      "step": 2024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.821336841242915e-05,
+      "loss": 1.2292,
+      "step": 2025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8211534150786166e-05,
+      "loss": 1.2338,
+      "step": 2026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.820969898297389e-05,
+      "loss": 1.2301,
+      "step": 2027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.820786290906396e-05,
+      "loss": 0.9533,
+      "step": 2028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8206025929128066e-05,
+      "loss": 1.1482,
+      "step": 2029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.820418804323792e-05,
+      "loss": 1.1555,
+      "step": 2030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8202349251465276e-05,
+      "loss": 1.2167,
+      "step": 2031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8200509553881903e-05,
+      "loss": 1.2649,
+      "step": 2032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8198668950559634e-05,
+      "loss": 1.0488,
+      "step": 2033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.819682744157033e-05,
+      "loss": 1.2033,
+      "step": 2034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8194985026985874e-05,
+      "loss": 1.1603,
+      "step": 2035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.81931417068782e-05,
+      "loss": 1.1185,
+      "step": 2036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.819129748131926e-05,
+      "loss": 1.2213,
+      "step": 2037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.818945235038107e-05,
+      "loss": 1.0896,
+      "step": 2038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.818760631413565e-05,
+      "loss": 1.2069,
+      "step": 2039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.818575937265507e-05,
+      "loss": 1.2372,
+      "step": 2040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8183911526011426e-05,
+      "loss": 1.1335,
+      "step": 2041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.818206277427687e-05,
+      "loss": 1.0945,
+      "step": 2042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8180213117523584e-05,
+      "loss": 1.2419,
+      "step": 2043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8178362555823754e-05,
+      "loss": 1.2465,
+      "step": 2044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8176511089249645e-05,
+      "loss": 1.1796,
+      "step": 2045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8174658717873525e-05,
+      "loss": 1.1536,
+      "step": 2046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.817280544176772e-05,
+      "loss": 1.2624,
+      "step": 2047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.817095126100457e-05,
+      "loss": 1.2542,
+      "step": 2048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.816909617565647e-05,
+      "loss": 1.1453,
+      "step": 2049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8167240185795835e-05,
+      "loss": 1.1443,
+      "step": 2050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.816538329149514e-05,
+      "loss": 1.1663,
+      "step": 2051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.816352549282685e-05,
+      "loss": 1.1436,
+      "step": 2052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.816166678986351e-05,
+      "loss": 1.2141,
+      "step": 2053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.815980718267768e-05,
+      "loss": 1.1101,
+      "step": 2054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8157946671341964e-05,
+      "loss": 1.1852,
+      "step": 2055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.815608525592898e-05,
+      "loss": 1.1661,
+      "step": 2056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8154222936511415e-05,
+      "loss": 1.1958,
+      "step": 2057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.815235971316196e-05,
+      "loss": 1.1792,
+      "step": 2058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.815049558595337e-05,
+      "loss": 1.2024,
+      "step": 2059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.81486305549584e-05,
+      "loss": 1.1162,
+      "step": 2060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.814676462024988e-05,
+      "loss": 1.0898,
+      "step": 2061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8144897781900625e-05,
+      "loss": 1.2132,
+      "step": 2062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.814303003998355e-05,
+      "loss": 1.1819,
+      "step": 2063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.814116139457155e-05,
+      "loss": 1.1535,
+      "step": 2064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.813929184573759e-05,
+      "loss": 1.1917,
+      "step": 2065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8137421393554636e-05,
+      "loss": 1.2758,
+      "step": 2066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.813555003809573e-05,
+      "loss": 1.123,
+      "step": 2067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8133677779433924e-05,
+      "loss": 1.1802,
+      "step": 2068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8131804617642307e-05,
+      "loss": 1.1582,
+      "step": 2069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8129930552794007e-05,
+      "loss": 1.2465,
+      "step": 2070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.812805558496218e-05,
+      "loss": 1.1668,
+      "step": 2071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.812617971422004e-05,
+      "loss": 1.1564,
+      "step": 2072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8124302940640816e-05,
+      "loss": 1.1593,
+      "step": 2073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8122425264297766e-05,
+      "loss": 1.1622,
+      "step": 2074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8120546685264197e-05,
+      "loss": 1.1938,
+      "step": 2075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.811866720361345e-05,
+      "loss": 1.249,
+      "step": 2076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.81167868194189e-05,
+      "loss": 1.1469,
+      "step": 2077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8114905532753965e-05,
+      "loss": 1.1847,
+      "step": 2078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8113023343692076e-05,
+      "loss": 1.2466,
+      "step": 2079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.811114025230672e-05,
+      "loss": 1.2476,
+      "step": 2080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8109256258671404e-05,
+      "loss": 1.1982,
+      "step": 2081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8107371362859686e-05,
+      "loss": 1.2101,
+      "step": 2082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.810548556494516e-05,
+      "loss": 1.268,
+      "step": 2083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.810359886500143e-05,
+      "loss": 1.2136,
+      "step": 2084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.810171126310215e-05,
+      "loss": 1.1444,
+      "step": 2085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8099822759321036e-05,
+      "loss": 1.228,
+      "step": 2086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.809793335373179e-05,
+      "loss": 1.1348,
+      "step": 2087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.809604304640818e-05,
+      "loss": 1.1473,
+      "step": 2088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.809415183742402e-05,
+      "loss": 1.0238,
+      "step": 2089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.809225972685311e-05,
+      "loss": 1.052,
+      "step": 2090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.809036671476934e-05,
+      "loss": 1.1105,
+      "step": 2091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.808847280124661e-05,
+      "loss": 1.1324,
+      "step": 2092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.808657798635886e-05,
+      "loss": 1.1262,
+      "step": 2093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8084682270180045e-05,
+      "loss": 1.0554,
+      "step": 2094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8082785652784195e-05,
+      "loss": 1.0387,
+      "step": 2095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.808088813424534e-05,
+      "loss": 1.0654,
+      "step": 2096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8078989714637566e-05,
+      "loss": 1.1058,
+      "step": 2097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807709039403498e-05,
+      "loss": 1.0652,
+      "step": 2098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8075190172511734e-05,
+      "loss": 1.3094,
+      "step": 2099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807328905014201e-05,
+      "loss": 1.3708,
+      "step": 2100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807138702700003e-05,
+      "loss": 1.3549,
+      "step": 2101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806948410316005e-05,
+      "loss": 1.3901,
+      "step": 2102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806758027869634e-05,
+      "loss": 1.3119,
+      "step": 2103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806567555368325e-05,
+      "loss": 1.3097,
+      "step": 2104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8063769928195144e-05,
+      "loss": 1.3003,
+      "step": 2105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806186340230639e-05,
+      "loss": 1.3201,
+      "step": 2106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805995597609143e-05,
+      "loss": 1.3237,
+      "step": 2107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805804764962474e-05,
+      "loss": 1.3141,
+      "step": 2108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.80561384229808e-05,
+      "loss": 1.1688,
+      "step": 2109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805422829623417e-05,
+      "loss": 1.2397,
+      "step": 2110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8052317269459396e-05,
+      "loss": 1.1902,
+      "step": 2111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8050405342731096e-05,
+      "loss": 1.1532,
+      "step": 2112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.804849251612391e-05,
+      "loss": 1.2583,
+      "step": 2113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8046578789712515e-05,
+      "loss": 1.2498,
+      "step": 2114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8044664163571616e-05,
+      "loss": 1.1367,
+      "step": 2115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8042748637775976e-05,
+      "loss": 1.246,
+      "step": 2116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.804083221240035e-05,
+      "loss": 1.2415,
+      "step": 2117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.803891488751958e-05,
+      "loss": 1.4308,
+      "step": 2118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.80369966632085e-05,
+      "loss": 1.3897,
+      "step": 2119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8035077539542004e-05,
+      "loss": 1.4722,
+      "step": 2120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8033157516595015e-05,
+      "loss": 1.3466,
+      "step": 2121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8031236594442485e-05,
+      "loss": 1.4106,
+      "step": 2122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.802931477315942e-05,
+      "loss": 1.4496,
+      "step": 2123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8027392052820824e-05,
+      "loss": 1.3589,
+      "step": 2124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8025468433501774e-05,
+      "loss": 1.3874,
+      "step": 2125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.802354391527737e-05,
+      "loss": 1.3769,
+      "step": 2126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.802161849822273e-05,
+      "loss": 1.5055,
+      "step": 2127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.801969218241304e-05,
+      "loss": 1.202,
+      "step": 2128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.801776496792348e-05,
+      "loss": 1.1941,
+      "step": 2129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8015836854829324e-05,
+      "loss": 1.201,
+      "step": 2130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8013907843205805e-05,
+      "loss": 1.2662,
+      "step": 2131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8011977933128247e-05,
+      "loss": 1.1731,
+      "step": 2132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8010047124672e-05,
+      "loss": 1.2178,
+      "step": 2133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8008115417912436e-05,
+      "loss": 1.1819,
+      "step": 2134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800618281292497e-05,
+      "loss": 1.2199,
+      "step": 2135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800424930978504e-05,
+      "loss": 1.1769,
+      "step": 2136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8002314908568134e-05,
+      "loss": 0.9481,
+      "step": 2137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800037960934979e-05,
+      "loss": 0.945,
+      "step": 2138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799844341220553e-05,
+      "loss": 1.1557,
+      "step": 2139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799650631721096e-05,
+      "loss": 1.006,
+      "step": 2140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799456832444171e-05,
+      "loss": 0.9576,
+      "step": 2141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799262943397342e-05,
+      "loss": 1.0072,
+      "step": 2142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799068964588179e-05,
+      "loss": 1.1229,
+      "step": 2143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7988748960242555e-05,
+      "loss": 1.0189,
+      "step": 2144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7986807377131484e-05,
+      "loss": 1.0516,
+      "step": 2145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798486489662436e-05,
+      "loss": 1.0081,
+      "step": 2146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7982921518797016e-05,
+      "loss": 1.2498,
+      "step": 2147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798097724372533e-05,
+      "loss": 1.2792,
+      "step": 2148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797903207148521e-05,
+      "loss": 1.237,
+      "step": 2149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797708600215258e-05,
+      "loss": 1.2753,
+      "step": 2150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797513903580343e-05,
+      "loss": 1.2431,
+      "step": 2151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797319117251376e-05,
+      "loss": 1.2414,
+      "step": 2152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797124241235962e-05,
+      "loss": 1.1842,
+      "step": 2153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796929275541707e-05,
+      "loss": 1.3169,
+      "step": 2154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796734220176225e-05,
+      "loss": 1.1737,
+      "step": 2155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796539075147129e-05,
+      "loss": 1.1896,
+      "step": 2156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796343840462038e-05,
+      "loss": 1.072,
+      "step": 2157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796148516128575e-05,
+      "loss": 0.9959,
+      "step": 2158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7959531021543624e-05,
+      "loss": 1.098,
+      "step": 2159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.795757598547032e-05,
+      "loss": 1.1079,
+      "step": 2160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7955620053142156e-05,
+      "loss": 1.0327,
+      "step": 2161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.795366322463548e-05,
+      "loss": 1.0914,
+      "step": 2162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.795170550002669e-05,
+      "loss": 1.0261,
+      "step": 2163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.794974687939222e-05,
+      "loss": 1.1156,
+      "step": 2164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.794778736280853e-05,
+      "loss": 1.1064,
+      "step": 2165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7945826950352124e-05,
+      "loss": 1.0062,
+      "step": 2166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.794386564209953e-05,
+      "loss": 1.1208,
+      "step": 2167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.794190343812732e-05,
+      "loss": 1.1576,
+      "step": 2168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.793994033851209e-05,
+      "loss": 1.1173,
+      "step": 2169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7937976343330496e-05,
+      "loss": 1.2457,
+      "step": 2170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7936011452659184e-05,
+      "loss": 1.1917,
+      "step": 2171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.793404566657489e-05,
+      "loss": 1.2657,
+      "step": 2172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7932078985154335e-05,
+      "loss": 1.2247,
+      "step": 2173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.793011140847432e-05,
+      "loss": 1.1021,
+      "step": 2174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.792814293661164e-05,
+      "loss": 1.1683,
+      "step": 2175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7926173569643164e-05,
+      "loss": 1.2488,
+      "step": 2176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.792420330764574e-05,
+      "loss": 1.1546,
+      "step": 2177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.792223215069633e-05,
+      "loss": 1.047,
+      "step": 2178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.792026009887185e-05,
+      "loss": 1.126,
+      "step": 2179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.791828715224931e-05,
+      "loss": 1.0964,
+      "step": 2180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.791631331090572e-05,
+      "loss": 1.1308,
+      "step": 2181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.791433857491816e-05,
+      "loss": 1.193,
+      "step": 2182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7912362944363696e-05,
+      "loss": 1.0822,
+      "step": 2183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.791038641931946e-05,
+      "loss": 1.1022,
+      "step": 2184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.790840899986263e-05,
+      "loss": 1.0968,
+      "step": 2185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7906430686070404e-05,
+      "loss": 1.1824,
+      "step": 2186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7904451478019996e-05,
+      "loss": 1.2573,
+      "step": 2187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7902471375788686e-05,
+      "loss": 1.4365,
+      "step": 2188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.790049037945378e-05,
+      "loss": 1.4498,
+      "step": 2189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7898508489092596e-05,
+      "loss": 1.5415,
+      "step": 2190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.789652570478254e-05,
+      "loss": 1.4477,
+      "step": 2191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.789454202660098e-05,
+      "loss": 1.4078,
+      "step": 2192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7892557454625383e-05,
+      "loss": 1.5172,
+      "step": 2193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.789057198893322e-05,
+      "loss": 1.4828,
+      "step": 2194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7888585629602e-05,
+      "loss": 1.3764,
+      "step": 2195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.788659837670928e-05,
+      "loss": 1.4832,
+      "step": 2196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7884610230332625e-05,
+      "loss": 1.4911,
+      "step": 2197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.788262119054967e-05,
+      "loss": 1.4722,
+      "step": 2198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7880631257438055e-05,
+      "loss": 1.1435,
+      "step": 2199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.787864043107546e-05,
+      "loss": 1.139,
+      "step": 2200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7876648711539625e-05,
+      "loss": 1.1343,
+      "step": 2201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7874656098908285e-05,
+      "loss": 1.1717,
+      "step": 2202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.787266259325925e-05,
+      "loss": 1.0691,
+      "step": 2203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.787066819467033e-05,
+      "loss": 1.1627,
+      "step": 2204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.786867290321939e-05,
+      "loss": 1.1261,
+      "step": 2205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7866676718984335e-05,
+      "loss": 1.1286,
+      "step": 2206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7864679642043085e-05,
+      "loss": 1.0698,
+      "step": 2207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7862681672473605e-05,
+      "loss": 1.1113,
+      "step": 2208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7860682810353905e-05,
+      "loss": 1.212,
+      "step": 2209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.785868305576201e-05,
+      "loss": 1.1304,
+      "step": 2210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.785668240877599e-05,
+      "loss": 1.1892,
+      "step": 2211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.785468086947395e-05,
+      "loss": 1.2266,
+      "step": 2212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7852678437934035e-05,
+      "loss": 1.1929,
+      "step": 2213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.785067511423441e-05,
+      "loss": 1.1848,
+      "step": 2214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7848670898453295e-05,
+      "loss": 1.1724,
+      "step": 2215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.784666579066892e-05,
+      "loss": 1.1792,
+      "step": 2216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7844659790959584e-05,
+      "loss": 1.1558,
+      "step": 2217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.784265289940358e-05,
+      "loss": 1.1404,
+      "step": 2218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7840645116079266e-05,
+      "loss": 1.2583,
+      "step": 2219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.783863644106502e-05,
+      "loss": 1.1903,
+      "step": 2220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7836626874439274e-05,
+      "loss": 1.1797,
+      "step": 2221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.783461641628046e-05,
+      "loss": 1.1352,
+      "step": 2222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.783260506666708e-05,
+      "loss": 1.2859,
+      "step": 2223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.783059282567765e-05,
+      "loss": 1.1795,
+      "step": 2224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782857969339073e-05,
+      "loss": 1.2508,
+      "step": 2225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782656566988491e-05,
+      "loss": 1.149,
+      "step": 2226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782455075523882e-05,
+      "loss": 1.2401,
+      "step": 2227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7822534949531115e-05,
+      "loss": 1.2635,
+      "step": 2228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78205182528405e-05,
+      "loss": 1.1746,
+      "step": 2229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7818500665245694e-05,
+      "loss": 1.313,
+      "step": 2230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.781648218682548e-05,
+      "loss": 1.3095,
+      "step": 2231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.781446281765865e-05,
+      "loss": 1.3463,
+      "step": 2232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.781244255782402e-05,
+      "loss": 1.201,
+      "step": 2233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7810421407400496e-05,
+      "loss": 1.217,
+      "step": 2234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.780839936646696e-05,
+      "loss": 1.2787,
+      "step": 2235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.780637643510236e-05,
+      "loss": 1.2236,
+      "step": 2236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.780435261338566e-05,
+      "loss": 1.27,
+      "step": 2237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.780232790139588e-05,
+      "loss": 1.2337,
+      "step": 2238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.780030229921206e-05,
+      "loss": 1.2648,
+      "step": 2239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7798275806913284e-05,
+      "loss": 1.3205,
+      "step": 2240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.779624842457866e-05,
+      "loss": 1.2414,
+      "step": 2241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7794220152287334e-05,
+      "loss": 1.3265,
+      "step": 2242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7792190990118495e-05,
+      "loss": 1.2255,
+      "step": 2243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.779016093815135e-05,
+      "loss": 1.2429,
+      "step": 2244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.778812999646517e-05,
+      "loss": 1.2128,
+      "step": 2245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7786098165139225e-05,
+      "loss": 1.1494,
+      "step": 2246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.778406544425284e-05,
+      "loss": 1.2489,
+      "step": 2247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7782031833885386e-05,
+      "loss": 1.2677,
+      "step": 2248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777999733411624e-05,
+      "loss": 1.2111,
+      "step": 2249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777796194502483e-05,
+      "loss": 1.1379,
+      "step": 2250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777592566669061e-05,
+      "loss": 1.2274,
+      "step": 2251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7773888499193095e-05,
+      "loss": 1.1322,
+      "step": 2252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77718504426118e-05,
+      "loss": 1.2525,
+      "step": 2253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77698114970263e-05,
+      "loss": 1.1542,
+      "step": 2254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776777166251618e-05,
+      "loss": 1.2669,
+      "step": 2255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776573093916108e-05,
+      "loss": 1.2124,
+      "step": 2256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7763689327040676e-05,
+      "loss": 1.2095,
+      "step": 2257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7761646826234676e-05,
+      "loss": 1.0543,
+      "step": 2258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7759603436822796e-05,
+      "loss": 1.177,
+      "step": 2259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775755915888483e-05,
+      "loss": 1.1577,
+      "step": 2260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775551399250058e-05,
+      "loss": 1.1649,
+      "step": 2261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7753467937749884e-05,
+      "loss": 1.2341,
+      "step": 2262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7751420994712626e-05,
+      "loss": 1.1997,
+      "step": 2263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774937316346871e-05,
+      "loss": 1.1967,
+      "step": 2264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7747324444098095e-05,
+      "loss": 1.1574,
+      "step": 2265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7745274836680755e-05,
+      "loss": 1.1376,
+      "step": 2266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774322434129669e-05,
+      "loss": 1.1712,
+      "step": 2267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774117295802598e-05,
+      "loss": 1.1487,
+      "step": 2268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7739120686948694e-05,
+      "loss": 1.2244,
+      "step": 2269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7737067528144955e-05,
+      "loss": 1.2512,
+      "step": 2270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773501348169491e-05,
+      "loss": 1.2532,
+      "step": 2271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7732958547678754e-05,
+      "loss": 1.262,
+      "step": 2272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773090272617672e-05,
+      "loss": 1.2472,
+      "step": 2273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7728846017269045e-05,
+      "loss": 1.2407,
+      "step": 2274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772678842103605e-05,
+      "loss": 1.2393,
+      "step": 2275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7724729937558035e-05,
+      "loss": 1.2747,
+      "step": 2276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7722670566915385e-05,
+      "loss": 1.2017,
+      "step": 2277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7720610309188485e-05,
+      "loss": 1.2818,
+      "step": 2278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771854916445776e-05,
+      "loss": 1.2202,
+      "step": 2279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7716487132803694e-05,
+      "loss": 1.1544,
+      "step": 2280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771442421430677e-05,
+      "loss": 1.1649,
+      "step": 2281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771236040904754e-05,
+      "loss": 1.172,
+      "step": 2282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7710295717106566e-05,
+      "loss": 1.1838,
+      "step": 2283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770823013856446e-05,
+      "loss": 1.1335,
+      "step": 2284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770616367350185e-05,
+      "loss": 1.1275,
+      "step": 2285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770409632199941e-05,
+      "loss": 1.1521,
+      "step": 2286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770202808413786e-05,
+      "loss": 1.1991,
+      "step": 2287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769995895999793e-05,
+      "loss": 1.1357,
+      "step": 2288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7697888949660404e-05,
+      "loss": 1.2167,
+      "step": 2289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76958180532061e-05,
+      "loss": 1.1492,
+      "step": 2290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7693746270715856e-05,
+      "loss": 1.28,
+      "step": 2291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769167360227056e-05,
+      "loss": 1.1945,
+      "step": 2292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768960004795113e-05,
+      "loss": 1.1599,
+      "step": 2293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768752560783851e-05,
+      "loss": 1.1918,
+      "step": 2294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768545028201368e-05,
+      "loss": 1.2829,
+      "step": 2295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7683374070557676e-05,
+      "loss": 1.2338,
+      "step": 2296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768129697355154e-05,
+      "loss": 1.238,
+      "step": 2297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767921899107637e-05,
+      "loss": 1.2751,
+      "step": 2298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767714012321328e-05,
+      "loss": 1.2408,
+      "step": 2299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767506037004344e-05,
+      "loss": 1.2071,
+      "step": 2300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7672979731648024e-05,
+      "loss": 1.1978,
+      "step": 2301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767089820810827e-05,
+      "loss": 1.213,
+      "step": 2302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766881579950545e-05,
+      "loss": 1.2246,
+      "step": 2303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766673250592085e-05,
+      "loss": 1.3133,
+      "step": 2304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76646483274358e-05,
+      "loss": 1.2152,
+      "step": 2305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766256326413168e-05,
+      "loss": 1.2572,
+      "step": 2306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7660477316089866e-05,
+      "loss": 1.2947,
+      "step": 2307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76583904833918e-05,
+      "loss": 1.2657,
+      "step": 2308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7656302766118965e-05,
+      "loss": 1.2455,
+      "step": 2309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765421416435285e-05,
+      "loss": 1.2287,
+      "step": 2310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7652124678175e-05,
+      "loss": 1.1796,
+      "step": 2311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765003430766699e-05,
+      "loss": 1.1556,
+      "step": 2312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7647943052910414e-05,
+      "loss": 1.1752,
+      "step": 2313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764585091398693e-05,
+      "loss": 1.1551,
+      "step": 2314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764375789097821e-05,
+      "loss": 1.1212,
+      "step": 2315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764166398396596e-05,
+      "loss": 1.0946,
+      "step": 2316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763956919303193e-05,
+      "loss": 1.1891,
+      "step": 2317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7637473518257893e-05,
+      "loss": 1.1346,
+      "step": 2318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7635376959725676e-05,
+      "loss": 1.1009,
+      "step": 2319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763327951751711e-05,
+      "loss": 1.2192,
+      "step": 2320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.763118119171409e-05,
+      "loss": 1.3799,
+      "step": 2321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.762908198239854e-05,
+      "loss": 1.3215,
+      "step": 2322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.762698188965239e-05,
+      "loss": 1.2735,
+      "step": 2323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.762488091355765e-05,
+      "loss": 1.298,
+      "step": 2324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.762277905419633e-05,
+      "loss": 1.3259,
+      "step": 2325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.762067631165049e-05,
+      "loss": 1.3874,
+      "step": 2326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7618572686002214e-05,
+      "loss": 1.2803,
+      "step": 2327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761646817733364e-05,
+      "loss": 1.2675,
+      "step": 2328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761436278572691e-05,
+      "loss": 1.3156,
+      "step": 2329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761225651126423e-05,
+      "loss": 0.9744,
+      "step": 2330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761014935402782e-05,
+      "loss": 1.0304,
+      "step": 2331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760804131409995e-05,
+      "loss": 1.0959,
+      "step": 2332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760593239156291e-05,
+      "loss": 1.1676,
+      "step": 2333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760382258649904e-05,
+      "loss": 1.0958,
+      "step": 2334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7601711898990706e-05,
+      "loss": 1.1383,
+      "step": 2335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75996003291203e-05,
+      "loss": 1.0986,
+      "step": 2336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759748787697025e-05,
+      "loss": 1.1195,
+      "step": 2337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7595374542623045e-05,
+      "loss": 1.0711,
+      "step": 2338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759326032616118e-05,
+      "loss": 1.0036,
+      "step": 2339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.759114522766719e-05,
+      "loss": 0.8924,
+      "step": 2340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758902924722365e-05,
+      "loss": 0.9393,
+      "step": 2341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758691238491317e-05,
+      "loss": 0.8415,
+      "step": 2342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758479464081839e-05,
+      "loss": 0.8754,
+      "step": 2343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7582676015021975e-05,
+      "loss": 0.9111,
+      "step": 2344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7580556507606655e-05,
+      "loss": 0.983,
+      "step": 2345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7578436118655164e-05,
+      "loss": 0.8554,
+      "step": 2346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757631484825028e-05,
+      "loss": 1.0077,
+      "step": 2347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757419269647482e-05,
+      "loss": 0.9476,
+      "step": 2348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757206966341163e-05,
+      "loss": 0.8681,
+      "step": 2349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756994574914359e-05,
+      "loss": 0.8913,
+      "step": 2350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756782095375362e-05,
+      "loss": 1.1503,
+      "step": 2351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7565695277324676e-05,
+      "loss": 1.3288,
+      "step": 2352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756356871993973e-05,
+      "loss": 1.3858,
+      "step": 2353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7561441281681824e-05,
+      "loss": 1.411,
+      "step": 2354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755931296263399e-05,
+      "loss": 1.3175,
+      "step": 2355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755718376287933e-05,
+      "loss": 1.3311,
+      "step": 2356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755505368250096e-05,
+      "loss": 1.3147,
+      "step": 2357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755292272158204e-05,
+      "loss": 1.3721,
+      "step": 2358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7550790880205764e-05,
+      "loss": 1.3825,
+      "step": 2359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754865815845535e-05,
+      "loss": 1.3437,
+      "step": 2360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7546524556414075e-05,
+      "loss": 1.3223,
+      "step": 2361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754439007416521e-05,
+      "loss": 1.1401,
+      "step": 2362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754225471179211e-05,
+      "loss": 1.0724,
+      "step": 2363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754011846937812e-05,
+      "loss": 1.0709,
+      "step": 2364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7537981347006657e-05,
+      "loss": 1.2383,
+      "step": 2365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753584334476113e-05,
+      "loss": 1.2371,
+      "step": 2366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7533704462725025e-05,
+      "loss": 1.2133,
+      "step": 2367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753156470098183e-05,
+      "loss": 1.1199,
+      "step": 2368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752942405961509e-05,
+      "loss": 1.1854,
+      "step": 2369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752728253870837e-05,
+      "loss": 1.1297,
+      "step": 2370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752514013834528e-05,
+      "loss": 1.2554,
+      "step": 2371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752299685860945e-05,
+      "loss": 1.3257,
+      "step": 2372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.752085269958455e-05,
+      "loss": 1.4062,
+      "step": 2373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75187076613543e-05,
+      "loss": 1.3895,
+      "step": 2374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751656174400243e-05,
+      "loss": 1.3695,
+      "step": 2375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751441494761272e-05,
+      "loss": 1.3019,
+      "step": 2376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751226727226898e-05,
+      "loss": 1.378,
+      "step": 2377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751011871805507e-05,
+      "loss": 1.4249,
+      "step": 2378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7507969285054845e-05,
+      "loss": 1.4036,
+      "step": 2379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750581897335222e-05,
+      "loss": 1.3527,
+      "step": 2380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.750366778303116e-05,
+      "loss": 1.3376,
+      "step": 2381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7501515714175626e-05,
+      "loss": 1.1908,
+      "step": 2382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749936276686965e-05,
+      "loss": 1.1685,
+      "step": 2383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7497208941197266e-05,
+      "loss": 1.2425,
+      "step": 2384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749505423724258e-05,
+      "loss": 1.1875,
+      "step": 2385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.749289865508969e-05,
+      "loss": 1.2322,
+      "step": 2386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7490742194822766e-05,
+      "loss": 1.2956,
+      "step": 2387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748858485652599e-05,
+      "loss": 1.2718,
+      "step": 2388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748642664028358e-05,
+      "loss": 1.2266,
+      "step": 2389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748426754617979e-05,
+      "loss": 1.2371,
+      "step": 2390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748210757429892e-05,
+      "loss": 1.1613,
+      "step": 2391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747994672472528e-05,
+      "loss": 0.896,
+      "step": 2392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747778499754324e-05,
+      "loss": 0.9955,
+      "step": 2393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.747562239283719e-05,
+      "loss": 0.85,
+      "step": 2394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7473458910691556e-05,
+      "loss": 0.7822,
+      "step": 2395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7471294551190804e-05,
+      "loss": 0.7506,
+      "step": 2396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746912931441942e-05,
+      "loss": 0.882,
+      "step": 2397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7466963200461945e-05,
+      "loss": 0.7375,
+      "step": 2398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746479620940294e-05,
+      "loss": 0.8617,
+      "step": 2399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7462628341326995e-05,
+      "loss": 0.8862,
+      "step": 2400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746045959631875e-05,
+      "loss": 0.9542,
+      "step": 2401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7458289974462884e-05,
+      "loss": 0.8051,
+      "step": 2402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7456119475844075e-05,
+      "loss": 1.2034,
+      "step": 2403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7453948100547074e-05,
+      "loss": 1.1074,
+      "step": 2404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7451775848656645e-05,
+      "loss": 1.1218,
+      "step": 2405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744960272025759e-05,
+      "loss": 1.1311,
+      "step": 2406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744742871543475e-05,
+      "loss": 1.1627,
+      "step": 2407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744525383427301e-05,
+      "loss": 1.2631,
+      "step": 2408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744307807685724e-05,
+      "loss": 1.1742,
+      "step": 2409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7440901443272424e-05,
+      "loss": 1.2273,
+      "step": 2410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7438723933603516e-05,
+      "loss": 1.1988,
+      "step": 2411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.743654554793552e-05,
+      "loss": 1.1585,
+      "step": 2412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.743436628635348e-05,
+      "loss": 1.2586,
+      "step": 2413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7432186148942495e-05,
+      "loss": 1.2616,
+      "step": 2414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.743000513578765e-05,
+      "loss": 1.1727,
+      "step": 2415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.742782324697411e-05,
+      "loss": 1.1696,
+      "step": 2416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.742564048258704e-05,
+      "loss": 1.1152,
+      "step": 2417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.742345684271167e-05,
+      "loss": 1.2463,
+      "step": 2418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.742127232743323e-05,
+      "loss": 1.2146,
+      "step": 2419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7419086936837017e-05,
+      "loss": 1.1914,
+      "step": 2420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7416900671008344e-05,
+      "loss": 1.2576,
+      "step": 2421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7414713530032555e-05,
+      "loss": 1.2046,
+      "step": 2422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.741252551399505e-05,
+      "loss": 1.1247,
+      "step": 2423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7410336622981234e-05,
+      "loss": 1.2009,
+      "step": 2424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7408146857076566e-05,
+      "loss": 1.0698,
+      "step": 2425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.740595621636654e-05,
+      "loss": 1.098,
+      "step": 2426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.740376470093667e-05,
+      "loss": 1.1608,
+      "step": 2427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.740157231087251e-05,
+      "loss": 1.1616,
+      "step": 2428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739937904625965e-05,
+      "loss": 1.1599,
+      "step": 2429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739718490718373e-05,
+      "loss": 1.1675,
+      "step": 2430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739498989373039e-05,
+      "loss": 1.1402,
+      "step": 2431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7392794005985326e-05,
+      "loss": 1.3558,
+      "step": 2432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739059724403426e-05,
+      "loss": 1.2282,
+      "step": 2433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.738839960796297e-05,
+      "loss": 1.3239,
+      "step": 2434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.738620109785724e-05,
+      "loss": 1.3174,
+      "step": 2435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7384001713802906e-05,
+      "loss": 1.1662,
+      "step": 2436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7381801455885814e-05,
+      "loss": 1.3523,
+      "step": 2437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7379600324191875e-05,
+      "loss": 1.1798,
+      "step": 2438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737739831880702e-05,
+      "loss": 1.2708,
+      "step": 2439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737519543981721e-05,
+      "loss": 1.2556,
+      "step": 2440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737299168730844e-05,
+      "loss": 1.1297,
+      "step": 2441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737078706136676e-05,
+      "loss": 1.016,
+      "step": 2442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.736858156207823e-05,
+      "loss": 1.0441,
+      "step": 2443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.736637518952895e-05,
+      "loss": 0.9672,
+      "step": 2444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.736416794380504e-05,
+      "loss": 1.0252,
+      "step": 2445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7361959824992695e-05,
+      "loss": 1.0084,
+      "step": 2446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.735975083317811e-05,
+      "loss": 0.9976,
+      "step": 2447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.735754096844752e-05,
+      "loss": 1.0679,
+      "step": 2448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.735533023088721e-05,
+      "loss": 0.9665,
+      "step": 2449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.735311862058346e-05,
+      "loss": 0.966,
+      "step": 2450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7350906137622644e-05,
+      "loss": 1.0005,
+      "step": 2451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.734869278209111e-05,
+      "loss": 1.1517,
+      "step": 2452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.734647855407528e-05,
+      "loss": 1.2937,
+      "step": 2453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.734426345366159e-05,
+      "loss": 1.321,
+      "step": 2454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7342047480936516e-05,
+      "loss": 1.2401,
+      "step": 2455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.733983063598657e-05,
+      "loss": 1.3015,
+      "step": 2456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.73376129188983e-05,
+      "loss": 1.261,
+      "step": 2457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7335394329758286e-05,
+      "loss": 1.3088,
+      "step": 2458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.733317486865314e-05,
+      "loss": 1.3007,
+      "step": 2459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.73309545356695e-05,
+      "loss": 1.3316,
+      "step": 2460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.732873333089407e-05,
+      "loss": 1.3954,
+      "step": 2461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.732651125441353e-05,
+      "loss": 1.3438,
+      "step": 2462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.732428830631466e-05,
+      "loss": 1.4284,
+      "step": 2463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.732206448668423e-05,
+      "loss": 1.4243,
+      "step": 2464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7319839795609056e-05,
+      "loss": 1.3647,
+      "step": 2465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.731761423317599e-05,
+      "loss": 1.3365,
+      "step": 2466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.731538779947192e-05,
+      "loss": 1.3372,
+      "step": 2467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.731316049458376e-05,
+      "loss": 1.3768,
+      "step": 2468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.731093231859847e-05,
+      "loss": 1.4016,
+      "step": 2469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.730870327160304e-05,
+      "loss": 1.3838,
+      "step": 2470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.730647335368448e-05,
+      "loss": 1.2943,
+      "step": 2471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7304242564929846e-05,
+      "loss": 1.2336,
+      "step": 2472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.730201090542623e-05,
+      "loss": 1.2139,
+      "step": 2473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.729977837526076e-05,
+      "loss": 1.2339,
+      "step": 2474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7297544974520594e-05,
+      "loss": 1.2145,
+      "step": 2475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.729531070329291e-05,
+      "loss": 1.2102,
+      "step": 2476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7293075561664946e-05,
+      "loss": 1.2097,
+      "step": 2477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7290839549723954e-05,
+      "loss": 1.2129,
+      "step": 2478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.728860266755722e-05,
+      "loss": 1.2599,
+      "step": 2479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.72863649152521e-05,
+      "loss": 1.2266,
+      "step": 2480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7284126292895924e-05,
+      "loss": 1.2417,
+      "step": 2481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7281886800576104e-05,
+      "loss": 1.1768,
+      "step": 2482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7279646438380065e-05,
+      "loss": 1.2031,
+      "step": 2483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.727740520639527e-05,
+      "loss": 1.235,
+      "step": 2484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.72751631047092e-05,
+      "loss": 1.202,
+      "step": 2485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7272920133409404e-05,
+      "loss": 1.232,
+      "step": 2486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.727067629258345e-05,
+      "loss": 1.2554,
+      "step": 2487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.726843158231893e-05,
+      "loss": 1.3799,
+      "step": 2488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.726618600270347e-05,
+      "loss": 1.1767,
+      "step": 2489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.726393955382474e-05,
+      "loss": 1.3079,
+      "step": 2490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.726169223577045e-05,
+      "loss": 1.479,
+      "step": 2491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.725944404862833e-05,
+      "loss": 1.7158,
+      "step": 2492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7257194992486145e-05,
+      "loss": 1.5901,
+      "step": 2493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.725494506743169e-05,
+      "loss": 1.6168,
+      "step": 2494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.725269427355281e-05,
+      "loss": 1.6187,
+      "step": 2495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.725044261093738e-05,
+      "loss": 1.5868,
+      "step": 2496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.724819007967329e-05,
+      "loss": 1.5713,
+      "step": 2497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7245936679848504e-05,
+      "loss": 1.6194,
+      "step": 2498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.724368241155096e-05,
+      "loss": 1.4581,
+      "step": 2499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.724142727486869e-05,
+      "loss": 1.1364,
+      "step": 2500
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 1.171353816986084,
+      "eval_runtime": 113.3861,
+      "eval_samples_per_second": 33.699,
+      "eval_steps_per_second": 0.353,
+      "step": 2500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7239171269889716e-05,
+      "loss": 1.2393,
+      "step": 2501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7236914396702114e-05,
+      "loss": 1.209,
+      "step": 2502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.723465665539401e-05,
+      "loss": 1.1695,
+      "step": 2503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.723239804605353e-05,
+      "loss": 1.2736,
+      "step": 2504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7230138568768846e-05,
+      "loss": 1.1293,
+      "step": 2505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.722787822362817e-05,
+      "loss": 1.1157,
+      "step": 2506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7225617010719755e-05,
+      "loss": 1.136,
+      "step": 2507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.722335493013187e-05,
+      "loss": 1.1808,
+      "step": 2508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.722109198195282e-05,
+      "loss": 1.2767,
+      "step": 2509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7218828166270954e-05,
+      "loss": 1.1889,
+      "step": 2510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.721656348317466e-05,
+      "loss": 1.2969,
+      "step": 2511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.721429793275234e-05,
+      "loss": 1.1958,
+      "step": 2512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.721203151509244e-05,
+      "loss": 1.2599,
+      "step": 2513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7209764230283435e-05,
+      "loss": 1.1912,
+      "step": 2514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.720749607841386e-05,
+      "loss": 1.1804,
+      "step": 2515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.720522705957224e-05,
+      "loss": 1.2242,
+      "step": 2516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.720295717384718e-05,
+      "loss": 1.2171,
+      "step": 2517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.720068642132727e-05,
+      "loss": 1.123,
+      "step": 2518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7198414802101173e-05,
+      "loss": 1.2483,
+      "step": 2519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7196142316257574e-05,
+      "loss": 1.1569,
+      "step": 2520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.719386896388518e-05,
+      "loss": 1.1921,
+      "step": 2521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.719159474507275e-05,
+      "loss": 1.2824,
+      "step": 2522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.718931965990907e-05,
+      "loss": 1.226,
+      "step": 2523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7187043708482946e-05,
+      "loss": 1.2614,
+      "step": 2524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.718476689088325e-05,
+      "loss": 1.29,
+      "step": 2525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.718248920719884e-05,
+      "loss": 1.2755,
+      "step": 2526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.718021065751867e-05,
+      "loss": 1.2488,
+      "step": 2527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.717793124193167e-05,
+      "loss": 1.2132,
+      "step": 2528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.717565096052684e-05,
+      "loss": 1.2361,
+      "step": 2529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.717336981339319e-05,
+      "loss": 1.2487,
+      "step": 2530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7171087800619774e-05,
+      "loss": 1.2291,
+      "step": 2531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.716880492229569e-05,
+      "loss": 1.1416,
+      "step": 2532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.716652117851007e-05,
+      "loss": 1.1996,
+      "step": 2533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.716423656935205e-05,
+      "loss": 1.2574,
+      "step": 2534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.716195109491083e-05,
+      "loss": 1.268,
+      "step": 2535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.715966475527563e-05,
+      "loss": 1.2342,
+      "step": 2536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.715737755053571e-05,
+      "loss": 1.1589,
+      "step": 2537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.715508948078037e-05,
+      "loss": 1.2181,
+      "step": 2538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7152800546098926e-05,
+      "loss": 1.2045,
+      "step": 2539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.715051074658073e-05,
+      "loss": 1.2435,
+      "step": 2540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.714822008231519e-05,
+      "loss": 1.2271,
+      "step": 2541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7145928553391726e-05,
+      "loss": 1.2448,
+      "step": 2542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.71436361598998e-05,
+      "loss": 1.2389,
+      "step": 2543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7141342901928906e-05,
+      "loss": 1.3207,
+      "step": 2544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.713904877956857e-05,
+      "loss": 1.3054,
+      "step": 2545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.713675379290835e-05,
+      "loss": 1.271,
+      "step": 2546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7134457942037846e-05,
+      "loss": 1.2845,
+      "step": 2547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.713216122704669e-05,
+      "loss": 1.28,
+      "step": 2548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7129863648024545e-05,
+      "loss": 1.1996,
+      "step": 2549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7127565205061096e-05,
+      "loss": 1.2808,
+      "step": 2550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.712526589824609e-05,
+      "loss": 1.2085,
+      "step": 2551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7122965727669275e-05,
+      "loss": 1.2144,
+      "step": 2552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.712066469342047e-05,
+      "loss": 1.1982,
+      "step": 2553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7118362795589474e-05,
+      "loss": 1.1773,
+      "step": 2554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.711606003426619e-05,
+      "loss": 1.2496,
+      "step": 2555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.711375640954048e-05,
+      "loss": 1.2211,
+      "step": 2556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7111451921502306e-05,
+      "loss": 1.2554,
+      "step": 2557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.710914657024163e-05,
+      "loss": 1.3055,
+      "step": 2558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7106840355848425e-05,
+      "loss": 1.2751,
+      "step": 2559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7104533278412763e-05,
+      "loss": 1.1724,
+      "step": 2560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.710222533802468e-05,
+      "loss": 1.2522,
+      "step": 2561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7099916534774295e-05,
+      "loss": 1.1438,
+      "step": 2562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7097606868751744e-05,
+      "loss": 1.2887,
+      "step": 2563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.709529634004718e-05,
+      "loss": 1.1911,
+      "step": 2564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.709298494875082e-05,
+      "loss": 1.2784,
+      "step": 2565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70906726949529e-05,
+      "loss": 1.2147,
+      "step": 2566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.708835957874367e-05,
+      "loss": 1.2648,
+      "step": 2567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7086045600213455e-05,
+      "loss": 1.194,
+      "step": 2568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7083730759452586e-05,
+      "loss": 1.1934,
+      "step": 2569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.708141505655143e-05,
+      "loss": 1.2146,
+      "step": 2570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7079098491600395e-05,
+      "loss": 1.1795,
+      "step": 2571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.707678106468991e-05,
+      "loss": 1.1278,
+      "step": 2572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.707446277591046e-05,
+      "loss": 1.1667,
+      "step": 2573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7072143625352536e-05,
+      "loss": 1.1563,
+      "step": 2574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.706982361310669e-05,
+      "loss": 1.1951,
+      "step": 2575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7067502739263483e-05,
+      "loss": 1.2066,
+      "step": 2576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.706518100391353e-05,
+      "loss": 1.245,
+      "step": 2577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.706285840714746e-05,
+      "loss": 1.291,
+      "step": 2578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.706053494905596e-05,
+      "loss": 1.2029,
+      "step": 2579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7058210629729725e-05,
+      "loss": 1.1979,
+      "step": 2580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.705588544925951e-05,
+      "loss": 1.2868,
+      "step": 2581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.705355940773607e-05,
+      "loss": 1.206,
+      "step": 2582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.705123250525022e-05,
+      "loss": 1.1774,
+      "step": 2583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.704890474189281e-05,
+      "loss": 1.3063,
+      "step": 2584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.704657611775471e-05,
+      "loss": 1.2541,
+      "step": 2585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.704424663292682e-05,
+      "loss": 1.1358,
+      "step": 2586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70419162875001e-05,
+      "loss": 1.3105,
+      "step": 2587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.703958508156551e-05,
+      "loss": 1.2467,
+      "step": 2588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.703725301521406e-05,
+      "loss": 1.2104,
+      "step": 2589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7034920088536804e-05,
+      "loss": 1.2589,
+      "step": 2590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70325863016248e-05,
+      "loss": 1.2796,
+      "step": 2591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.703025165456919e-05,
+      "loss": 1.2946,
+      "step": 2592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7027916147461085e-05,
+      "loss": 1.3281,
+      "step": 2593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.702557978039168e-05,
+      "loss": 1.1779,
+      "step": 2594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.702324255345218e-05,
+      "loss": 1.2067,
+      "step": 2595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.702090446673383e-05,
+      "loss": 1.2707,
+      "step": 2596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.701856552032791e-05,
+      "loss": 1.1903,
+      "step": 2597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.701622571432572e-05,
+      "loss": 1.2069,
+      "step": 2598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.701388504881863e-05,
+      "loss": 1.1998,
+      "step": 2599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7011543523897996e-05,
+      "loss": 1.2357,
+      "step": 2600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.700920113965524e-05,
+      "loss": 1.2125,
+      "step": 2601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7006857896181805e-05,
+      "loss": 1.1706,
+      "step": 2602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.700451379356917e-05,
+      "loss": 1.156,
+      "step": 2603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.700216883190886e-05,
+      "loss": 1.1754,
+      "step": 2604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.69998230112924e-05,
+      "loss": 1.1941,
+      "step": 2605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6997476331811374e-05,
+      "loss": 1.2572,
+      "step": 2606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6995128793557416e-05,
+      "loss": 1.191,
+      "step": 2607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.699278039662215e-05,
+      "loss": 1.2206,
+      "step": 2608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6990431141097266e-05,
+      "loss": 1.2127,
+      "step": 2609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.698808102707448e-05,
+      "loss": 1.1906,
+      "step": 2610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.698573005464554e-05,
+      "loss": 1.2476,
+      "step": 2611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.698337822390222e-05,
+      "loss": 1.2149,
+      "step": 2612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.698102553493634e-05,
+      "loss": 1.2695,
+      "step": 2613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.697867198783974e-05,
+      "loss": 1.3244,
+      "step": 2614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.697631758270432e-05,
+      "loss": 1.2836,
+      "step": 2615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.697396231962197e-05,
+      "loss": 1.2724,
+      "step": 2616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.697160619868466e-05,
+      "loss": 1.2359,
+      "step": 2617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.696924921998437e-05,
+      "loss": 1.3076,
+      "step": 2618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.69668913836131e-05,
+      "loss": 1.3021,
+      "step": 2619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.696453268966291e-05,
+      "loss": 1.3188,
+      "step": 2620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.696217313822588e-05,
+      "loss": 1.2691,
+      "step": 2621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.695981272939413e-05,
+      "loss": 1.1159,
+      "step": 2622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6957451463259807e-05,
+      "loss": 1.2822,
+      "step": 2623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6955089339915084e-05,
+      "loss": 1.1901,
+      "step": 2624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6952726359452193e-05,
+      "loss": 1.1782,
+      "step": 2625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.695036252196338e-05,
+      "loss": 1.2094,
+      "step": 2626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.694799782754093e-05,
+      "loss": 1.1844,
+      "step": 2627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6945632276277144e-05,
+      "loss": 1.2085,
+      "step": 2628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.694326586826438e-05,
+      "loss": 1.2656,
+      "step": 2629
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.694089860359504e-05,
+      "loss": 1.1867,
+      "step": 2630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.693853048236152e-05,
+      "loss": 1.2031,
+      "step": 2631
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.693616150465627e-05,
+      "loss": 1.1898,
+      "step": 2632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.693379167057179e-05,
+      "loss": 1.1362,
+      "step": 2633
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.693142098020059e-05,
+      "loss": 1.1329,
+      "step": 2634
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.692904943363522e-05,
+      "loss": 1.1627,
+      "step": 2635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.692667703096826e-05,
+      "loss": 1.1998,
+      "step": 2636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.692430377229233e-05,
+      "loss": 1.1276,
+      "step": 2637
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.692192965770008e-05,
+      "loss": 1.2265,
+      "step": 2638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.69195546872842e-05,
+      "loss": 1.2158,
+      "step": 2639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.69171788611374e-05,
+      "loss": 1.1884,
+      "step": 2640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.691480217935245e-05,
+      "loss": 1.0525,
+      "step": 2641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.69124246420221e-05,
+      "loss": 1.1342,
+      "step": 2642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.69100462492392e-05,
+      "loss": 1.0455,
+      "step": 2643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.690766700109659e-05,
+      "loss": 1.0907,
+      "step": 2644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.690528689768716e-05,
+      "loss": 1.1634,
+      "step": 2645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.690290593910381e-05,
+      "loss": 1.1121,
+      "step": 2646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.690052412543952e-05,
+      "loss": 0.9444,
+      "step": 2647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.689814145678726e-05,
+      "loss": 1.1101,
+      "step": 2648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.689575793324004e-05,
+      "loss": 1.0542,
+      "step": 2649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6893373554890925e-05,
+      "loss": 1.0712,
+      "step": 2650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6890988321833e-05,
+      "loss": 1.0758,
+      "step": 2651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.688860223415937e-05,
+      "loss": 1.1197,
+      "step": 2652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6886215291963197e-05,
+      "loss": 1.1432,
+      "step": 2653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.688382749533767e-05,
+      "loss": 1.1884,
+      "step": 2654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6881438844376e-05,
+      "loss": 1.2324,
+      "step": 2655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.687904933917145e-05,
+      "loss": 1.0674,
+      "step": 2656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.687665897981729e-05,
+      "loss": 1.1206,
+      "step": 2657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.687426776640685e-05,
+      "loss": 1.0811,
+      "step": 2658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.687187569903348e-05,
+      "loss": 1.1484,
+      "step": 2659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6869482777790565e-05,
+      "loss": 1.0893,
+      "step": 2660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.686708900277151e-05,
+      "loss": 1.1129,
+      "step": 2661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6864694374069796e-05,
+      "loss": 1.1853,
+      "step": 2662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.686229889177889e-05,
+      "loss": 1.1074,
+      "step": 2663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6859902555992305e-05,
+      "loss": 1.0923,
+      "step": 2664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.68575053668036e-05,
+      "loss": 1.1116,
+      "step": 2665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.685510732430637e-05,
+      "loss": 1.0844,
+      "step": 2666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6852708428594214e-05,
+      "loss": 1.0429,
+      "step": 2667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.68503086797608e-05,
+      "loss": 1.1578,
+      "step": 2668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.684790807789981e-05,
+      "loss": 1.0629,
+      "step": 2669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.684550662310496e-05,
+      "loss": 1.0204,
+      "step": 2670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.684310431547e-05,
+      "loss": 1.224,
+      "step": 2671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.684070115508872e-05,
+      "loss": 1.3826,
+      "step": 2672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6838297142054934e-05,
+      "loss": 1.3147,
+      "step": 2673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6835892276462496e-05,
+      "loss": 1.2721,
+      "step": 2674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.683348655840529e-05,
+      "loss": 1.3061,
+      "step": 2675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.683107998797723e-05,
+      "loss": 1.3396,
+      "step": 2676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.682867256527229e-05,
+      "loss": 1.332,
+      "step": 2677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.682626429038443e-05,
+      "loss": 1.361,
+      "step": 2678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.682385516340766e-05,
+      "loss": 1.3743,
+      "step": 2679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6821445184436066e-05,
+      "loss": 1.2532,
+      "step": 2680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.68190343535637e-05,
+      "loss": 1.3722,
+      "step": 2681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.68166226708847e-05,
+      "loss": 1.1682,
+      "step": 2682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.681421013649322e-05,
+      "loss": 1.2699,
+      "step": 2683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6811796750483416e-05,
+      "loss": 1.1777,
+      "step": 2684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.680938251294954e-05,
+      "loss": 1.3413,
+      "step": 2685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6806967423985815e-05,
+      "loss": 1.3365,
+      "step": 2686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6804551483686545e-05,
+      "loss": 1.2459,
+      "step": 2687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.680213469214604e-05,
+      "loss": 1.215,
+      "step": 2688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.679971704945865e-05,
+      "loss": 1.304,
+      "step": 2689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.679729855571876e-05,
+      "loss": 1.2034,
+      "step": 2690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6794879211020794e-05,
+      "loss": 1.1776,
+      "step": 2691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.679245901545918e-05,
+      "loss": 1.123,
+      "step": 2692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.679003796912842e-05,
+      "loss": 1.1524,
+      "step": 2693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6787616072123036e-05,
+      "loss": 1.1892,
+      "step": 2694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.678519332453757e-05,
+      "loss": 1.2207,
+      "step": 2695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.678276972646659e-05,
+      "loss": 1.1791,
+      "step": 2696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.678034527800474e-05,
+      "loss": 1.186,
+      "step": 2697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.677791997924665e-05,
+      "loss": 1.1552,
+      "step": 2698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.677549383028701e-05,
+      "loss": 1.0848,
+      "step": 2699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.677306683122054e-05,
+      "loss": 1.154,
+      "step": 2700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6770638982141966e-05,
+      "loss": 1.0979,
+      "step": 2701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.676821028314611e-05,
+      "loss": 1.0224,
+      "step": 2702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6765780734327743e-05,
+      "loss": 1.1898,
+      "step": 2703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.676335033578175e-05,
+      "loss": 1.1511,
+      "step": 2704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.676091908760301e-05,
+      "loss": 1.1534,
+      "step": 2705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.675848698988641e-05,
+      "loss": 1.1995,
+      "step": 2706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6756054042726915e-05,
+      "loss": 1.2309,
+      "step": 2707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.675362024621951e-05,
+      "loss": 1.1032,
+      "step": 2708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6751185600459216e-05,
+      "loss": 1.0986,
+      "step": 2709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6748750105541056e-05,
+      "loss": 1.1523,
+      "step": 2710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.674631376156013e-05,
+      "loss": 1.1127,
+      "step": 2711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.674387656861154e-05,
+      "loss": 1.2562,
+      "step": 2712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6741438526790454e-05,
+      "loss": 1.2527,
+      "step": 2713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.673899963619204e-05,
+      "loss": 1.2967,
+      "step": 2714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6736559896911496e-05,
+      "loss": 1.2074,
+      "step": 2715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6734119309044086e-05,
+      "loss": 1.215,
+      "step": 2716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.673167787268509e-05,
+      "loss": 1.2505,
+      "step": 2717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.672923558792982e-05,
+      "loss": 1.1771,
+      "step": 2718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.672679245487361e-05,
+      "loss": 1.2372,
+      "step": 2719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.672434847361186e-05,
+      "loss": 1.1834,
+      "step": 2720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.672190364423996e-05,
+      "loss": 1.3033,
+      "step": 2721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.671945796685336e-05,
+      "loss": 1.381,
+      "step": 2722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6717011441547545e-05,
+      "loss": 1.2691,
+      "step": 2723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.671456406841803e-05,
+      "loss": 1.2439,
+      "step": 2724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6712115847560355e-05,
+      "loss": 1.3344,
+      "step": 2725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.670966677907009e-05,
+      "loss": 1.2374,
+      "step": 2726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6707216863042855e-05,
+      "loss": 1.3244,
+      "step": 2727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6704766099574294e-05,
+      "loss": 1.2838,
+      "step": 2728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.670231448876008e-05,
+      "loss": 1.364,
+      "step": 2729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6699862030695924e-05,
+      "loss": 1.2997,
+      "step": 2730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6697408725477566e-05,
+      "loss": 1.3226,
+      "step": 2731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.669495457320079e-05,
+      "loss": 1.2427,
+      "step": 2732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.669249957396139e-05,
+      "loss": 1.177,
+      "step": 2733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.669004372785523e-05,
+      "loss": 1.1508,
+      "step": 2734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6687587034978166e-05,
+      "loss": 1.1781,
+      "step": 2735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.668512949542612e-05,
+      "loss": 1.1251,
+      "step": 2736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.668267110929502e-05,
+      "loss": 1.2288,
+      "step": 2737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6680211876680846e-05,
+      "loss": 1.2508,
+      "step": 2738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.667775179767961e-05,
+      "loss": 1.1528,
+      "step": 2739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.667529087238736e-05,
+      "loss": 1.1702,
+      "step": 2740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.667282910090015e-05,
+      "loss": 1.1946,
+      "step": 2741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.667036648331409e-05,
+      "loss": 1.2384,
+      "step": 2742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.666790301972533e-05,
+      "loss": 1.1782,
+      "step": 2743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6665438710230044e-05,
+      "loss": 1.2019,
+      "step": 2744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.666297355492443e-05,
+      "loss": 1.2341,
+      "step": 2745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6660507553904725e-05,
+      "loss": 1.1918,
+      "step": 2746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.66580407072672e-05,
+      "loss": 1.1384,
+      "step": 2747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.665557301510818e-05,
+      "loss": 1.1059,
+      "step": 2748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.665310447752397e-05,
+      "loss": 1.2371,
+      "step": 2749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.665063509461097e-05,
+      "loss": 1.3707,
+      "step": 2750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6648164866465563e-05,
+      "loss": 1.314,
+      "step": 2751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.66456937931842e-05,
+      "loss": 1.3391,
+      "step": 2752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.664322187486334e-05,
+      "loss": 1.3011,
+      "step": 2753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6640749111599494e-05,
+      "loss": 1.4052,
+      "step": 2754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.663827550348919e-05,
+      "loss": 1.3556,
+      "step": 2755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6635801050629003e-05,
+      "loss": 1.3604,
+      "step": 2756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.663332575311553e-05,
+      "loss": 1.4151,
+      "step": 2757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6630849611045416e-05,
+      "loss": 1.3155,
+      "step": 2758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.662837262451532e-05,
+      "loss": 1.413,
+      "step": 2759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.662589479362193e-05,
+      "loss": 1.2875,
+      "step": 2760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6623416118462005e-05,
+      "loss": 1.3487,
+      "step": 2761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.66209365991323e-05,
+      "loss": 1.3472,
+      "step": 2762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.661845623572961e-05,
+      "loss": 1.2496,
+      "step": 2763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.661597502835077e-05,
+      "loss": 1.2327,
+      "step": 2764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6613492977092656e-05,
+      "loss": 1.2288,
+      "step": 2765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.661101008205214e-05,
+      "loss": 1.219,
+      "step": 2766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.660852634332619e-05,
+      "loss": 1.2348,
+      "step": 2767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6606041761011736e-05,
+      "loss": 1.1958,
+      "step": 2768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6603556335205796e-05,
+      "loss": 1.2311,
+      "step": 2769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.660107006600539e-05,
+      "loss": 1.3407,
+      "step": 2770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6598582953507586e-05,
+      "loss": 1.363,
+      "step": 2771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6596094997809484e-05,
+      "loss": 1.3951,
+      "step": 2772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.65936061990082e-05,
+      "loss": 1.4004,
+      "step": 2773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.659111655720091e-05,
+      "loss": 1.3533,
+      "step": 2774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.65886260724848e-05,
+      "loss": 1.3917,
+      "step": 2775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.65861347449571e-05,
+      "loss": 1.3397,
+      "step": 2776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.658364257471507e-05,
+      "loss": 1.3558,
+      "step": 2777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6581149561855996e-05,
+      "loss": 1.2787,
+      "step": 2778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.657865570647722e-05,
+      "loss": 1.2927,
+      "step": 2779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.657616100867609e-05,
+      "loss": 1.2725,
+      "step": 2780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.657366546855e-05,
+      "loss": 1.3434,
+      "step": 2781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.657116908619638e-05,
+      "loss": 1.2654,
+      "step": 2782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.656867186171267e-05,
+      "loss": 1.2617,
+      "step": 2783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.656617379519639e-05,
+      "loss": 1.3021,
+      "step": 2784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.656367488674505e-05,
+      "loss": 1.4044,
+      "step": 2785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6561175136456183e-05,
+      "loss": 1.3189,
+      "step": 2786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.655867454442742e-05,
+      "loss": 1.2497,
+      "step": 2787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6556173110756354e-05,
+      "loss": 1.3248,
+      "step": 2788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.655367083554064e-05,
+      "loss": 1.2067,
+      "step": 2789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6551167718877986e-05,
+      "loss": 1.2335,
+      "step": 2790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.65486637608661e-05,
+      "loss": 1.2476,
+      "step": 2791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6546158961602736e-05,
+      "loss": 1.2016,
+      "step": 2792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6543653321185676e-05,
+      "loss": 1.2342,
+      "step": 2793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.654114683971275e-05,
+      "loss": 1.3885,
+      "step": 2794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.65386395172818e-05,
+      "loss": 1.3132,
+      "step": 2795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6536131353990716e-05,
+      "loss": 1.1908,
+      "step": 2796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.653362234993742e-05,
+      "loss": 1.2586,
+      "step": 2797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6531112505219854e-05,
+      "loss": 1.3029,
+      "step": 2798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.652860181993601e-05,
+      "loss": 1.246,
+      "step": 2799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.652609029418389e-05,
+      "loss": 1.103,
+      "step": 2800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6523577928061556e-05,
+      "loss": 1.1982,
+      "step": 2801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.652106472166708e-05,
+      "loss": 1.2915,
+      "step": 2802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.65185506750986e-05,
+      "loss": 1.1202,
+      "step": 2803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6516035788454224e-05,
+      "loss": 1.2794,
+      "step": 2804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.651352006183217e-05,
+      "loss": 1.2001,
+      "step": 2805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.651100349533062e-05,
+      "loss": 1.1851,
+      "step": 2806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.650848608904785e-05,
+      "loss": 1.124,
+      "step": 2807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6505967843082116e-05,
+      "loss": 1.2477,
+      "step": 2808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.650344875753174e-05,
+      "loss": 1.1564,
+      "step": 2809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.650092883249506e-05,
+      "loss": 1.3362,
+      "step": 2810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.649840806807045e-05,
+      "loss": 1.2542,
+      "step": 2811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6495886464356335e-05,
+      "loss": 1.2046,
+      "step": 2812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.649336402145115e-05,
+      "loss": 1.3337,
+      "step": 2813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6490840739453354e-05,
+      "loss": 1.2827,
+      "step": 2814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.648831661846148e-05,
+      "loss": 1.2576,
+      "step": 2815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.648579165857405e-05,
+      "loss": 1.2895,
+      "step": 2816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.648326585988966e-05,
+      "loss": 1.2955,
+      "step": 2817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6480739222506885e-05,
+      "loss": 1.1792,
+      "step": 2818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6478211746524385e-05,
+      "loss": 1.2318,
+      "step": 2819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.647568343204083e-05,
+      "loss": 1.2722,
+      "step": 2820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6473154279154915e-05,
+      "loss": 1.1874,
+      "step": 2821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.647062428796539e-05,
+      "loss": 1.2634,
+      "step": 2822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.646809345857102e-05,
+      "loss": 1.1787,
+      "step": 2823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.64655617910706e-05,
+      "loss": 1.1928,
+      "step": 2824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6463029285562974e-05,
+      "loss": 1.1652,
+      "step": 2825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.646049594214701e-05,
+      "loss": 1.1262,
+      "step": 2826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.64579617609216e-05,
+      "loss": 1.1895,
+      "step": 2827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.645542674198568e-05,
+      "loss": 1.1734,
+      "step": 2828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.645289088543824e-05,
+      "loss": 1.1294,
+      "step": 2829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6450354191378244e-05,
+      "loss": 1.0522,
+      "step": 2830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.644781665990473e-05,
+      "loss": 1.1448,
+      "step": 2831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.644527829111678e-05,
+      "loss": 1.1151,
+      "step": 2832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6442739085113484e-05,
+      "loss": 1.1343,
+      "step": 2833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.644019904199396e-05,
+      "loss": 1.0907,
+      "step": 2834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.643765816185739e-05,
+      "loss": 1.1438,
+      "step": 2835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.643511644480295e-05,
+      "loss": 1.1526,
+      "step": 2836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.643257389092988e-05,
+      "loss": 1.1622,
+      "step": 2837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.643003050033743e-05,
+      "loss": 1.2104,
+      "step": 2838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.642748627312491e-05,
+      "loss": 1.2601,
+      "step": 2839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.642494120939163e-05,
+      "loss": 1.24,
+      "step": 2840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.642239530923695e-05,
+      "loss": 1.3657,
+      "step": 2841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.641984857276027e-05,
+      "loss": 1.3928,
+      "step": 2842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6417301000060996e-05,
+      "loss": 1.3153,
+      "step": 2843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.641475259123861e-05,
+      "loss": 1.3115,
+      "step": 2844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.641220334639258e-05,
+      "loss": 1.3061,
+      "step": 2845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.640965326562243e-05,
+      "loss": 1.3734,
+      "step": 2846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.640710234902772e-05,
+      "loss": 1.3752,
+      "step": 2847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.640455059670804e-05,
+      "loss": 1.2526,
+      "step": 2848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6401998008763005e-05,
+      "loss": 1.3863,
+      "step": 2849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.639944458529226e-05,
+      "loss": 1.2264,
+      "step": 2850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.63968903263955e-05,
+      "loss": 1.2501,
+      "step": 2851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6394335232172435e-05,
+      "loss": 1.2059,
+      "step": 2852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.639177930272283e-05,
+      "loss": 1.0954,
+      "step": 2853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.638922253814645e-05,
+      "loss": 1.2403,
+      "step": 2854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.638666493854311e-05,
+      "loss": 1.0841,
+      "step": 2855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.638410650401267e-05,
+      "loss": 1.1867,
+      "step": 2856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6381547234655e-05,
+      "loss": 1.2387,
+      "step": 2857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6378987130570016e-05,
+      "loss": 1.1711,
+      "step": 2858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.637642619185767e-05,
+      "loss": 1.2581,
+      "step": 2859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6373864418617935e-05,
+      "loss": 1.3306,
+      "step": 2860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.637130181095082e-05,
+      "loss": 1.2983,
+      "step": 2861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6368738368956364e-05,
+      "loss": 1.2906,
+      "step": 2862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.636617409273466e-05,
+      "loss": 1.3356,
+      "step": 2863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.636360898238579e-05,
+      "loss": 1.2266,
+      "step": 2864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.636104303800992e-05,
+      "loss": 1.3034,
+      "step": 2865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.635847625970721e-05,
+      "loss": 1.2654,
+      "step": 2866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.635590864757786e-05,
+      "loss": 1.4172,
+      "step": 2867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.635334020172213e-05,
+      "loss": 1.2937,
+      "step": 2868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.635077092224027e-05,
+      "loss": 1.2298,
+      "step": 2869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.63482008092326e-05,
+      "loss": 1.2213,
+      "step": 2870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.634562986279945e-05,
+      "loss": 1.1927,
+      "step": 2871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.634305808304118e-05,
+      "loss": 1.2792,
+      "step": 2872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.634048547005819e-05,
+      "loss": 1.2609,
+      "step": 2873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6337912023950936e-05,
+      "loss": 1.1638,
+      "step": 2874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6335337744819873e-05,
+      "loss": 1.1425,
+      "step": 2875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6332762632765495e-05,
+      "loss": 1.2014,
+      "step": 2876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6330186687888325e-05,
+      "loss": 1.2468,
+      "step": 2877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.632760991028895e-05,
+      "loss": 1.1638,
+      "step": 2878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.632503230006795e-05,
+      "loss": 1.2054,
+      "step": 2879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.632245385732596e-05,
+      "loss": 1.2541,
+      "step": 2880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6319874582163636e-05,
+      "loss": 1.0967,
+      "step": 2881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.631729447468167e-05,
+      "loss": 1.2124,
+      "step": 2882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.63147135349808e-05,
+      "loss": 1.1261,
+      "step": 2883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.631213176316179e-05,
+      "loss": 1.0699,
+      "step": 2884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.63095491593254e-05,
+      "loss": 1.1489,
+      "step": 2885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6306965723572494e-05,
+      "loss": 1.132,
+      "step": 2886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.630438145600389e-05,
+      "loss": 1.1504,
+      "step": 2887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6301796356720506e-05,
+      "loss": 1.2846,
+      "step": 2888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.629921042582325e-05,
+      "loss": 1.3005,
+      "step": 2889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.629662366341307e-05,
+      "loss": 1.2698,
+      "step": 2890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.629403606959097e-05,
+      "loss": 1.2849,
+      "step": 2891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.629144764445796e-05,
+      "loss": 1.3924,
+      "step": 2892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6288858388115086e-05,
+      "loss": 1.316,
+      "step": 2893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6286268300663435e-05,
+      "loss": 1.2875,
+      "step": 2894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6283677382204125e-05,
+      "loss": 1.3491,
+      "step": 2895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6281085632838304e-05,
+      "loss": 1.3595,
+      "step": 2896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.627849305266715e-05,
+      "loss": 1.2195,
+      "step": 2897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.627589964179188e-05,
+      "loss": 1.1688,
+      "step": 2898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.627330540031374e-05,
+      "loss": 1.1224,
+      "step": 2899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6270710328334004e-05,
+      "loss": 1.2531,
+      "step": 2900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.626811442595399e-05,
+      "loss": 1.3093,
+      "step": 2901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6265517693275027e-05,
+      "loss": 1.2429,
+      "step": 2902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.62629201303985e-05,
+      "loss": 1.1221,
+      "step": 2903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.626032173742582e-05,
+      "loss": 1.2434,
+      "step": 2904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6257722514458426e-05,
+      "loss": 1.226,
+      "step": 2905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.625512246159778e-05,
+      "loss": 1.2356,
+      "step": 2906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6252521578945395e-05,
+      "loss": 1.3211,
+      "step": 2907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6249919866602815e-05,
+      "loss": 1.3181,
+      "step": 2908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6247317324671605e-05,
+      "loss": 1.2609,
+      "step": 2909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.624471395325336e-05,
+      "loss": 1.2549,
+      "step": 2910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.624210975244972e-05,
+      "loss": 1.26,
+      "step": 2911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6239504722362356e-05,
+      "loss": 1.3228,
+      "step": 2912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.623689886309296e-05,
+      "loss": 1.2561,
+      "step": 2913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6234292174743267e-05,
+      "loss": 1.2789,
+      "step": 2914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.623168465741504e-05,
+      "loss": 1.5097,
+      "step": 2915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6229076311210085e-05,
+      "loss": 1.2722,
+      "step": 2916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.622646713623022e-05,
+      "loss": 1.2035,
+      "step": 2917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.622385713257731e-05,
+      "loss": 1.2932,
+      "step": 2918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.622124630035324e-05,
+      "loss": 1.3448,
+      "step": 2919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6218634639659954e-05,
+      "loss": 1.3451,
+      "step": 2920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.621602215059939e-05,
+      "loss": 1.2289,
+      "step": 2921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.621340883327355e-05,
+      "loss": 1.381,
+      "step": 2922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6210794687784456e-05,
+      "loss": 1.2526,
+      "step": 2923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6208179714234165e-05,
+      "loss": 1.2066,
+      "step": 2924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.620556391272476e-05,
+      "loss": 1.1601,
+      "step": 2925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.620294728335836e-05,
+      "loss": 1.1526,
+      "step": 2926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.620032982623712e-05,
+      "loss": 1.1372,
+      "step": 2927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6197711541463226e-05,
+      "loss": 1.1613,
+      "step": 2928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6195092429138894e-05,
+      "loss": 1.1451,
+      "step": 2929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6192472489366365e-05,
+      "loss": 1.1315,
+      "step": 2930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.618985172224794e-05,
+      "loss": 1.2039,
+      "step": 2931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6187230127885915e-05,
+      "loss": 1.0796,
+      "step": 2932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.618460770638264e-05,
+      "loss": 1.0736,
+      "step": 2933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.61819844578405e-05,
+      "loss": 1.2207,
+      "step": 2934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.617936038236189e-05,
+      "loss": 1.1724,
+      "step": 2935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.617673548004927e-05,
+      "loss": 1.2777,
+      "step": 2936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6174109751005114e-05,
+      "loss": 1.2398,
+      "step": 2937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.617148319533191e-05,
+      "loss": 1.2125,
+      "step": 2938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.616885581313223e-05,
+      "loss": 1.2837,
+      "step": 2939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6166227604508616e-05,
+      "loss": 1.1963,
+      "step": 2940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6163598569563684e-05,
+      "loss": 1.2278,
+      "step": 2941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.616096870840007e-05,
+      "loss": 1.2451,
+      "step": 2942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.615833802112044e-05,
+      "loss": 1.193,
+      "step": 2943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.615570650782751e-05,
+      "loss": 1.1087,
+      "step": 2944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.615307416862399e-05,
+      "loss": 1.1729,
+      "step": 2945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.615044100361267e-05,
+      "loss": 1.1531,
+      "step": 2946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.614780701289632e-05,
+      "loss": 1.1219,
+      "step": 2947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.614517219657779e-05,
+      "loss": 1.2309,
+      "step": 2948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.614253655475994e-05,
+      "loss": 1.2466,
+      "step": 2949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613990008754565e-05,
+      "loss": 1.2443,
+      "step": 2950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613726279503787e-05,
+      "loss": 1.1775,
+      "step": 2951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613462467733954e-05,
+      "loss": 1.0913,
+      "step": 2952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6131985734553664e-05,
+      "loss": 1.223,
+      "step": 2953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6129345966783255e-05,
+      "loss": 1.2475,
+      "step": 2954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.612670537413138e-05,
+      "loss": 1.2003,
+      "step": 2955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.612406395670111e-05,
+      "loss": 1.2071,
+      "step": 2956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.612142171459558e-05,
+      "loss": 1.226,
+      "step": 2957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.611877864791793e-05,
+      "loss": 1.2257,
+      "step": 2958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.611613475677136e-05,
+      "loss": 1.1696,
+      "step": 2959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.611349004125907e-05,
+      "loss": 1.254,
+      "step": 2960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.611084450148432e-05,
+      "loss": 1.2518,
+      "step": 2961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.610819813755038e-05,
+      "loss": 1.2565,
+      "step": 2962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.610555094956057e-05,
+      "loss": 1.2902,
+      "step": 2963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.610290293761824e-05,
+      "loss": 1.2734,
+      "step": 2964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.610025410182677e-05,
+      "loss": 1.2024,
+      "step": 2965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6097604442289544e-05,
+      "loss": 1.2562,
+      "step": 2966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.609495395911003e-05,
+      "loss": 1.2419,
+      "step": 2967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.609230265239169e-05,
+      "loss": 1.2603,
+      "step": 2968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6089650522238045e-05,
+      "loss": 1.3012,
+      "step": 2969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.608699756875262e-05,
+      "loss": 1.3108,
+      "step": 2970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6084343792038974e-05,
+      "loss": 1.2576,
+      "step": 2971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.608168919220073e-05,
+      "loss": 1.2735,
+      "step": 2972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.607903376934152e-05,
+      "loss": 1.2936,
+      "step": 2973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.607637752356499e-05,
+      "loss": 1.2225,
+      "step": 2974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6073720454974864e-05,
+      "loss": 1.2581,
+      "step": 2975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.607106256367486e-05,
+      "loss": 1.1536,
+      "step": 2976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.606840384976875e-05,
+      "loss": 1.2824,
+      "step": 2977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.606574431336032e-05,
+      "loss": 1.2592,
+      "step": 2978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.60630839545534e-05,
+      "loss": 1.199,
+      "step": 2979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.606042277345185e-05,
+      "loss": 1.2844,
+      "step": 2980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6057760770159566e-05,
+      "loss": 1.2586,
+      "step": 2981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.605509794478046e-05,
+      "loss": 1.2164,
+      "step": 2982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6052434297418495e-05,
+      "loss": 1.2466,
+      "step": 2983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6049769828177665e-05,
+      "loss": 1.2774,
+      "step": 2984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6047104537161975e-05,
+      "loss": 1.2212,
+      "step": 2985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.604443842447549e-05,
+      "loss": 1.1987,
+      "step": 2986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.604177149022229e-05,
+      "loss": 1.2509,
+      "step": 2987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.603910373450648e-05,
+      "loss": 1.2549,
+      "step": 2988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6036435157432235e-05,
+      "loss": 1.3268,
+      "step": 2989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.603376575910371e-05,
+      "loss": 1.236,
+      "step": 2990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.603109553962512e-05,
+      "loss": 1.3011,
+      "step": 2991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6028424499100734e-05,
+      "loss": 1.3047,
+      "step": 2992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6025752637634786e-05,
+      "loss": 1.3152,
+      "step": 2993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6023079955331625e-05,
+      "loss": 1.3373,
+      "step": 2994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.602040645229557e-05,
+      "loss": 1.2996,
+      "step": 2995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6017732128631e-05,
+      "loss": 1.1637,
+      "step": 2996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.601505698444231e-05,
+      "loss": 1.3412,
+      "step": 2997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.601238101983395e-05,
+      "loss": 1.2755,
+      "step": 2998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6009704234910383e-05,
+      "loss": 1.2822,
+      "step": 2999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6007026629776104e-05,
+      "loss": 1.278,
+      "step": 3000
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.17154860496521,
+      "eval_runtime": 116.7089,
+      "eval_samples_per_second": 32.74,
+      "eval_steps_per_second": 0.343,
+      "step": 3000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.600434820453566e-05,
+      "loss": 1.1962,
+      "step": 3001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.60016689592936e-05,
+      "loss": 1.1573,
+      "step": 3002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.599898889415454e-05,
+      "loss": 1.2295,
+      "step": 3003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5996308009223085e-05,
+      "loss": 1.1824,
+      "step": 3004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.599362630460391e-05,
+      "loss": 1.1798,
+      "step": 3005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5990943780401695e-05,
+      "loss": 1.1868,
+      "step": 3006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.598826043672119e-05,
+      "loss": 1.1649,
+      "step": 3007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.598557627366712e-05,
+      "loss": 1.1141,
+      "step": 3008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.59828912913443e-05,
+      "loss": 1.2034,
+      "step": 3009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.598020548985754e-05,
+      "loss": 1.1936,
+      "step": 3010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.597751886931169e-05,
+      "loss": 1.2505,
+      "step": 3011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.597483142981165e-05,
+      "loss": 1.2906,
+      "step": 3012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.597214317146231e-05,
+      "loss": 1.2031,
+      "step": 3013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5969454094368634e-05,
+      "loss": 1.2415,
+      "step": 3014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5966764198635606e-05,
+      "loss": 1.31,
+      "step": 3015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.596407348436823e-05,
+      "loss": 1.2396,
+      "step": 3016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5961381951671554e-05,
+      "loss": 1.2626,
+      "step": 3017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.595868960065066e-05,
+      "loss": 1.275,
+      "step": 3018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5955996431410645e-05,
+      "loss": 1.2753,
+      "step": 3019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5953302444056655e-05,
+      "loss": 1.1274,
+      "step": 3020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.595060763869387e-05,
+      "loss": 1.2847,
+      "step": 3021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5947912015427486e-05,
+      "loss": 1.2664,
+      "step": 3022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.594521557436273e-05,
+      "loss": 1.2031,
+      "step": 3023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.594251831560489e-05,
+      "loss": 1.2437,
+      "step": 3024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.593982023925926e-05,
+      "loss": 1.2541,
+      "step": 3025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.593712134543116e-05,
+      "loss": 1.2674,
+      "step": 3026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.593442163422596e-05,
+      "loss": 1.159,
+      "step": 3027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.593172110574905e-05,
+      "loss": 1.2515,
+      "step": 3028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.592901976010587e-05,
+      "loss": 1.1575,
+      "step": 3029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.592631759740188e-05,
+      "loss": 1.204,
+      "step": 3030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.592361461774256e-05,
+      "loss": 1.1851,
+      "step": 3031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5920910821233445e-05,
+      "loss": 1.1518,
+      "step": 3032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.591820620798008e-05,
+      "loss": 1.1978,
+      "step": 3033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5915500778088054e-05,
+      "loss": 1.1894,
+      "step": 3034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.591279453166299e-05,
+      "loss": 1.256,
+      "step": 3035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5910087468810534e-05,
+      "loss": 1.2339,
+      "step": 3036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.590737958963638e-05,
+      "loss": 1.2001,
+      "step": 3037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5904670894246224e-05,
+      "loss": 1.2633,
+      "step": 3038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5901961382745825e-05,
+      "loss": 1.2602,
+      "step": 3039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5899251055240963e-05,
+      "loss": 1.4193,
+      "step": 3040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.589653991183744e-05,
+      "loss": 1.3032,
+      "step": 3041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.589382795264111e-05,
+      "loss": 1.3406,
+      "step": 3042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5891115177757826e-05,
+      "loss": 1.5028,
+      "step": 3043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.588840158729351e-05,
+      "loss": 1.3549,
+      "step": 3044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5885687181354095e-05,
+      "loss": 1.4421,
+      "step": 3045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.588297196004556e-05,
+      "loss": 1.2929,
+      "step": 3046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.588025592347389e-05,
+      "loss": 1.3211,
+      "step": 3047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.587753907174512e-05,
+      "loss": 1.5841,
+      "step": 3048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5874821404965335e-05,
+      "loss": 1.4387,
+      "step": 3049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.587210292324061e-05,
+      "loss": 1.3701,
+      "step": 3050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5869383626677074e-05,
+      "loss": 1.4465,
+      "step": 3051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.58666635153809e-05,
+      "loss": 1.2992,
+      "step": 3052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.586394258945827e-05,
+      "loss": 1.3077,
+      "step": 3053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.586122084901542e-05,
+      "loss": 1.457,
+      "step": 3054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5858498294158594e-05,
+      "loss": 1.3922,
+      "step": 3055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5855774924994074e-05,
+      "loss": 1.5346,
+      "step": 3056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5853050741628194e-05,
+      "loss": 1.6082,
+      "step": 3057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.58503257441673e-05,
+      "loss": 1.5003,
+      "step": 3058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.584759993271778e-05,
+      "loss": 1.2595,
+      "step": 3059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5844873307386025e-05,
+      "loss": 1.1351,
+      "step": 3060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.584214586827852e-05,
+      "loss": 1.1479,
+      "step": 3061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.583941761550171e-05,
+      "loss": 1.1981,
+      "step": 3062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.583668854916212e-05,
+      "loss": 1.0981,
+      "step": 3063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.583395866936628e-05,
+      "loss": 1.1523,
+      "step": 3064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.583122797622078e-05,
+      "loss": 1.1742,
+      "step": 3065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.582849646983222e-05,
+      "loss": 1.0453,
+      "step": 3066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5825764150307236e-05,
+      "loss": 1.1675,
+      "step": 3067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5823031017752485e-05,
+      "loss": 1.1579,
+      "step": 3068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.582029707227469e-05,
+      "loss": 1.1309,
+      "step": 3069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.581756231398057e-05,
+      "loss": 1.1183,
+      "step": 3070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.581482674297688e-05,
+      "loss": 1.2403,
+      "step": 3071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.581209035937043e-05,
+      "loss": 1.1587,
+      "step": 3072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.580935316326804e-05,
+      "loss": 1.1885,
+      "step": 3073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.580661515477657e-05,
+      "loss": 1.176,
+      "step": 3074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.580387633400292e-05,
+      "loss": 1.2018,
+      "step": 3075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5801136701054e-05,
+      "loss": 1.1746,
+      "step": 3076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.579839625603677e-05,
+      "loss": 1.1425,
+      "step": 3077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.579565499905821e-05,
+      "loss": 1.185,
+      "step": 3078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.579291293022535e-05,
+      "loss": 1.0988,
+      "step": 3079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.579017004964523e-05,
+      "loss": 1.1964,
+      "step": 3080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.578742635742493e-05,
+      "loss": 1.3045,
+      "step": 3081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5784681853671575e-05,
+      "loss": 1.0835,
+      "step": 3082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.578193653849229e-05,
+      "loss": 1.1712,
+      "step": 3083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.577919041199427e-05,
+      "loss": 1.2133,
+      "step": 3084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.57764434742847e-05,
+      "loss": 1.1765,
+      "step": 3085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.577369572547084e-05,
+      "loss": 1.2004,
+      "step": 3086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.577094716565996e-05,
+      "loss": 1.1543,
+      "step": 3087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.576819779495935e-05,
+      "loss": 1.0517,
+      "step": 3088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5765447613476354e-05,
+      "loss": 1.1547,
+      "step": 3089
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5762696621318335e-05,
+      "loss": 1.1815,
+      "step": 3090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5759944818592694e-05,
+      "loss": 1.1581,
+      "step": 3091
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.575719220540685e-05,
+      "loss": 1.2917,
+      "step": 3092
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.575443878186828e-05,
+      "loss": 1.1619,
+      "step": 3093
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.575168454808446e-05,
+      "loss": 1.1565,
+      "step": 3094
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574892950416293e-05,
+      "loss": 1.2122,
+      "step": 3095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5746173650211234e-05,
+      "loss": 1.1119,
+      "step": 3096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5743416986336965e-05,
+      "loss": 1.2191,
+      "step": 3097
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574065951264773e-05,
+      "loss": 1.1272,
+      "step": 3098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.57379012292512e-05,
+      "loss": 1.1889,
+      "step": 3099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.573514213625505e-05,
+      "loss": 1.099,
+      "step": 3100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5732382233766985e-05,
+      "loss": 1.0982,
+      "step": 3101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5729621521894764e-05,
+      "loss": 1.1923,
+      "step": 3102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.572686000074615e-05,
+      "loss": 1.2229,
+      "step": 3103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.572409767042895e-05,
+      "loss": 1.1384,
+      "step": 3104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.572133453105103e-05,
+      "loss": 1.149,
+      "step": 3105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.571857058272024e-05,
+      "loss": 1.1607,
+      "step": 3106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.571580582554449e-05,
+      "loss": 1.1438,
+      "step": 3107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.571304025963171e-05,
+      "loss": 1.3579,
+      "step": 3108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5710273885089875e-05,
+      "loss": 1.3862,
+      "step": 3109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.570750670202697e-05,
+      "loss": 1.4014,
+      "step": 3110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.570473871055103e-05,
+      "loss": 1.4615,
+      "step": 3111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.570196991077013e-05,
+      "loss": 1.4262,
+      "step": 3112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.569920030279234e-05,
+      "loss": 1.4575,
+      "step": 3113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.569642988672581e-05,
+      "loss": 1.4327,
+      "step": 3114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.569365866267868e-05,
+      "loss": 1.4188,
+      "step": 3115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.569088663075914e-05,
+      "loss": 1.4177,
+      "step": 3116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5688113791075406e-05,
+      "loss": 1.3848,
+      "step": 3117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5685340143735735e-05,
+      "loss": 1.3728,
+      "step": 3118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.56825656888484e-05,
+      "loss": 1.2199,
+      "step": 3119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.567979042652172e-05,
+      "loss": 1.2388,
+      "step": 3120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.567701435686404e-05,
+      "loss": 1.2886,
+      "step": 3121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.567423747998375e-05,
+      "loss": 1.354,
+      "step": 3122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.567145979598924e-05,
+      "loss": 1.2511,
+      "step": 3123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5668681304988944e-05,
+      "loss": 1.264,
+      "step": 3124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.566590200709136e-05,
+      "loss": 1.2773,
+      "step": 3125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.566312190240497e-05,
+      "loss": 1.2883,
+      "step": 3126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.566034099103831e-05,
+      "loss": 1.2636,
+      "step": 3127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5657559273099945e-05,
+      "loss": 1.2926,
+      "step": 3128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.565477674869849e-05,
+      "loss": 1.2844,
+      "step": 3129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.565199341794255e-05,
+      "loss": 1.3174,
+      "step": 3130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.56492092809408e-05,
+      "loss": 1.23,
+      "step": 3131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.564642433780192e-05,
+      "loss": 1.2254,
+      "step": 3132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.564363858863464e-05,
+      "loss": 1.3422,
+      "step": 3133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.564085203354772e-05,
+      "loss": 1.3009,
+      "step": 3134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.563806467264994e-05,
+      "loss": 1.293,
+      "step": 3135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.563527650605012e-05,
+      "loss": 1.3504,
+      "step": 3136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5632487533857096e-05,
+      "loss": 1.3494,
+      "step": 3137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5629697756179766e-05,
+      "loss": 1.2477,
+      "step": 3138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5626907173127044e-05,
+      "loss": 1.2475,
+      "step": 3139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.562411578480785e-05,
+      "loss": 1.1649,
+      "step": 3140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5621323591331176e-05,
+      "loss": 1.1956,
+      "step": 3141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.561853059280603e-05,
+      "loss": 1.2374,
+      "step": 3142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.561573678934145e-05,
+      "loss": 1.2423,
+      "step": 3143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.561294218104649e-05,
+      "loss": 1.1764,
+      "step": 3144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.561014676803026e-05,
+      "loss": 1.2239,
+      "step": 3145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.56073505504019e-05,
+      "loss": 1.2182,
+      "step": 3146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5604553528270555e-05,
+      "loss": 1.1773,
+      "step": 3147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.560175570174544e-05,
+      "loss": 1.1925,
+      "step": 3148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.559895707093577e-05,
+      "loss": 1.1681,
+      "step": 3149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.559615763595079e-05,
+      "loss": 1.3758,
+      "step": 3150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.559335739689982e-05,
+      "loss": 1.4268,
+      "step": 3151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.559055635389216e-05,
+      "loss": 1.3875,
+      "step": 3152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.558775450703715e-05,
+      "loss": 1.3104,
+      "step": 3153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5584951856444206e-05,
+      "loss": 1.3211,
+      "step": 3154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.55821484022227e-05,
+      "loss": 1.3641,
+      "step": 3155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5579344144482116e-05,
+      "loss": 1.2945,
+      "step": 3156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.557653908333192e-05,
+      "loss": 1.203,
+      "step": 3157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.557373321888161e-05,
+      "loss": 1.252,
+      "step": 3158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.557092655124075e-05,
+      "loss": 1.3686,
+      "step": 3159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5568119080518864e-05,
+      "loss": 1.339,
+      "step": 3160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.556531080682561e-05,
+      "loss": 1.3649,
+      "step": 3161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.556250173027059e-05,
+      "loss": 1.4198,
+      "step": 3162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.555969185096347e-05,
+      "loss": 1.3071,
+      "step": 3163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.555688116901395e-05,
+      "loss": 1.3406,
+      "step": 3164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.555406968453178e-05,
+      "loss": 1.3897,
+      "step": 3165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5551257397626676e-05,
+      "loss": 1.3111,
+      "step": 3166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.554844430840847e-05,
+      "loss": 1.4271,
+      "step": 3167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.554563041698695e-05,
+      "loss": 1.3531,
+      "step": 3168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5542815723472e-05,
+      "loss": 1.3514,
+      "step": 3169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.554000022797348e-05,
+      "loss": 1.1991,
+      "step": 3170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.553718393060132e-05,
+      "loss": 1.3034,
+      "step": 3171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.553436683146547e-05,
+      "loss": 1.2773,
+      "step": 3172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5531548930675886e-05,
+      "loss": 1.2826,
+      "step": 3173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5528730228342605e-05,
+      "loss": 1.2966,
+      "step": 3174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.552591072457565e-05,
+      "loss": 1.3094,
+      "step": 3175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.55230904194851e-05,
+      "loss": 1.264,
+      "step": 3176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.552026931318106e-05,
+      "loss": 1.2121,
+      "step": 3177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.551744740577367e-05,
+      "loss": 1.3765,
+      "step": 3178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.551462469737307e-05,
+      "loss": 1.3709,
+      "step": 3179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5511801188089497e-05,
+      "loss": 1.2505,
+      "step": 3180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.550897687803315e-05,
+      "loss": 1.2349,
+      "step": 3181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.550615176731431e-05,
+      "loss": 1.2823,
+      "step": 3182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.550332585604324e-05,
+      "loss": 1.1931,
+      "step": 3183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.550049914433029e-05,
+      "loss": 1.2425,
+      "step": 3184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.549767163228579e-05,
+      "loss": 1.209,
+      "step": 3185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5494843320020155e-05,
+      "loss": 1.3753,
+      "step": 3186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.549201420764377e-05,
+      "loss": 1.2105,
+      "step": 3187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.54891842952671e-05,
+      "loss": 1.2534,
+      "step": 3188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5486353583000625e-05,
+      "loss": 1.2865,
+      "step": 3189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5483522070954845e-05,
+      "loss": 1.1901,
+      "step": 3190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.548068975924031e-05,
+      "loss": 1.1156,
+      "step": 3191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.547785664796759e-05,
+      "loss": 1.2755,
+      "step": 3192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.547502273724727e-05,
+      "loss": 1.2708,
+      "step": 3193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5472188027190024e-05,
+      "loss": 1.2199,
+      "step": 3194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.546935251790648e-05,
+      "loss": 1.3284,
+      "step": 3195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5466516209507365e-05,
+      "loss": 1.177,
+      "step": 3196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.546367910210338e-05,
+      "loss": 1.2168,
+      "step": 3197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.546084119580531e-05,
+      "loss": 1.2331,
+      "step": 3198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5458002490723926e-05,
+      "loss": 1.1785,
+      "step": 3199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.545516298697006e-05,
+      "loss": 1.0521,
+      "step": 3200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.545232268465457e-05,
+      "loss": 1.0302,
+      "step": 3201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.544948158388832e-05,
+      "loss": 1.1535,
+      "step": 3202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.544663968478224e-05,
+      "loss": 1.0536,
+      "step": 3203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.544379698744728e-05,
+      "loss": 1.0867,
+      "step": 3204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.544095349199442e-05,
+      "loss": 1.015,
+      "step": 3205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.543810919853465e-05,
+      "loss": 1.1303,
+      "step": 3206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.543526410717903e-05,
+      "loss": 1.0079,
+      "step": 3207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.543241821803862e-05,
+      "loss": 1.1945,
+      "step": 3208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.542957153122452e-05,
+      "loss": 1.0151,
+      "step": 3209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.542672404684788e-05,
+      "loss": 1.1667,
+      "step": 3210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.542387576501985e-05,
+      "loss": 1.2389,
+      "step": 3211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.542102668585163e-05,
+      "loss": 1.2701,
+      "step": 3212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5418176809454445e-05,
+      "loss": 1.2427,
+      "step": 3213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.541532613593955e-05,
+      "loss": 1.2192,
+      "step": 3214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.541247466541825e-05,
+      "loss": 1.1067,
+      "step": 3215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.540962239800185e-05,
+      "loss": 1.2126,
+      "step": 3216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.54067693338017e-05,
+      "loss": 1.3123,
+      "step": 3217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.540391547292919e-05,
+      "loss": 1.165,
+      "step": 3218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.540106081549574e-05,
+      "loss": 1.3023,
+      "step": 3219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.539820536161278e-05,
+      "loss": 1.2724,
+      "step": 3220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5395349111391795e-05,
+      "loss": 1.231,
+      "step": 3221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.539249206494429e-05,
+      "loss": 1.1705,
+      "step": 3222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5389634222381795e-05,
+      "loss": 1.211,
+      "step": 3223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5386775583815896e-05,
+      "loss": 1.2718,
+      "step": 3224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5383916149358175e-05,
+      "loss": 1.2852,
+      "step": 3225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.538105591912028e-05,
+      "loss": 1.372,
+      "step": 3226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.537819489321386e-05,
+      "loss": 1.254,
+      "step": 3227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.537533307175061e-05,
+      "loss": 1.2948,
+      "step": 3228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5372470454842264e-05,
+      "loss": 1.0065,
+      "step": 3229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.536960704260057e-05,
+      "loss": 0.9944,
+      "step": 3230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5366742835137314e-05,
+      "loss": 1.1018,
+      "step": 3231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5363877832564315e-05,
+      "loss": 1.052,
+      "step": 3232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.536101203499342e-05,
+      "loss": 1.1735,
+      "step": 3233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5358145442536516e-05,
+      "loss": 1.0795,
+      "step": 3234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.535527805530551e-05,
+      "loss": 1.1076,
+      "step": 3235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.535240987341233e-05,
+      "loss": 1.0164,
+      "step": 3236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.534954089696897e-05,
+      "loss": 1.0571,
+      "step": 3237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5346671126087425e-05,
+      "loss": 1.1145,
+      "step": 3238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.534380056087972e-05,
+      "loss": 1.1749,
+      "step": 3239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.534092920145794e-05,
+      "loss": 1.4365,
+      "step": 3240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5338057047934165e-05,
+      "loss": 1.2896,
+      "step": 3241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5335184100420535e-05,
+      "loss": 1.3344,
+      "step": 3242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.533231035902919e-05,
+      "loss": 1.3437,
+      "step": 3243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5329435823872345e-05,
+      "loss": 1.3821,
+      "step": 3244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.532656049506221e-05,
+      "loss": 1.3968,
+      "step": 3245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.532368437271103e-05,
+      "loss": 1.4528,
+      "step": 3246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.532080745693109e-05,
+      "loss": 1.4623,
+      "step": 3247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.531792974783471e-05,
+      "loss": 1.3281,
+      "step": 3248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.531505124553423e-05,
+      "loss": 1.3796,
+      "step": 3249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5312171950142034e-05,
+      "loss": 1.4083,
+      "step": 3250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.530929186177052e-05,
+      "loss": 1.2772,
+      "step": 3251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.530641098053212e-05,
+      "loss": 1.1628,
+      "step": 3252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.530352930653931e-05,
+      "loss": 1.191,
+      "step": 3253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.530064683990459e-05,
+      "loss": 1.1573,
+      "step": 3254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.529776358074049e-05,
+      "loss": 1.2274,
+      "step": 3255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.529487952915957e-05,
+      "loss": 1.184,
+      "step": 3256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.529199468527443e-05,
+      "loss": 1.1617,
+      "step": 3257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.528910904919768e-05,
+      "loss": 1.3063,
+      "step": 3258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5286222621041976e-05,
+      "loss": 1.262,
+      "step": 3259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5283335400920016e-05,
+      "loss": 1.1697,
+      "step": 3260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.52804473889445e-05,
+      "loss": 1.3491,
+      "step": 3261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.527755858522819e-05,
+      "loss": 1.3584,
+      "step": 3262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.527466898988384e-05,
+      "loss": 1.2255,
+      "step": 3263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5271778603024286e-05,
+      "loss": 1.2797,
+      "step": 3264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526888742476235e-05,
+      "loss": 1.2417,
+      "step": 3265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526599545521092e-05,
+      "loss": 1.2898,
+      "step": 3266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526310269448288e-05,
+      "loss": 1.266,
+      "step": 3267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5260209142691166e-05,
+      "loss": 1.2544,
+      "step": 3268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.525731479994875e-05,
+      "loss": 1.3268,
+      "step": 3269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.525441966636861e-05,
+      "loss": 1.275,
+      "step": 3270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.525152374206378e-05,
+      "loss": 1.3202,
+      "step": 3271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.524862702714732e-05,
+      "loss": 1.1787,
+      "step": 3272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5245729521732314e-05,
+      "loss": 1.1273,
+      "step": 3273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5242831225931874e-05,
+      "loss": 1.1488,
+      "step": 3274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5239932139859154e-05,
+      "loss": 1.0972,
+      "step": 3275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.523703226362733e-05,
+      "loss": 1.113,
+      "step": 3276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.523413159734962e-05,
+      "loss": 1.1813,
+      "step": 3277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.523123014113926e-05,
+      "loss": 1.1782,
+      "step": 3278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.522832789510951e-05,
+      "loss": 1.0932,
+      "step": 3279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.522542485937369e-05,
+      "loss": 1.1234,
+      "step": 3280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.522252103404512e-05,
+      "loss": 1.0704,
+      "step": 3281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5219616419237175e-05,
+      "loss": 1.1026,
+      "step": 3282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.521671101506325e-05,
+      "loss": 1.1456,
+      "step": 3283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.521380482163676e-05,
+      "loss": 1.1608,
+      "step": 3284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5210897839071166e-05,
+      "loss": 1.1121,
+      "step": 3285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.520799006747996e-05,
+      "loss": 1.1091,
+      "step": 3286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.520508150697666e-05,
+      "loss": 1.169,
+      "step": 3287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.520217215767481e-05,
+      "loss": 1.2144,
+      "step": 3288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5199262019687996e-05,
+      "loss": 1.1351,
+      "step": 3289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.519635109312982e-05,
+      "loss": 1.1138,
+      "step": 3290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5193439378113936e-05,
+      "loss": 1.252,
+      "step": 3291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5190526874754e-05,
+      "loss": 1.0887,
+      "step": 3292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5187613583163736e-05,
+      "loss": 1.1036,
+      "step": 3293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.518469950345685e-05,
+      "loss": 1.1556,
+      "step": 3294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5181784635747136e-05,
+      "loss": 1.1844,
+      "step": 3295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.517886898014837e-05,
+      "loss": 1.2456,
+      "step": 3296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5175952536774385e-05,
+      "loss": 1.1258,
+      "step": 3297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5173035305739034e-05,
+      "loss": 1.2438,
+      "step": 3298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5170117287156213e-05,
+      "loss": 1.0783,
+      "step": 3299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5167198481139825e-05,
+      "loss": 1.2536,
+      "step": 3300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.516427888780383e-05,
+      "loss": 1.019,
+      "step": 3301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.516135850726222e-05,
+      "loss": 1.0738,
+      "step": 3302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.515843733962898e-05,
+      "loss": 1.1841,
+      "step": 3303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.515551538501817e-05,
+      "loss": 1.0969,
+      "step": 3304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.515259264354384e-05,
+      "loss": 1.0885,
+      "step": 3305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5149669115320115e-05,
+      "loss": 1.1545,
+      "step": 3306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5146744800461124e-05,
+      "loss": 1.056,
+      "step": 3307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.514381969908103e-05,
+      "loss": 0.9716,
+      "step": 3308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5140893811294026e-05,
+      "loss": 1.1855,
+      "step": 3309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5137967137214333e-05,
+      "loss": 1.0705,
+      "step": 3310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.513503967695621e-05,
+      "loss": 1.1784,
+      "step": 3311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5132111430633954e-05,
+      "loss": 1.0863,
+      "step": 3312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.512918239836187e-05,
+      "loss": 1.1762,
+      "step": 3313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5126252580254314e-05,
+      "loss": 1.2496,
+      "step": 3314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.512332197642566e-05,
+      "loss": 1.1091,
+      "step": 3315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.512039058699031e-05,
+      "loss": 1.0508,
+      "step": 3316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5117458412062726e-05,
+      "loss": 1.2064,
+      "step": 3317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5114525451757363e-05,
+      "loss": 1.1894,
+      "step": 3318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5111591706188714e-05,
+      "loss": 1.1565,
+      "step": 3319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.510865717547134e-05,
+      "loss": 1.2151,
+      "step": 3320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.510572185971978e-05,
+      "loss": 1.046,
+      "step": 3321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.510278575904864e-05,
+      "loss": 1.2501,
+      "step": 3322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.509984887357253e-05,
+      "loss": 1.3438,
+      "step": 3323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.509691120340612e-05,
+      "loss": 1.2465,
+      "step": 3324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.509397274866409e-05,
+      "loss": 1.2764,
+      "step": 3325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5091033509461155e-05,
+      "loss": 1.2889,
+      "step": 3326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.508809348591206e-05,
+      "loss": 1.3222,
+      "step": 3327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5085152678131584e-05,
+      "loss": 1.3239,
+      "step": 3328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.508221108623454e-05,
+      "loss": 1.3045,
+      "step": 3329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.507926871033576e-05,
+      "loss": 1.3769,
+      "step": 3330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5076325550550114e-05,
+      "loss": 1.2278,
+      "step": 3331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5073381606992506e-05,
+      "loss": 1.3987,
+      "step": 3332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5070436879777865e-05,
+      "loss": 1.3013,
+      "step": 3333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5067491369021156e-05,
+      "loss": 1.3592,
+      "step": 3334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.506454507483735e-05,
+      "loss": 1.2512,
+      "step": 3335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.50615979973415e-05,
+      "loss": 1.3848,
+      "step": 3336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.505865013664863e-05,
+      "loss": 1.3337,
+      "step": 3337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.505570149287385e-05,
+      "loss": 1.4001,
+      "step": 3338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.505275206613225e-05,
+      "loss": 1.2908,
+      "step": 3339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.504980185653899e-05,
+      "loss": 1.3391,
+      "step": 3340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.504685086420924e-05,
+      "loss": 1.3464,
+      "step": 3341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.50438990892582e-05,
+      "loss": 1.2978,
+      "step": 3342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.504094653180112e-05,
+      "loss": 1.3445,
+      "step": 3343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.503799319195326e-05,
+      "loss": 1.3201,
+      "step": 3344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.503503906982991e-05,
+      "loss": 1.3481,
+      "step": 3345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.50320841655464e-05,
+      "loss": 1.3636,
+      "step": 3346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5029128479218094e-05,
+      "loss": 1.3204,
+      "step": 3347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5026172010960385e-05,
+      "loss": 1.3385,
+      "step": 3348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5023214760888685e-05,
+      "loss": 1.3659,
+      "step": 3349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5020256729118446e-05,
+      "loss": 1.3029,
+      "step": 3350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5017297915765136e-05,
+      "loss": 1.3238,
+      "step": 3351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.501433832094429e-05,
+      "loss": 1.2569,
+      "step": 3352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.501137794477143e-05,
+      "loss": 1.2676,
+      "step": 3353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.500841678736214e-05,
+      "loss": 1.3563,
+      "step": 3354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5005454848832014e-05,
+      "loss": 1.385,
+      "step": 3355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.500249212929669e-05,
+      "loss": 1.3564,
+      "step": 3356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.499952862887183e-05,
+      "loss": 1.3615,
+      "step": 3357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.499656434767312e-05,
+      "loss": 1.3006,
+      "step": 3358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.49935992858163e-05,
+      "loss": 1.3305,
+      "step": 3359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.499063344341711e-05,
+      "loss": 1.3238,
+      "step": 3360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.498766682059135e-05,
+      "loss": 1.3272,
+      "step": 3361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.498469941745482e-05,
+      "loss": 1.2594,
+      "step": 3362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4981731234123384e-05,
+      "loss": 1.213,
+      "step": 3363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.49787622707129e-05,
+      "loss": 1.2609,
+      "step": 3364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.497579252733929e-05,
+      "loss": 1.3217,
+      "step": 3365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.497282200411848e-05,
+      "loss": 1.2786,
+      "step": 3366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.496985070116645e-05,
+      "loss": 1.3226,
+      "step": 3367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4966878618599184e-05,
+      "loss": 1.2062,
+      "step": 3368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.496390575653273e-05,
+      "loss": 1.3126,
+      "step": 3369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.496093211508313e-05,
+      "loss": 1.286,
+      "step": 3370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.495795769436648e-05,
+      "loss": 1.148,
+      "step": 3371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.49549824944989e-05,
+      "loss": 1.2245,
+      "step": 3372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.495200651559654e-05,
+      "loss": 1.1983,
+      "step": 3373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.494902975777558e-05,
+      "loss": 1.2438,
+      "step": 3374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.494605222115225e-05,
+      "loss": 1.1986,
+      "step": 3375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4943073905842756e-05,
+      "loss": 1.3314,
+      "step": 3376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.49400948119634e-05,
+      "loss": 1.221,
+      "step": 3377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.493711493963047e-05,
+      "loss": 1.1769,
+      "step": 3378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.493413428896031e-05,
+      "loss": 1.0681,
+      "step": 3379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4931152860069275e-05,
+      "loss": 1.1789,
+      "step": 3380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4928170653073765e-05,
+      "loss": 1.1605,
+      "step": 3381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4925187668090196e-05,
+      "loss": 1.2018,
+      "step": 3382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4922203905235024e-05,
+      "loss": 1.25,
+      "step": 3383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.491921936462473e-05,
+      "loss": 1.2522,
+      "step": 3384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.491623404637585e-05,
+      "loss": 1.2372,
+      "step": 3385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.491324795060491e-05,
+      "loss": 1.1391,
+      "step": 3386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.49102610774285e-05,
+      "loss": 1.1531,
+      "step": 3387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.49072734269632e-05,
+      "loss": 1.2882,
+      "step": 3388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4904284999325676e-05,
+      "loss": 1.3544,
+      "step": 3389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.490129579463258e-05,
+      "loss": 1.3317,
+      "step": 3390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.489830581300062e-05,
+      "loss": 1.3883,
+      "step": 3391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.489531505454651e-05,
+      "loss": 1.414,
+      "step": 3392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.489232351938702e-05,
+      "loss": 1.3422,
+      "step": 3393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.488933120763893e-05,
+      "loss": 1.3469,
+      "step": 3394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4886338119419064e-05,
+      "loss": 1.2935,
+      "step": 3395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.488334425484426e-05,
+      "loss": 1.375,
+      "step": 3396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.488034961403143e-05,
+      "loss": 1.3562,
+      "step": 3397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.487735419709743e-05,
+      "loss": 1.3706,
+      "step": 3398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.487435800415925e-05,
+      "loss": 1.3211,
+      "step": 3399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4871361035333836e-05,
+      "loss": 1.3452,
+      "step": 3400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.486836329073819e-05,
+      "loss": 1.2477,
+      "step": 3401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.486536477048935e-05,
+      "loss": 1.4231,
+      "step": 3402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4862365474704366e-05,
+      "loss": 1.2694,
+      "step": 3403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4859365403500343e-05,
+      "loss": 1.2995,
+      "step": 3404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4856364556994386e-05,
+      "loss": 1.3409,
+      "step": 3405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.485336293530366e-05,
+      "loss": 1.3335,
+      "step": 3406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.485036053854534e-05,
+      "loss": 1.2671,
+      "step": 3407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.484735736683665e-05,
+      "loss": 1.0836,
+      "step": 3408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4844353420294826e-05,
+      "loss": 1.1318,
+      "step": 3409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.484134869903713e-05,
+      "loss": 1.0889,
+      "step": 3410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.483834320318088e-05,
+      "loss": 1.2328,
+      "step": 3411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.483533693284341e-05,
+      "loss": 1.183,
+      "step": 3412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4832329888142065e-05,
+      "loss": 1.2158,
+      "step": 3413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.482932206919426e-05,
+      "loss": 1.1586,
+      "step": 3414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.482631347611741e-05,
+      "loss": 1.1538,
+      "step": 3415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.482330410902896e-05,
+      "loss": 1.1671,
+      "step": 3416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.482029396804642e-05,
+      "loss": 1.1155,
+      "step": 3417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.481728305328728e-05,
+      "loss": 1.0313,
+      "step": 3418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4814271364869094e-05,
+      "loss": 0.991,
+      "step": 3419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4811258902909435e-05,
+      "loss": 0.9585,
+      "step": 3420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4808245667525916e-05,
+      "loss": 1.0044,
+      "step": 3421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4805231658836166e-05,
+      "loss": 0.9918,
+      "step": 3422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.480221687695785e-05,
+      "loss": 1.0631,
+      "step": 3423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.479920132200867e-05,
+      "loss": 1.0853,
+      "step": 3424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.479618499410634e-05,
+      "loss": 0.9573,
+      "step": 3425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.479316789336863e-05,
+      "loss": 0.9821,
+      "step": 3426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4790150019913325e-05,
+      "loss": 1.1785,
+      "step": 3427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.478713137385823e-05,
+      "loss": 1.2932,
+      "step": 3428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.47841119553212e-05,
+      "loss": 1.2209,
+      "step": 3429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.478109176442011e-05,
+      "loss": 1.2095,
+      "step": 3430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.477807080127286e-05,
+      "loss": 1.2252,
+      "step": 3431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.477504906599741e-05,
+      "loss": 1.3235,
+      "step": 3432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.477202655871171e-05,
+      "loss": 1.2716,
+      "step": 3433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4769003279533754e-05,
+      "loss": 1.2855,
+      "step": 3434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.476597922858158e-05,
+      "loss": 1.1952,
+      "step": 3435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.476295440597324e-05,
+      "loss": 1.2563,
+      "step": 3436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4759928811826824e-05,
+      "loss": 1.2744,
+      "step": 3437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.475690244626045e-05,
+      "loss": 1.2002,
+      "step": 3438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4753875309392266e-05,
+      "loss": 1.3636,
+      "step": 3439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.475084740134045e-05,
+      "loss": 1.2625,
+      "step": 3440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.47478187222232e-05,
+      "loss": 1.2429,
+      "step": 3441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.474478927215877e-05,
+      "loss": 1.2694,
+      "step": 3442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.474175905126543e-05,
+      "loss": 1.1927,
+      "step": 3443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.473872805966147e-05,
+      "loss": 1.2094,
+      "step": 3444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.473569629746522e-05,
+      "loss": 1.1941,
+      "step": 3445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.473266376479504e-05,
+      "loss": 1.1234,
+      "step": 3446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.472963046176931e-05,
+      "loss": 1.3123,
+      "step": 3447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.472659638850646e-05,
+      "loss": 1.3061,
+      "step": 3448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4723561545124946e-05,
+      "loss": 1.3554,
+      "step": 3449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.472052593174323e-05,
+      "loss": 1.3378,
+      "step": 3450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.471748954847983e-05,
+      "loss": 1.1777,
+      "step": 3451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4714452395453285e-05,
+      "loss": 1.3444,
+      "step": 3452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.471141447278216e-05,
+      "loss": 1.2479,
+      "step": 3453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4708375780585054e-05,
+      "loss": 1.3395,
+      "step": 3454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4705336318980606e-05,
+      "loss": 1.3834,
+      "step": 3455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.470229608808747e-05,
+      "loss": 1.3195,
+      "step": 3456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.469925508802433e-05,
+      "loss": 1.2915,
+      "step": 3457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.469621331890992e-05,
+      "loss": 1.295,
+      "step": 3458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.469317078086297e-05,
+      "loss": 1.2919,
+      "step": 3459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.469012747400227e-05,
+      "loss": 1.2832,
+      "step": 3460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.468708339844663e-05,
+      "loss": 1.2291,
+      "step": 3461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4684038554314886e-05,
+      "loss": 1.2877,
+      "step": 3462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.468099294172591e-05,
+      "loss": 1.2743,
+      "step": 3463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.46779465607986e-05,
+      "loss": 1.2677,
+      "step": 3464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.467489941165189e-05,
+      "loss": 1.2932,
+      "step": 3465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4671851494404736e-05,
+      "loss": 1.1916,
+      "step": 3466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.466880280917613e-05,
+      "loss": 1.3106,
+      "step": 3467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.466575335608508e-05,
+      "loss": 1.2736,
+      "step": 3468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.466270313525066e-05,
+      "loss": 1.3458,
+      "step": 3469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.465965214679192e-05,
+      "loss": 1.2677,
+      "step": 3470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.465660039082799e-05,
+      "loss": 1.2226,
+      "step": 3471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.465354786747801e-05,
+      "loss": 1.2176,
+      "step": 3472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.465049457686112e-05,
+      "loss": 1.1681,
+      "step": 3473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4647440519096564e-05,
+      "loss": 1.2391,
+      "step": 3474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.464438569430354e-05,
+      "loss": 1.2368,
+      "step": 3475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4641330102601315e-05,
+      "loss": 1.3635,
+      "step": 3476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.463827374410918e-05,
+      "loss": 1.2932,
+      "step": 3477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.463521661894645e-05,
+      "loss": 1.3082,
+      "step": 3478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4632158727232484e-05,
+      "loss": 1.3171,
+      "step": 3479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4629100069086646e-05,
+      "loss": 1.2169,
+      "step": 3480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.462604064462836e-05,
+      "loss": 1.3923,
+      "step": 3481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4622980453977054e-05,
+      "loss": 1.2865,
+      "step": 3482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.46199194972522e-05,
+      "loss": 1.2334,
+      "step": 3483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4616857774573296e-05,
+      "loss": 1.2973,
+      "step": 3484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.461379528605988e-05,
+      "loss": 1.3229,
+      "step": 3485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4610732031831495e-05,
+      "loss": 1.2172,
+      "step": 3486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.460766801200773e-05,
+      "loss": 1.194,
+      "step": 3487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.460460322670821e-05,
+      "loss": 1.1809,
+      "step": 3488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.460153767605258e-05,
+      "loss": 1.1997,
+      "step": 3489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4598471360160534e-05,
+      "loss": 1.169,
+      "step": 3490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.459540427915176e-05,
+      "loss": 1.1294,
+      "step": 3491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4592336433146e-05,
+      "loss": 1.2496,
+      "step": 3492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.458926782226302e-05,
+      "loss": 1.2096,
+      "step": 3493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.458619844662262e-05,
+      "loss": 1.2607,
+      "step": 3494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.458312830634464e-05,
+      "loss": 1.2662,
+      "step": 3495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4580057401548904e-05,
+      "loss": 1.074,
+      "step": 3496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.457698573235533e-05,
+      "loss": 1.3738,
+      "step": 3497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.457391329888383e-05,
+      "loss": 1.2837,
+      "step": 3498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.457084010125434e-05,
+      "loss": 1.319,
+      "step": 3499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.456776613958683e-05,
+      "loss": 1.2622,
+      "step": 3500
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.17086660861969,
+      "eval_runtime": 116.4609,
+      "eval_samples_per_second": 32.809,
+      "eval_steps_per_second": 0.343,
+      "step": 3500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.456469141400134e-05,
+      "loss": 1.2965,
+      "step": 3501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.456161592461786e-05,
+      "loss": 1.3026,
+      "step": 3502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4558539671556496e-05,
+      "loss": 1.4932,
+      "step": 3503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.455546265493732e-05,
+      "loss": 1.2582,
+      "step": 3504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.455238487488047e-05,
+      "loss": 1.2925,
+      "step": 3505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.454930633150609e-05,
+      "loss": 1.2764,
+      "step": 3506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.454622702493438e-05,
+      "loss": 1.1789,
+      "step": 3507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.454314695528554e-05,
+      "loss": 1.1541,
+      "step": 3508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4540066122679825e-05,
+      "loss": 1.1981,
+      "step": 3509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.45369845272375e-05,
+      "loss": 1.2131,
+      "step": 3510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.453390216907888e-05,
+      "loss": 1.258,
+      "step": 3511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.453081904832429e-05,
+      "loss": 1.1614,
+      "step": 3512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.45277351650941e-05,
+      "loss": 1.1982,
+      "step": 3513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4524650519508706e-05,
+      "loss": 1.0871,
+      "step": 3514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.452156511168851e-05,
+      "loss": 1.1596,
+      "step": 3515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4518478941753996e-05,
+      "loss": 1.2278,
+      "step": 3516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.451539200982563e-05,
+      "loss": 1.3055,
+      "step": 3517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4512304316023935e-05,
+      "loss": 1.0933,
+      "step": 3518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4509215860469436e-05,
+      "loss": 1.1041,
+      "step": 3519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.450612664328271e-05,
+      "loss": 1.2664,
+      "step": 3520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.450303666458438e-05,
+      "loss": 1.0958,
+      "step": 3521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4499945924495046e-05,
+      "loss": 1.1056,
+      "step": 3522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4496854423135395e-05,
+      "loss": 1.2071,
+      "step": 3523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.44937621606261e-05,
+      "loss": 1.1994,
+      "step": 3524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.449066913708789e-05,
+      "loss": 1.1706,
+      "step": 3525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.448757535264152e-05,
+      "loss": 1.0078,
+      "step": 3526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.448448080740776e-05,
+      "loss": 1.0021,
+      "step": 3527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.448138550150742e-05,
+      "loss": 1.1225,
+      "step": 3528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.447828943506136e-05,
+      "loss": 1.1007,
+      "step": 3529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.447519260819042e-05,
+      "loss": 1.0605,
+      "step": 3530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4472095021015515e-05,
+      "loss": 1.0645,
+      "step": 3531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.446899667365757e-05,
+      "loss": 1.0431,
+      "step": 3532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.446589756623754e-05,
+      "loss": 1.0523,
+      "step": 3533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.446279769887644e-05,
+      "loss": 1.0813,
+      "step": 3534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4459697071695236e-05,
+      "loss": 1.2215,
+      "step": 3535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.445659568481502e-05,
+      "loss": 1.1938,
+      "step": 3536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4453493538356846e-05,
+      "loss": 1.1758,
+      "step": 3537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.445039063244184e-05,
+      "loss": 1.2499,
+      "step": 3538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4447286967191116e-05,
+      "loss": 1.2537,
+      "step": 3539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.444418254272585e-05,
+      "loss": 1.2143,
+      "step": 3540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.444107735916724e-05,
+      "loss": 1.1832,
+      "step": 3541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4437971416636504e-05,
+      "loss": 1.2483,
+      "step": 3542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.443486471525491e-05,
+      "loss": 1.194,
+      "step": 3543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4431757255143725e-05,
+      "loss": 1.2018,
+      "step": 3544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.442864903642428e-05,
+      "loss": 1.3458,
+      "step": 3545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4425540059217896e-05,
+      "loss": 1.3469,
+      "step": 3546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.442243032364597e-05,
+      "loss": 1.3625,
+      "step": 3547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.44193198298299e-05,
+      "loss": 1.3051,
+      "step": 3548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.441620857789111e-05,
+      "loss": 1.2853,
+      "step": 3549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.441309656795106e-05,
+      "loss": 1.2382,
+      "step": 3550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.440998380013125e-05,
+      "loss": 1.3225,
+      "step": 3551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.440687027455321e-05,
+      "loss": 1.3203,
+      "step": 3552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.440375599133847e-05,
+      "loss": 1.2633,
+      "step": 3553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.440064095060862e-05,
+      "loss": 1.1994,
+      "step": 3554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4397525152485275e-05,
+      "loss": 1.1797,
+      "step": 3555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4394408597090064e-05,
+      "loss": 1.1955,
+      "step": 3556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4391291284544664e-05,
+      "loss": 1.1368,
+      "step": 3557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.438817321497077e-05,
+      "loss": 1.2212,
+      "step": 3558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.438505438849012e-05,
+      "loss": 1.127,
+      "step": 3559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4381934805224454e-05,
+      "loss": 1.2542,
+      "step": 3560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4378814465295575e-05,
+      "loss": 1.2453,
+      "step": 3561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.437569336882529e-05,
+      "loss": 1.2132,
+      "step": 3562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.437257151593545e-05,
+      "loss": 1.1421,
+      "step": 3563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.436944890674794e-05,
+      "loss": 1.1912,
+      "step": 3564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4366325541384636e-05,
+      "loss": 1.1634,
+      "step": 3565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4363201419967506e-05,
+      "loss": 1.2873,
+      "step": 3566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.43600765426185e-05,
+      "loss": 1.2158,
+      "step": 3567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4356950909459606e-05,
+      "loss": 1.2123,
+      "step": 3568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.435382452061286e-05,
+      "loss": 1.2546,
+      "step": 3569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.43506973762003e-05,
+      "loss": 1.1959,
+      "step": 3570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.434756947634403e-05,
+      "loss": 1.2694,
+      "step": 3571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4344440821166145e-05,
+      "loss": 1.2788,
+      "step": 3572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.434131141078879e-05,
+      "loss": 1.2112,
+      "step": 3573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4338181245334145e-05,
+      "loss": 1.2217,
+      "step": 3574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.43350503249244e-05,
+      "loss": 1.1163,
+      "step": 3575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.433191864968178e-05,
+      "loss": 1.2341,
+      "step": 3576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4328786219728556e-05,
+      "loss": 1.2526,
+      "step": 3577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.432565303518702e-05,
+      "loss": 1.2188,
+      "step": 3578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.432251909617948e-05,
+      "loss": 1.1946,
+      "step": 3579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.431938440282828e-05,
+      "loss": 1.2195,
+      "step": 3580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.431624895525581e-05,
+      "loss": 1.1481,
+      "step": 3581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.431311275358447e-05,
+      "loss": 0.9496,
+      "step": 3582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.43099757979367e-05,
+      "loss": 0.9778,
+      "step": 3583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.430683808843496e-05,
+      "loss": 0.929,
+      "step": 3584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4303699625201745e-05,
+      "loss": 0.9305,
+      "step": 3585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.430056040835959e-05,
+      "loss": 0.9387,
+      "step": 3586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4297420438031036e-05,
+      "loss": 1.02,
+      "step": 3587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4294279714338666e-05,
+      "loss": 1.0389,
+      "step": 3588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.42911382374051e-05,
+      "loss": 0.9252,
+      "step": 3589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.428799600735298e-05,
+      "loss": 1.0546,
+      "step": 3590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.428485302430497e-05,
+      "loss": 1.0173,
+      "step": 3591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.428170928838379e-05,
+      "loss": 1.3636,
+      "step": 3592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.427856479971214e-05,
+      "loss": 1.3054,
+      "step": 3593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4275419558412805e-05,
+      "loss": 1.2573,
+      "step": 3594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.427227356460856e-05,
+      "loss": 1.3282,
+      "step": 3595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.426912681842224e-05,
+      "loss": 1.2995,
+      "step": 3596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4265979319976666e-05,
+      "loss": 1.3632,
+      "step": 3597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.426283106939474e-05,
+      "loss": 1.2173,
+      "step": 3598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4259682066799346e-05,
+      "loss": 1.3275,
+      "step": 3599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.425653231231344e-05,
+      "loss": 1.3283,
+      "step": 3600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.425338180605998e-05,
+      "loss": 1.2684,
+      "step": 3601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.425023054816197e-05,
+      "loss": 1.2817,
+      "step": 3602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.42470785387424e-05,
+      "loss": 1.3409,
+      "step": 3603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.424392577792437e-05,
+      "loss": 1.3134,
+      "step": 3604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4240772265830925e-05,
+      "loss": 1.2355,
+      "step": 3605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.42376180025852e-05,
+      "loss": 1.2353,
+      "step": 3606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.423446298831032e-05,
+      "loss": 1.2529,
+      "step": 3607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.423130722312947e-05,
+      "loss": 1.3507,
+      "step": 3608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.422815070716585e-05,
+      "loss": 1.3648,
+      "step": 3609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4224993440542675e-05,
+      "loss": 1.1915,
+      "step": 3610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4221835423383205e-05,
+      "loss": 1.1575,
+      "step": 3611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4218676655810745e-05,
+      "loss": 1.142,
+      "step": 3612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.421551713794859e-05,
+      "loss": 1.1609,
+      "step": 3613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.42123568699201e-05,
+      "loss": 1.064,
+      "step": 3614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.420919585184865e-05,
+      "loss": 1.196,
+      "step": 3615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.420603408385764e-05,
+      "loss": 1.074,
+      "step": 3616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.420287156607051e-05,
+      "loss": 1.0777,
+      "step": 3617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.419970829861072e-05,
+      "loss": 1.1521,
+      "step": 3618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.419654428160176e-05,
+      "loss": 1.1623,
+      "step": 3619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.419337951516715e-05,
+      "loss": 1.0132,
+      "step": 3620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.419021399943046e-05,
+      "loss": 1.1339,
+      "step": 3621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.418704773451525e-05,
+      "loss": 1.108,
+      "step": 3622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.418388072054513e-05,
+      "loss": 1.1752,
+      "step": 3623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.418071295764376e-05,
+      "loss": 0.9265,
+      "step": 3624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.417754444593478e-05,
+      "loss": 1.1776,
+      "step": 3625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.417437518554192e-05,
+      "loss": 1.0438,
+      "step": 3626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.417120517658887e-05,
+      "loss": 0.9993,
+      "step": 3627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4168034419199405e-05,
+      "loss": 1.0865,
+      "step": 3628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4164862913497316e-05,
+      "loss": 1.0928,
+      "step": 3629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.41616906596064e-05,
+      "loss": 1.3044,
+      "step": 3630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.415851765765052e-05,
+      "loss": 1.3579,
+      "step": 3631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4155343907753546e-05,
+      "loss": 1.313,
+      "step": 3632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.415216941003936e-05,
+      "loss": 1.2484,
+      "step": 3633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4148994164631916e-05,
+      "loss": 1.2451,
+      "step": 3634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.414581817165516e-05,
+      "loss": 1.2789,
+      "step": 3635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.414264143123309e-05,
+      "loss": 1.4335,
+      "step": 3636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.413946394348971e-05,
+      "loss": 1.3323,
+      "step": 3637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4136285708549097e-05,
+      "loss": 1.3772,
+      "step": 3638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.41331067265353e-05,
+      "loss": 1.1931,
+      "step": 3639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.412992699757244e-05,
+      "loss": 1.0602,
+      "step": 3640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.412674652178465e-05,
+      "loss": 1.1777,
+      "step": 3641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.412356529929608e-05,
+      "loss": 1.1745,
+      "step": 3642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.412038333023094e-05,
+      "loss": 1.2416,
+      "step": 3643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.411720061471346e-05,
+      "loss": 1.1395,
+      "step": 3644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.411401715286787e-05,
+      "loss": 1.2041,
+      "step": 3645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.411083294481847e-05,
+      "loss": 1.193,
+      "step": 3646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.410764799068955e-05,
+      "loss": 1.0567,
+      "step": 3647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.410446229060548e-05,
+      "loss": 1.212,
+      "step": 3648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.410127584469059e-05,
+      "loss": 1.3339,
+      "step": 3649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.409808865306931e-05,
+      "loss": 1.2856,
+      "step": 3650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4094900715866064e-05,
+      "loss": 1.373,
+      "step": 3651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.409171203320528e-05,
+      "loss": 1.3396,
+      "step": 3652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.408852260521147e-05,
+      "loss": 1.2819,
+      "step": 3653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.408533243200914e-05,
+      "loss": 1.3054,
+      "step": 3654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.408214151372283e-05,
+      "loss": 1.3225,
+      "step": 3655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4078949850477115e-05,
+      "loss": 1.415,
+      "step": 3656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.407575744239659e-05,
+      "loss": 1.3009,
+      "step": 3657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.40725642896059e-05,
+      "loss": 1.275,
+      "step": 3658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.406937039222969e-05,
+      "loss": 1.2185,
+      "step": 3659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4066175750392654e-05,
+      "loss": 1.3029,
+      "step": 3660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.406298036421952e-05,
+      "loss": 1.2267,
+      "step": 3661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.405978423383502e-05,
+      "loss": 1.264,
+      "step": 3662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.405658735936393e-05,
+      "loss": 1.2369,
+      "step": 3663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.405338974093106e-05,
+      "loss": 1.2645,
+      "step": 3664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.405019137866125e-05,
+      "loss": 1.205,
+      "step": 3665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.404699227267935e-05,
+      "loss": 1.31,
+      "step": 3666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4043792423110255e-05,
+      "loss": 1.156,
+      "step": 3667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4040591830078894e-05,
+      "loss": 0.9589,
+      "step": 3668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.403739049371021e-05,
+      "loss": 0.9247,
+      "step": 3669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.403418841412917e-05,
+      "loss": 0.9377,
+      "step": 3670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.403098559146081e-05,
+      "loss": 0.9853,
+      "step": 3671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4027782025830144e-05,
+      "loss": 0.9878,
+      "step": 3672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.402457771736226e-05,
+      "loss": 1.0294,
+      "step": 3673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4021372666182226e-05,
+      "loss": 1.0354,
+      "step": 3674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4018166872415176e-05,
+      "loss": 0.9736,
+      "step": 3675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4014960336186274e-05,
+      "loss": 0.9643,
+      "step": 3676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.40117530576207e-05,
+      "loss": 0.9921,
+      "step": 3677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.400854503684364e-05,
+      "loss": 1.701,
+      "step": 3678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.400533627398037e-05,
+      "loss": 2.0553,
+      "step": 3679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.400212676915613e-05,
+      "loss": 2.4389,
+      "step": 3680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.399891652249624e-05,
+      "loss": 2.1204,
+      "step": 3681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.399570553412601e-05,
+      "loss": 2.1076,
+      "step": 3682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.39924938041708e-05,
+      "loss": 2.1878,
+      "step": 3683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3989281332755996e-05,
+      "loss": 2.618,
+      "step": 3684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.398606812000702e-05,
+      "loss": 1.928,
+      "step": 3685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.39828541660493e-05,
+      "loss": 2.0937,
+      "step": 3686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.397963947100831e-05,
+      "loss": 2.3618,
+      "step": 3687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.397642403500956e-05,
+      "loss": 2.2407,
+      "step": 3688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.397320785817858e-05,
+      "loss": 1.4737,
+      "step": 3689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3969990940640917e-05,
+      "loss": 1.2937,
+      "step": 3690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.396677328252216e-05,
+      "loss": 1.2888,
+      "step": 3691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.396355488394793e-05,
+      "loss": 1.3109,
+      "step": 3692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3960335745043874e-05,
+      "loss": 1.2599,
+      "step": 3693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.395711586593566e-05,
+      "loss": 1.3437,
+      "step": 3694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.395389524674899e-05,
+      "loss": 1.3127,
+      "step": 3695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.395067388760961e-05,
+      "loss": 1.3804,
+      "step": 3696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3947451788643265e-05,
+      "loss": 1.1991,
+      "step": 3697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.394422894997575e-05,
+      "loss": 1.2371,
+      "step": 3698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.394100537173288e-05,
+      "loss": 1.1414,
+      "step": 3699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3937781054040505e-05,
+      "loss": 1.1124,
+      "step": 3700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.39345559970245e-05,
+      "loss": 1.2172,
+      "step": 3701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.393133020081077e-05,
+      "loss": 1.0979,
+      "step": 3702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.392810366552526e-05,
+      "loss": 1.1434,
+      "step": 3703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3924876391293915e-05,
+      "loss": 1.1774,
+      "step": 3704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.392164837824273e-05,
+      "loss": 1.1347,
+      "step": 3705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3918419626497736e-05,
+      "loss": 1.2374,
+      "step": 3706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.391519013618498e-05,
+      "loss": 1.192,
+      "step": 3707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.391195990743052e-05,
+      "loss": 1.317,
+      "step": 3708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.390872894036049e-05,
+      "loss": 1.2988,
+      "step": 3709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3905497235101015e-05,
+      "loss": 1.4047,
+      "step": 3710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.390226479177826e-05,
+      "loss": 1.4358,
+      "step": 3711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.389903161051841e-05,
+      "loss": 1.3743,
+      "step": 3712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.38957976914477e-05,
+      "loss": 1.425,
+      "step": 3713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.389256303469237e-05,
+      "loss": 1.3596,
+      "step": 3714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.388932764037871e-05,
+      "loss": 1.4144,
+      "step": 3715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.388609150863302e-05,
+      "loss": 1.4398,
+      "step": 3716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3882854639581636e-05,
+      "loss": 1.2948,
+      "step": 3717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.387961703335094e-05,
+      "loss": 1.2703,
+      "step": 3718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.387637869006732e-05,
+      "loss": 1.2102,
+      "step": 3719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.387313960985718e-05,
+      "loss": 1.2664,
+      "step": 3720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3869899792847e-05,
+      "loss": 1.2444,
+      "step": 3721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.386665923916324e-05,
+      "loss": 1.1723,
+      "step": 3722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3863417948932424e-05,
+      "loss": 1.2628,
+      "step": 3723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3860175922281087e-05,
+      "loss": 1.2633,
+      "step": 3724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.385693315933579e-05,
+      "loss": 1.1381,
+      "step": 3725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.385368966022314e-05,
+      "loss": 1.226,
+      "step": 3726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3850445425069755e-05,
+      "loss": 1.1525,
+      "step": 3727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.384720045400229e-05,
+      "loss": 1.0978,
+      "step": 3728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3843954747147424e-05,
+      "loss": 1.0953,
+      "step": 3729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.384070830463187e-05,
+      "loss": 1.1421,
+      "step": 3730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.383746112658238e-05,
+      "loss": 1.0995,
+      "step": 3731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.383421321312571e-05,
+      "loss": 1.0706,
+      "step": 3732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.383096456438865e-05,
+      "loss": 1.1033,
+      "step": 3733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.382771518049805e-05,
+      "loss": 1.0539,
+      "step": 3734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.382446506158073e-05,
+      "loss": 1.0296,
+      "step": 3735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3821214207763614e-05,
+      "loss": 1.0944,
+      "step": 3736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.381796261917358e-05,
+      "loss": 0.9806,
+      "step": 3737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.381471029593759e-05,
+      "loss": 1.1288,
+      "step": 3738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.38114572381826e-05,
+      "loss": 1.2393,
+      "step": 3739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.380820344603562e-05,
+      "loss": 1.2621,
+      "step": 3740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.380494891962367e-05,
+      "loss": 1.1771,
+      "step": 3741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3801693659073794e-05,
+      "loss": 1.1561,
+      "step": 3742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.37984376645131e-05,
+      "loss": 1.1367,
+      "step": 3743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.379518093606869e-05,
+      "loss": 1.1947,
+      "step": 3744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.37919234738677e-05,
+      "loss": 1.1891,
+      "step": 3745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3788665278037305e-05,
+      "loss": 1.1625,
+      "step": 3746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.378540634870471e-05,
+      "loss": 1.2449,
+      "step": 3747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3782146685997125e-05,
+      "loss": 1.3025,
+      "step": 3748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.377888629004182e-05,
+      "loss": 1.2435,
+      "step": 3749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.377562516096607e-05,
+      "loss": 1.3244,
+      "step": 3750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.37723632988972e-05,
+      "loss": 1.2549,
+      "step": 3751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.376910070396255e-05,
+      "loss": 1.2409,
+      "step": 3752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.376583737628949e-05,
+      "loss": 1.3804,
+      "step": 3753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.376257331600541e-05,
+      "loss": 1.3315,
+      "step": 3754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.375930852323773e-05,
+      "loss": 1.2729,
+      "step": 3755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.375604299811393e-05,
+      "loss": 1.181,
+      "step": 3756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.375277674076149e-05,
+      "loss": 1.042,
+      "step": 3757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3749509751307914e-05,
+      "loss": 1.1237,
+      "step": 3758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.374624202988075e-05,
+      "loss": 1.0385,
+      "step": 3759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.374297357660756e-05,
+      "loss": 1.1894,
+      "step": 3760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3739704391615954e-05,
+      "loss": 0.921,
+      "step": 3761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.373643447503355e-05,
+      "loss": 1.1007,
+      "step": 3762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3733163826988016e-05,
+      "loss": 0.9408,
+      "step": 3763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.372989244760703e-05,
+      "loss": 1.1184,
+      "step": 3764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3726620337018306e-05,
+      "loss": 1.0882,
+      "step": 3765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.372334749534958e-05,
+      "loss": 0.9853,
+      "step": 3766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.372007392272863e-05,
+      "loss": 1.2345,
+      "step": 3767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3716799619283255e-05,
+      "loss": 1.2622,
+      "step": 3768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.371352458514128e-05,
+      "loss": 1.3049,
+      "step": 3769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.371024882043057e-05,
+      "loss": 1.1732,
+      "step": 3770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3706972325278985e-05,
+      "loss": 1.1931,
+      "step": 3771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3703695099814463e-05,
+      "loss": 1.0668,
+      "step": 3772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.370041714416493e-05,
+      "loss": 1.1711,
+      "step": 3773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3697138458458374e-05,
+      "loss": 1.2894,
+      "step": 3774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3693859042822774e-05,
+      "loss": 1.198,
+      "step": 3775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.369057889738617e-05,
+      "loss": 1.1702,
+      "step": 3776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.368729802227661e-05,
+      "loss": 1.2465,
+      "step": 3777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.368401641762219e-05,
+      "loss": 1.2856,
+      "step": 3778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3680734083551015e-05,
+      "loss": 1.2069,
+      "step": 3779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3677451020191215e-05,
+      "loss": 1.2552,
+      "step": 3780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.367416722767097e-05,
+      "loss": 1.2118,
+      "step": 3781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.367088270611848e-05,
+      "loss": 1.197,
+      "step": 3782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.366759745566197e-05,
+      "loss": 1.2001,
+      "step": 3783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.36643114764297e-05,
+      "loss": 1.2425,
+      "step": 3784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.366102476854994e-05,
+      "loss": 1.2357,
+      "step": 3785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3657737332151e-05,
+      "loss": 1.009,
+      "step": 3786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.365444916736124e-05,
+      "loss": 1.0793,
+      "step": 3787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.365116027430902e-05,
+      "loss": 1.1344,
+      "step": 3788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3647870653122735e-05,
+      "loss": 1.0013,
+      "step": 3789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.364458030393081e-05,
+      "loss": 1.0571,
+      "step": 3790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.36412892268617e-05,
+      "loss": 1.0875,
+      "step": 3791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3637997422043886e-05,
+      "loss": 1.1084,
+      "step": 3792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.363470488960588e-05,
+      "loss": 1.141,
+      "step": 3793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.363141162967622e-05,
+      "loss": 1.1699,
+      "step": 3794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.362811764238347e-05,
+      "loss": 1.0964,
+      "step": 3795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.362482292785624e-05,
+      "loss": 1.1647,
+      "step": 3796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.362152748622313e-05,
+      "loss": 1.2091,
+      "step": 3797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3618231317612825e-05,
+      "loss": 1.1467,
+      "step": 3798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3614934422153985e-05,
+      "loss": 1.1465,
+      "step": 3799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.361163679997532e-05,
+      "loss": 1.1078,
+      "step": 3800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.360833845120557e-05,
+      "loss": 1.134,
+      "step": 3801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3605039375973504e-05,
+      "loss": 1.1887,
+      "step": 3802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.360173957440792e-05,
+      "loss": 1.2878,
+      "step": 3803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3598439046637626e-05,
+      "loss": 1.1178,
+      "step": 3804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.35951377927915e-05,
+      "loss": 1.1739,
+      "step": 3805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3591835812998385e-05,
+      "loss": 1.1379,
+      "step": 3806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.358853310738722e-05,
+      "loss": 1.1566,
+      "step": 3807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.358522967608692e-05,
+      "loss": 1.0447,
+      "step": 3808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.358192551922647e-05,
+      "loss": 1.2178,
+      "step": 3809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.357862063693486e-05,
+      "loss": 1.1113,
+      "step": 3810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3575315029341096e-05,
+      "loss": 1.1544,
+      "step": 3811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.357200869657423e-05,
+      "loss": 1.0852,
+      "step": 3812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.356870163876336e-05,
+      "loss": 1.12,
+      "step": 3813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3565393856037574e-05,
+      "loss": 1.2287,
+      "step": 3814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3562085348526004e-05,
+      "loss": 1.2061,
+      "step": 3815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3558776116357825e-05,
+      "loss": 1.1957,
+      "step": 3816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3555466159662214e-05,
+      "loss": 1.2049,
+      "step": 3817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3552155478568414e-05,
+      "loss": 1.1948,
+      "step": 3818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3548844073205655e-05,
+      "loss": 1.1699,
+      "step": 3819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.354553194370321e-05,
+      "loss": 1.1983,
+      "step": 3820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3542219090190384e-05,
+      "loss": 1.0448,
+      "step": 3821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.353890551279652e-05,
+      "loss": 1.2016,
+      "step": 3822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.353559121165098e-05,
+      "loss": 1.1804,
+      "step": 3823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.353227618688314e-05,
+      "loss": 1.0941,
+      "step": 3824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3528960438622425e-05,
+      "loss": 1.0492,
+      "step": 3825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3525643966998284e-05,
+      "loss": 1.0239,
+      "step": 3826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3522326772140184e-05,
+      "loss": 1.1116,
+      "step": 3827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3519008854177626e-05,
+      "loss": 1.1731,
+      "step": 3828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3515690213240145e-05,
+      "loss": 1.1023,
+      "step": 3829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3512370849457296e-05,
+      "loss": 1.1031,
+      "step": 3830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.350905076295867e-05,
+      "loss": 1.1475,
+      "step": 3831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3505729953873873e-05,
+      "loss": 1.1571,
+      "step": 3832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3502408422332566e-05,
+      "loss": 1.0863,
+      "step": 3833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.34990861684644e-05,
+      "loss": 1.0668,
+      "step": 3834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3495763192399085e-05,
+      "loss": 1.3622,
+      "step": 3835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3492439494266355e-05,
+      "loss": 1.423,
+      "step": 3836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3489115074195944e-05,
+      "loss": 1.361,
+      "step": 3837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.348578993231766e-05,
+      "loss": 1.3978,
+      "step": 3838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3482464068761305e-05,
+      "loss": 1.3682,
+      "step": 3839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.347913748365671e-05,
+      "loss": 1.3049,
+      "step": 3840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3475810177133755e-05,
+      "loss": 1.4535,
+      "step": 3841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3472482149322346e-05,
+      "loss": 1.3893,
+      "step": 3842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.346915340035239e-05,
+      "loss": 1.3577,
+      "step": 3843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3465823930353846e-05,
+      "loss": 1.3098,
+      "step": 3844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.34624937394567e-05,
+      "loss": 1.2232,
+      "step": 3845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3459162827790956e-05,
+      "loss": 1.1364,
+      "step": 3846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.345583119548665e-05,
+      "loss": 1.1787,
+      "step": 3847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.345249884267386e-05,
+      "loss": 1.1347,
+      "step": 3848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.344916576948266e-05,
+      "loss": 1.0754,
+      "step": 3849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.344583197604318e-05,
+      "loss": 1.1597,
+      "step": 3850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.344249746248559e-05,
+      "loss": 1.118,
+      "step": 3851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3439162228940034e-05,
+      "loss": 1.2869,
+      "step": 3852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.343582627553674e-05,
+      "loss": 1.1245,
+      "step": 3853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3432489602405934e-05,
+      "loss": 1.2501,
+      "step": 3854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.342915220967788e-05,
+      "loss": 1.1533,
+      "step": 3855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.342581409748287e-05,
+      "loss": 1.1499,
+      "step": 3856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.342247526595123e-05,
+      "loss": 1.1341,
+      "step": 3857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.34191357152133e-05,
+      "loss": 1.1783,
+      "step": 3858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3415795445399446e-05,
+      "loss": 1.1643,
+      "step": 3859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.341245445664008e-05,
+      "loss": 1.1476,
+      "step": 3860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.340911274906564e-05,
+      "loss": 1.2178,
+      "step": 3861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.340577032280657e-05,
+      "loss": 1.0843,
+      "step": 3862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3402427177993366e-05,
+      "loss": 1.1141,
+      "step": 3863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.339908331475653e-05,
+      "loss": 1.1014,
+      "step": 3864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.339573873322663e-05,
+      "loss": 1.2609,
+      "step": 3865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.339239343353422e-05,
+      "loss": 1.1562,
+      "step": 3866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.33890474158099e-05,
+      "loss": 1.285,
+      "step": 3867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3385700680184296e-05,
+      "loss": 1.2552,
+      "step": 3868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3382353226788075e-05,
+      "loss": 1.3112,
+      "step": 3869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3379005055751905e-05,
+      "loss": 1.2276,
+      "step": 3870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3375656167206504e-05,
+      "loss": 1.209,
+      "step": 3871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3372306561282616e-05,
+      "loss": 1.3186,
+      "step": 3872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3368956238111e-05,
+      "loss": 1.3133,
+      "step": 3873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.336560519782246e-05,
+      "loss": 1.1082,
+      "step": 3874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.336225344054782e-05,
+      "loss": 1.106,
+      "step": 3875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3358900966417914e-05,
+      "loss": 1.1311,
+      "step": 3876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.335554777556363e-05,
+      "loss": 1.0949,
+      "step": 3877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.335219386811589e-05,
+      "loss": 1.1934,
+      "step": 3878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.334883924420562e-05,
+      "loss": 1.1084,
+      "step": 3879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.334548390396377e-05,
+      "loss": 1.1731,
+      "step": 3880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.334212784752135e-05,
+      "loss": 1.2137,
+      "step": 3881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.333877107500937e-05,
+      "loss": 1.1345,
+      "step": 3882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.333541358655888e-05,
+      "loss": 1.0197,
+      "step": 3883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.333205538230094e-05,
+      "loss": 1.0858,
+      "step": 3884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.332869646236668e-05,
+      "loss": 1.2026,
+      "step": 3885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.332533682688721e-05,
+      "loss": 1.1114,
+      "step": 3886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3321976475993706e-05,
+      "loss": 1.1308,
+      "step": 3887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3318615409817345e-05,
+      "loss": 1.182,
+      "step": 3888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3315253628489326e-05,
+      "loss": 1.2436,
+      "step": 3889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.331189113214092e-05,
+      "loss": 1.203,
+      "step": 3890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.330852792090338e-05,
+      "loss": 1.126,
+      "step": 3891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.330516399490802e-05,
+      "loss": 1.2304,
+      "step": 3892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.330179935428614e-05,
+      "loss": 1.208,
+      "step": 3893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.329843399916912e-05,
+      "loss": 1.1383,
+      "step": 3894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.329506792968833e-05,
+      "loss": 1.2506,
+      "step": 3895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3291701145975185e-05,
+      "loss": 1.1407,
+      "step": 3896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.328833364816112e-05,
+      "loss": 1.1502,
+      "step": 3897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.328496543637759e-05,
+      "loss": 1.2658,
+      "step": 3898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.328159651075611e-05,
+      "loss": 1.1738,
+      "step": 3899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.327822687142819e-05,
+      "loss": 1.279,
+      "step": 3900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3274856518525375e-05,
+      "loss": 1.2442,
+      "step": 3901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.327148545217925e-05,
+      "loss": 1.2279,
+      "step": 3902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.326811367252143e-05,
+      "loss": 1.2717,
+      "step": 3903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.326474117968352e-05,
+      "loss": 1.3335,
+      "step": 3904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.326136797379721e-05,
+      "loss": 1.2777,
+      "step": 3905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3257994054994165e-05,
+      "loss": 1.2646,
+      "step": 3906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.325461942340611e-05,
+      "loss": 1.1928,
+      "step": 3907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.32512440791648e-05,
+      "loss": 1.3188,
+      "step": 3908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.324786802240199e-05,
+      "loss": 1.2293,
+      "step": 3909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.32444912532495e-05,
+      "loss": 1.2496,
+      "step": 3910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.324111377183914e-05,
+      "loss": 1.1839,
+      "step": 3911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.323773557830276e-05,
+      "loss": 1.2062,
+      "step": 3912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.323435667277227e-05,
+      "loss": 1.2503,
+      "step": 3913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.323097705537956e-05,
+      "loss": 1.2408,
+      "step": 3914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.322759672625658e-05,
+      "loss": 1.2374,
+      "step": 3915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3224215685535294e-05,
+      "loss": 1.2105,
+      "step": 3916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.322083393334768e-05,
+      "loss": 1.1792,
+      "step": 3917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.321745146982579e-05,
+      "loss": 1.1635,
+      "step": 3918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.321406829510165e-05,
+      "loss": 1.2695,
+      "step": 3919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.321068440930735e-05,
+      "loss": 1.205,
+      "step": 3920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.320729981257499e-05,
+      "loss": 1.1674,
+      "step": 3921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3203914505036706e-05,
+      "loss": 1.1421,
+      "step": 3922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.320052848682466e-05,
+      "loss": 1.1981,
+      "step": 3923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.319714175807104e-05,
+      "loss": 1.2081,
+      "step": 3924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.319375431890806e-05,
+      "loss": 1.1411,
+      "step": 3925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3190366169467975e-05,
+      "loss": 1.1032,
+      "step": 3926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3186977309883035e-05,
+      "loss": 1.185,
+      "step": 3927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.318358774028556e-05,
+      "loss": 1.1629,
+      "step": 3928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.318019746080787e-05,
+      "loss": 1.1666,
+      "step": 3929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.317680647158232e-05,
+      "loss": 1.1949,
+      "step": 3930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3173414772741294e-05,
+      "loss": 1.2109,
+      "step": 3931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.31700223644172e-05,
+      "loss": 1.3182,
+      "step": 3932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.316662924674248e-05,
+      "loss": 1.2893,
+      "step": 3933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.31632354198496e-05,
+      "loss": 1.3865,
+      "step": 3934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.315984088387105e-05,
+      "loss": 1.353,
+      "step": 3935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3156445638939357e-05,
+      "loss": 1.296,
+      "step": 3936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.315304968518706e-05,
+      "loss": 1.3449,
+      "step": 3937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.314965302274675e-05,
+      "loss": 1.3042,
+      "step": 3938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.314625565175102e-05,
+      "loss": 1.3189,
+      "step": 3939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3142857572332504e-05,
+      "loss": 1.2612,
+      "step": 3940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.313945878462388e-05,
+      "loss": 1.1383,
+      "step": 3941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.31360592887578e-05,
+      "loss": 1.2518,
+      "step": 3942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3132659084867e-05,
+      "loss": 1.2207,
+      "step": 3943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.312925817308422e-05,
+      "loss": 1.2303,
+      "step": 3944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.312585655354224e-05,
+      "loss": 1.2092,
+      "step": 3945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3122454226373845e-05,
+      "loss": 1.3165,
+      "step": 3946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.311905119171186e-05,
+      "loss": 1.2566,
+      "step": 3947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.311564744968915e-05,
+      "loss": 1.2523,
+      "step": 3948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.311224300043858e-05,
+      "loss": 1.1536,
+      "step": 3949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.310883784409306e-05,
+      "loss": 1.2431,
+      "step": 3950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.310543198078555e-05,
+      "loss": 1.1589,
+      "step": 3951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.310202541064899e-05,
+      "loss": 1.2609,
+      "step": 3952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3098618133816376e-05,
+      "loss": 1.1875,
+      "step": 3953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3095210150420726e-05,
+      "loss": 1.2544,
+      "step": 3954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.309180146059509e-05,
+      "loss": 1.3223,
+      "step": 3955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.308839206447254e-05,
+      "loss": 1.212,
+      "step": 3956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3084981962186175e-05,
+      "loss": 1.234,
+      "step": 3957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.308157115386914e-05,
+      "loss": 1.2429,
+      "step": 3958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.307815963965457e-05,
+      "loss": 1.3617,
+      "step": 3959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.307474741967565e-05,
+      "loss": 1.2774,
+      "step": 3960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3071334494065606e-05,
+      "loss": 1.3263,
+      "step": 3961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.306792086295767e-05,
+      "loss": 1.2944,
+      "step": 3962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3064506526485116e-05,
+      "loss": 1.379,
+      "step": 3963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3061091484781224e-05,
+      "loss": 1.2667,
+      "step": 3964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.305767573797933e-05,
+      "loss": 1.2482,
+      "step": 3965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.305425928621277e-05,
+      "loss": 1.3314,
+      "step": 3966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.305084212961493e-05,
+      "loss": 1.3714,
+      "step": 3967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.304742426831922e-05,
+      "loss": 1.287,
+      "step": 3968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.304400570245906e-05,
+      "loss": 1.2127,
+      "step": 3969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.304058643216791e-05,
+      "loss": 1.2143,
+      "step": 3970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.303716645757927e-05,
+      "loss": 1.2655,
+      "step": 3971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.303374577882664e-05,
+      "loss": 1.3324,
+      "step": 3972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.303032439604356e-05,
+      "loss": 1.2328,
+      "step": 3973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.302690230936362e-05,
+      "loss": 1.2556,
+      "step": 3974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.30234795189204e-05,
+      "loss": 1.3188,
+      "step": 3975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.302005602484753e-05,
+      "loss": 1.2727,
+      "step": 3976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.301663182727866e-05,
+      "loss": 1.2785,
+      "step": 3977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3013206926347466e-05,
+      "loss": 0.9762,
+      "step": 3978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.300978132218766e-05,
+      "loss": 0.9677,
+      "step": 3979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.300635501493297e-05,
+      "loss": 0.9641,
+      "step": 3980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.300292800471717e-05,
+      "loss": 1.0263,
+      "step": 3981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.299950029167404e-05,
+      "loss": 0.998,
+      "step": 3982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2996071875937405e-05,
+      "loss": 1.084,
+      "step": 3983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.299264275764109e-05,
+      "loss": 1.0742,
+      "step": 3984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.298921293691899e-05,
+      "loss": 1.054,
+      "step": 3985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.298578241390498e-05,
+      "loss": 1.0358,
+      "step": 3986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2982351188733004e-05,
+      "loss": 1.0003,
+      "step": 3987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2978919261537016e-05,
+      "loss": 1.1862,
+      "step": 3988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2975486632450984e-05,
+      "loss": 1.096,
+      "step": 3989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.297205330160893e-05,
+      "loss": 1.0745,
+      "step": 3990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.296861926914488e-05,
+      "loss": 1.1421,
+      "step": 3991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2965184535192905e-05,
+      "loss": 1.0921,
+      "step": 3992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2961749099887085e-05,
+      "loss": 1.1868,
+      "step": 3993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.295831296336155e-05,
+      "loss": 1.2221,
+      "step": 3994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.295487612575044e-05,
+      "loss": 1.1501,
+      "step": 3995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.295143858718793e-05,
+      "loss": 1.1698,
+      "step": 3996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.294800034780822e-05,
+      "loss": 1.1732,
+      "step": 3997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.294456140774553e-05,
+      "loss": 1.2524,
+      "step": 3998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2941121767134124e-05,
+      "loss": 1.3348,
+      "step": 3999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.293768142610828e-05,
+      "loss": 1.3818,
+      "step": 4000
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.1710116863250732,
+      "eval_runtime": 113.3528,
+      "eval_samples_per_second": 33.709,
+      "eval_steps_per_second": 0.353,
+      "step": 4000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.293424038480231e-05,
+      "loss": 1.365,
+      "step": 4001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.293079864335055e-05,
+      "loss": 1.3069,
+      "step": 4002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.292735620188737e-05,
+      "loss": 1.3533,
+      "step": 4003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.292391306054714e-05,
+      "loss": 1.3431,
+      "step": 4004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.292046921946431e-05,
+      "loss": 1.3785,
+      "step": 4005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2917024678773296e-05,
+      "loss": 1.3342,
+      "step": 4006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.29135794386086e-05,
+      "loss": 1.3056,
+      "step": 4007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2910133499104696e-05,
+      "loss": 1.1919,
+      "step": 4008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.290668686039614e-05,
+      "loss": 1.1877,
+      "step": 4009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2903239522617464e-05,
+      "loss": 1.1906,
+      "step": 4010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.289979148590326e-05,
+      "loss": 1.2928,
+      "step": 4011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.289634275038814e-05,
+      "loss": 1.2123,
+      "step": 4012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2892893316206735e-05,
+      "loss": 1.3374,
+      "step": 4013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.288944318349371e-05,
+      "loss": 1.1172,
+      "step": 4014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2885992352383765e-05,
+      "loss": 1.1718,
+      "step": 4015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.288254082301162e-05,
+      "loss": 1.2266,
+      "step": 4016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2879088595512004e-05,
+      "loss": 1.2744,
+      "step": 4017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2875635670019706e-05,
+      "loss": 1.2518,
+      "step": 4018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2872182046669527e-05,
+      "loss": 1.2814,
+      "step": 4019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2868727725596286e-05,
+      "loss": 1.2864,
+      "step": 4020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2865272706934844e-05,
+      "loss": 1.2423,
+      "step": 4021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2861816990820084e-05,
+      "loss": 1.2819,
+      "step": 4022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.285836057738692e-05,
+      "loss": 1.2872,
+      "step": 4023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.285490346677028e-05,
+      "loss": 1.3013,
+      "step": 4024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2851445659105126e-05,
+      "loss": 1.2674,
+      "step": 4025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.284798715452646e-05,
+      "loss": 1.1167,
+      "step": 4026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.28445279531693e-05,
+      "loss": 1.1624,
+      "step": 4027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.284106805516869e-05,
+      "loss": 1.1314,
+      "step": 4028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.28376074606597e-05,
+      "loss": 1.114,
+      "step": 4029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2834146169777423e-05,
+      "loss": 1.1872,
+      "step": 4030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2830684182656996e-05,
+      "loss": 1.1049,
+      "step": 4031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.282722149943358e-05,
+      "loss": 1.0719,
+      "step": 4032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.282375812024235e-05,
+      "loss": 1.1505,
+      "step": 4033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.282029404521851e-05,
+      "loss": 1.1655,
+      "step": 4034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.28168292744973e-05,
+      "loss": 1.2861,
+      "step": 4035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2813363808213994e-05,
+      "loss": 1.2925,
+      "step": 4036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2809897646503863e-05,
+      "loss": 1.3216,
+      "step": 4037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2806430789502225e-05,
+      "loss": 1.3329,
+      "step": 4038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.280296323734445e-05,
+      "loss": 1.3035,
+      "step": 4039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.279949499016589e-05,
+      "loss": 1.2769,
+      "step": 4040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.279602604810194e-05,
+      "loss": 1.3622,
+      "step": 4041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.279255641128804e-05,
+      "loss": 1.2647,
+      "step": 4042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.278908607985964e-05,
+      "loss": 1.3646,
+      "step": 4043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.278561505395221e-05,
+      "loss": 1.3241,
+      "step": 4044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.278214333370126e-05,
+      "loss": 1.2187,
+      "step": 4045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2778670919242346e-05,
+      "loss": 1.1508,
+      "step": 4046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2775197810711e-05,
+      "loss": 1.2191,
+      "step": 4047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.277172400824283e-05,
+      "loss": 1.2304,
+      "step": 4048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.276824951197345e-05,
+      "loss": 1.2345,
+      "step": 4049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2764774322038494e-05,
+      "loss": 1.2092,
+      "step": 4050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.276129843857364e-05,
+      "loss": 1.2605,
+      "step": 4051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.275782186171457e-05,
+      "loss": 1.224,
+      "step": 4052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2754344591597035e-05,
+      "loss": 1.2468,
+      "step": 4053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.275086662835677e-05,
+      "loss": 1.1984,
+      "step": 4054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.274738797212955e-05,
+      "loss": 1.1696,
+      "step": 4055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.274390862305118e-05,
+      "loss": 1.1819,
+      "step": 4056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.274042858125751e-05,
+      "loss": 1.1695,
+      "step": 4057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.273694784688438e-05,
+      "loss": 1.126,
+      "step": 4058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.273346642006768e-05,
+      "loss": 1.1057,
+      "step": 4059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.272998430094334e-05,
+      "loss": 1.1832,
+      "step": 4060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.272650148964728e-05,
+      "loss": 1.1847,
+      "step": 4061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.272301798631547e-05,
+      "loss": 1.1664,
+      "step": 4062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.271953379108392e-05,
+      "loss": 1.1502,
+      "step": 4063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.271604890408863e-05,
+      "loss": 1.3126,
+      "step": 4064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2712563325465675e-05,
+      "loss": 1.3759,
+      "step": 4065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.270907705535111e-05,
+      "loss": 1.4123,
+      "step": 4066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2705590093881045e-05,
+      "loss": 1.3464,
+      "step": 4067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.270210244119161e-05,
+      "loss": 1.334,
+      "step": 4068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2698614097418956e-05,
+      "loss": 1.3044,
+      "step": 4069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.269512506269927e-05,
+      "loss": 1.2666,
+      "step": 4070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.269163533716877e-05,
+      "loss": 1.3392,
+      "step": 4071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.268814492096368e-05,
+      "loss": 1.3369,
+      "step": 4072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2684653814220266e-05,
+      "loss": 1.3184,
+      "step": 4073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.268116201707483e-05,
+      "loss": 1.3324,
+      "step": 4074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.267766952966369e-05,
+      "loss": 1.0382,
+      "step": 4075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2674176352123183e-05,
+      "loss": 1.044,
+      "step": 4076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2670682484589686e-05,
+      "loss": 1.099,
+      "step": 4077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.266718792719959e-05,
+      "loss": 0.9592,
+      "step": 4078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.266369268008934e-05,
+      "loss": 0.8739,
+      "step": 4079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.266019674339537e-05,
+      "loss": 0.9884,
+      "step": 4080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.265670011725417e-05,
+      "loss": 1.1424,
+      "step": 4081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.265320280180224e-05,
+      "loss": 0.9673,
+      "step": 4082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.264970479717612e-05,
+      "loss": 1.1131,
+      "step": 4083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.264620610351237e-05,
+      "loss": 1.0183,
+      "step": 4084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.264270672094757e-05,
+      "loss": 1.0257,
+      "step": 4085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2639206649618346e-05,
+      "loss": 1.3248,
+      "step": 4086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.263570588966133e-05,
+      "loss": 1.3072,
+      "step": 4087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.26322044412132e-05,
+      "loss": 1.2887,
+      "step": 4088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.262870230441064e-05,
+      "loss": 1.3398,
+      "step": 4089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.262519947939039e-05,
+      "loss": 1.2859,
+      "step": 4090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2621695966289174e-05,
+      "loss": 1.3036,
+      "step": 4091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2618191765243785e-05,
+      "loss": 1.3229,
+      "step": 4092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.261468687639102e-05,
+      "loss": 1.2188,
+      "step": 4093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2611181299867706e-05,
+      "loss": 1.2677,
+      "step": 4094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.260767503581071e-05,
+      "loss": 1.3166,
+      "step": 4095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.260416808435691e-05,
+      "loss": 1.2092,
+      "step": 4096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.260066044564321e-05,
+      "loss": 1.281,
+      "step": 4097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.259715211980655e-05,
+      "loss": 1.1924,
+      "step": 4098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2593643106983896e-05,
+      "loss": 1.2199,
+      "step": 4099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.259013340731224e-05,
+      "loss": 1.205,
+      "step": 4100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.25866230209286e-05,
+      "loss": 1.1514,
+      "step": 4101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.258311194797001e-05,
+      "loss": 1.1349,
+      "step": 4102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2579600188573544e-05,
+      "loss": 1.234,
+      "step": 4103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2576087742876306e-05,
+      "loss": 1.1701,
+      "step": 4104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.257257461101542e-05,
+      "loss": 1.1036,
+      "step": 4105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.256906079312804e-05,
+      "loss": 1.0359,
+      "step": 4106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.256554628935133e-05,
+      "loss": 1.1141,
+      "step": 4107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.256203109982251e-05,
+      "loss": 1.1218,
+      "step": 4108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2558515224678806e-05,
+      "loss": 1.2146,
+      "step": 4109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2554998664057476e-05,
+      "loss": 1.0728,
+      "step": 4110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2551481418095804e-05,
+      "loss": 1.1459,
+      "step": 4111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2547963486931105e-05,
+      "loss": 0.9948,
+      "step": 4112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2544444870700714e-05,
+      "loss": 1.1419,
+      "step": 4113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.254092556954199e-05,
+      "loss": 1.0932,
+      "step": 4114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.253740558359235e-05,
+      "loss": 1.035,
+      "step": 4115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2533884912989186e-05,
+      "loss": 1.0681,
+      "step": 4116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2530363557869954e-05,
+      "loss": 1.0252,
+      "step": 4117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.252684151837213e-05,
+      "loss": 1.081,
+      "step": 4118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.252331879463321e-05,
+      "loss": 1.0845,
+      "step": 4119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2519795386790716e-05,
+      "loss": 1.029,
+      "step": 4120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.251627129498221e-05,
+      "loss": 1.0823,
+      "step": 4121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.251274651934526e-05,
+      "loss": 1.1187,
+      "step": 4122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.250922106001748e-05,
+      "loss": 0.9818,
+      "step": 4123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2505694917136496e-05,
+      "loss": 1.3451,
+      "step": 4124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.250216809083997e-05,
+      "loss": 1.3439,
+      "step": 4125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.249864058126559e-05,
+      "loss": 1.3997,
+      "step": 4126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.249511238855107e-05,
+      "loss": 1.3322,
+      "step": 4127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.249158351283414e-05,
+      "loss": 1.2823,
+      "step": 4128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.248805395425257e-05,
+      "loss": 1.3206,
+      "step": 4129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.248452371294416e-05,
+      "loss": 1.4196,
+      "step": 4130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2480992789046725e-05,
+      "loss": 1.3056,
+      "step": 4131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2477461182698116e-05,
+      "loss": 1.304,
+      "step": 4132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2473928894036184e-05,
+      "loss": 1.2873,
+      "step": 4133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.247039592319886e-05,
+      "loss": 1.3778,
+      "step": 4134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.246686227032404e-05,
+      "loss": 1.3366,
+      "step": 4135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2463327935549704e-05,
+      "loss": 1.2318,
+      "step": 4136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.245979291901381e-05,
+      "loss": 1.2289,
+      "step": 4137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2456257220854366e-05,
+      "loss": 1.3147,
+      "step": 4138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2452720841209415e-05,
+      "loss": 1.2842,
+      "step": 4139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2449183780217e-05,
+      "loss": 1.1922,
+      "step": 4140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.244564603801523e-05,
+      "loss": 1.2137,
+      "step": 4141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.24421076147422e-05,
+      "loss": 1.2632,
+      "step": 4142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.243856851053605e-05,
+      "loss": 1.3067,
+      "step": 4143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.243502872553494e-05,
+      "loss": 1.2834,
+      "step": 4144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.243148825987708e-05,
+      "loss": 1.2696,
+      "step": 4145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.242794711370068e-05,
+      "loss": 1.1433,
+      "step": 4146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.242440528714397e-05,
+      "loss": 1.1074,
+      "step": 4147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2420862780345245e-05,
+      "loss": 1.2031,
+      "step": 4148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.241731959344279e-05,
+      "loss": 1.2177,
+      "step": 4149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.241377572657493e-05,
+      "loss": 1.1874,
+      "step": 4150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.241023117988002e-05,
+      "loss": 1.1882,
+      "step": 4151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.240668595349643e-05,
+      "loss": 1.2367,
+      "step": 4152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.240314004756257e-05,
+      "loss": 1.3905,
+      "step": 4153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.239959346221688e-05,
+      "loss": 1.3556,
+      "step": 4154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.23960461975978e-05,
+      "loss": 1.2578,
+      "step": 4155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2392498253843833e-05,
+      "loss": 1.3458,
+      "step": 4156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.238894963109347e-05,
+      "loss": 1.317,
+      "step": 4157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2385400329485255e-05,
+      "loss": 1.3457,
+      "step": 4158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.238185034915775e-05,
+      "loss": 1.2923,
+      "step": 4159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.237829969024956e-05,
+      "loss": 1.305,
+      "step": 4160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.237474835289927e-05,
+      "loss": 1.2901,
+      "step": 4161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.237119633724556e-05,
+      "loss": 1.2965,
+      "step": 4162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2367643643427065e-05,
+      "loss": 1.4065,
+      "step": 4163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.236409027158251e-05,
+      "loss": 1.4102,
+      "step": 4164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.23605362218506e-05,
+      "loss": 1.3515,
+      "step": 4165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.235698149437009e-05,
+      "loss": 1.3418,
+      "step": 4166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.235342608927975e-05,
+      "loss": 1.2958,
+      "step": 4167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2349870006718385e-05,
+      "loss": 1.3529,
+      "step": 4168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.234631324682482e-05,
+      "loss": 1.3576,
+      "step": 4169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.234275580973792e-05,
+      "loss": 1.2847,
+      "step": 4170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.233919769559655e-05,
+      "loss": 1.3353,
+      "step": 4171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2335638904539626e-05,
+      "loss": 1.2396,
+      "step": 4172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.233207943670609e-05,
+      "loss": 1.2965,
+      "step": 4173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.232851929223488e-05,
+      "loss": 1.1837,
+      "step": 4174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2324958471265006e-05,
+      "loss": 1.2507,
+      "step": 4175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.232139697393547e-05,
+      "loss": 1.2199,
+      "step": 4176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.231783480038531e-05,
+      "loss": 1.1834,
+      "step": 4177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.231427195075359e-05,
+      "loss": 1.2575,
+      "step": 4178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.231070842517941e-05,
+      "loss": 1.2341,
+      "step": 4179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.23071442238019e-05,
+      "loss": 1.2052,
+      "step": 4180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.230357934676017e-05,
+      "loss": 1.2959,
+      "step": 4181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2300013794193426e-05,
+      "loss": 1.2025,
+      "step": 4182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.229644756624085e-05,
+      "loss": 1.142,
+      "step": 4183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.229288066304166e-05,
+      "loss": 0.99,
+      "step": 4184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.228931308473511e-05,
+      "loss": 0.9522,
+      "step": 4185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.22857448314605e-05,
+      "loss": 1.084,
+      "step": 4186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.22821759033571e-05,
+      "loss": 0.9696,
+      "step": 4187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.227860630056426e-05,
+      "loss": 1.0139,
+      "step": 4188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.227503602322133e-05,
+      "loss": 1.0365,
+      "step": 4189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2271465071467694e-05,
+      "loss": 1.0538,
+      "step": 4190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.226789344544276e-05,
+      "loss": 0.9916,
+      "step": 4191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2264321145285956e-05,
+      "loss": 1.1156,
+      "step": 4192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.226074817113675e-05,
+      "loss": 1.1255,
+      "step": 4193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.225717452313464e-05,
+      "loss": 1.3004,
+      "step": 4194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.225360020141911e-05,
+      "loss": 1.3789,
+      "step": 4195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.225002520612973e-05,
+      "loss": 1.3248,
+      "step": 4196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2246449537406054e-05,
+      "loss": 1.4244,
+      "step": 4197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2242873195387666e-05,
+      "loss": 1.4495,
+      "step": 4198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2239296180214204e-05,
+      "loss": 1.3805,
+      "step": 4199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.22357184920253e-05,
+      "loss": 1.311,
+      "step": 4200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2232140130960626e-05,
+      "loss": 1.3116,
+      "step": 4201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2228561097159894e-05,
+      "loss": 1.3659,
+      "step": 4202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2224981390762805e-05,
+      "loss": 1.3577,
+      "step": 4203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.222140101190913e-05,
+      "loss": 1.2878,
+      "step": 4204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.221781996073865e-05,
+      "loss": 1.0917,
+      "step": 4205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.221423823739114e-05,
+      "loss": 1.1616,
+      "step": 4206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.221065584200644e-05,
+      "loss": 1.1177,
+      "step": 4207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2207072774724416e-05,
+      "loss": 1.1256,
+      "step": 4208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.220348903568494e-05,
+      "loss": 1.1309,
+      "step": 4209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.219990462502793e-05,
+      "loss": 1.089,
+      "step": 4210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2196319542893306e-05,
+      "loss": 1.0871,
+      "step": 4211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.219273378942104e-05,
+      "loss": 1.229,
+      "step": 4212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.218914736475111e-05,
+      "loss": 1.1435,
+      "step": 4213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2185560269023534e-05,
+      "loss": 1.2757,
+      "step": 4214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2181972502378345e-05,
+      "loss": 1.162,
+      "step": 4215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.217838406495562e-05,
+      "loss": 1.1018,
+      "step": 4216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2174794956895436e-05,
+      "loss": 1.1407,
+      "step": 4217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.217120517833792e-05,
+      "loss": 1.2316,
+      "step": 4218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.216761472942321e-05,
+      "loss": 1.2791,
+      "step": 4219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.216402361029148e-05,
+      "loss": 1.2187,
+      "step": 4220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.216043182108292e-05,
+      "loss": 1.2142,
+      "step": 4221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2156839361937764e-05,
+      "loss": 1.1603,
+      "step": 4222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.215324623299624e-05,
+      "loss": 1.3126,
+      "step": 4223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.214965243439865e-05,
+      "loss": 1.2746,
+      "step": 4224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.214605796628527e-05,
+      "loss": 1.1301,
+      "step": 4225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.214246282879643e-05,
+      "loss": 1.1207,
+      "step": 4226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.21388670220725e-05,
+      "loss": 1.2295,
+      "step": 4227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.213527054625383e-05,
+      "loss": 1.1886,
+      "step": 4228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2131673401480865e-05,
+      "loss": 1.215,
+      "step": 4229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2128075587894e-05,
+      "loss": 1.1873,
+      "step": 4230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2124477105633706e-05,
+      "loss": 1.1247,
+      "step": 4231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.212087795484047e-05,
+      "loss": 1.3084,
+      "step": 4232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.211727813565479e-05,
+      "loss": 1.2449,
+      "step": 4233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.211367764821722e-05,
+      "loss": 1.3317,
+      "step": 4234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.211007649266831e-05,
+      "loss": 1.2912,
+      "step": 4235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2106474669148645e-05,
+      "loss": 1.2914,
+      "step": 4236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.210287217779883e-05,
+      "loss": 1.3349,
+      "step": 4237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.209926901875954e-05,
+      "loss": 1.3495,
+      "step": 4238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2095665192171395e-05,
+      "loss": 1.3267,
+      "step": 4239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.209206069817513e-05,
+      "loss": 1.2315,
+      "step": 4240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.208845553691143e-05,
+      "loss": 1.2983,
+      "step": 4241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.208484970852106e-05,
+      "loss": 1.2486,
+      "step": 4242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.208124321314477e-05,
+      "loss": 1.3291,
+      "step": 4243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.207763605092339e-05,
+      "loss": 1.2848,
+      "step": 4244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.20740282219977e-05,
+      "loss": 1.2312,
+      "step": 4245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.207041972650858e-05,
+      "loss": 1.3062,
+      "step": 4246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2066810564596895e-05,
+      "loss": 1.295,
+      "step": 4247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2063200736403527e-05,
+      "loss": 1.2353,
+      "step": 4248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.205959024206944e-05,
+      "loss": 1.2505,
+      "step": 4249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.205597908173555e-05,
+      "loss": 1.2189,
+      "step": 4250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2052367255542854e-05,
+      "loss": 0.9748,
+      "step": 4251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.204875476363236e-05,
+      "loss": 1.0428,
+      "step": 4252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2045141606145085e-05,
+      "loss": 0.9514,
+      "step": 4253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.204152778322209e-05,
+      "loss": 1.0094,
+      "step": 4254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.203791329500446e-05,
+      "loss": 0.9509,
+      "step": 4255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.203429814163331e-05,
+      "loss": 0.9799,
+      "step": 4256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2030682323249756e-05,
+      "loss": 1.0673,
+      "step": 4257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2027065839994964e-05,
+      "loss": 1.0778,
+      "step": 4258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.202344869201013e-05,
+      "loss": 0.9856,
+      "step": 4259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.201983087943646e-05,
+      "loss": 0.883,
+      "step": 4260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2016212402415194e-05,
+      "loss": 1.0634,
+      "step": 4261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.201259326108759e-05,
+      "loss": 0.9598,
+      "step": 4262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.200897345559495e-05,
+      "loss": 1.0791,
+      "step": 4263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2005352986078574e-05,
+      "loss": 0.9132,
+      "step": 4264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.200173185267981e-05,
+      "loss": 0.9781,
+      "step": 4265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.199811005554003e-05,
+      "loss": 1.0192,
+      "step": 4266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.199448759480063e-05,
+      "loss": 1.1442,
+      "step": 4267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1990864470603014e-05,
+      "loss": 1.0744,
+      "step": 4268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.198724068308865e-05,
+      "loss": 1.1297,
+      "step": 4269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1983616232398986e-05,
+      "loss": 0.9647,
+      "step": 4270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.197999111867553e-05,
+      "loss": 0.9773,
+      "step": 4271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.197636534205981e-05,
+      "loss": 1.0452,
+      "step": 4272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.197273890269337e-05,
+      "loss": 1.1304,
+      "step": 4273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.196911180071778e-05,
+      "loss": 1.1792,
+      "step": 4274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.196548403627465e-05,
+      "loss": 1.1737,
+      "step": 4275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.196185560950561e-05,
+      "loss": 1.2033,
+      "step": 4276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.195822652055229e-05,
+      "loss": 1.1409,
+      "step": 4277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1954596769556384e-05,
+      "loss": 1.2176,
+      "step": 4278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1950966356659606e-05,
+      "loss": 1.2126,
+      "step": 4279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.194733528200366e-05,
+      "loss": 1.1953,
+      "step": 4280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.194370354573033e-05,
+      "loss": 1.2719,
+      "step": 4281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1940071147981375e-05,
+      "loss": 1.247,
+      "step": 4282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.193643808889861e-05,
+      "loss": 1.1952,
+      "step": 4283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.193280436862387e-05,
+      "loss": 1.2199,
+      "step": 4284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1929169987299015e-05,
+      "loss": 1.2091,
+      "step": 4285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.192553494506593e-05,
+      "loss": 1.2495,
+      "step": 4286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.192189924206652e-05,
+      "loss": 1.237,
+      "step": 4287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.191826287844273e-05,
+      "loss": 1.2164,
+      "step": 4288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.191462585433651e-05,
+      "loss": 1.1969,
+      "step": 4289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1910988169889865e-05,
+      "loss": 1.2547,
+      "step": 4290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.19073498252448e-05,
+      "loss": 1.2323,
+      "step": 4291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1903710820543344e-05,
+      "loss": 1.2624,
+      "step": 4292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.190007115592758e-05,
+      "loss": 1.2573,
+      "step": 4293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1896430831539586e-05,
+      "loss": 1.2917,
+      "step": 4294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1892789847521485e-05,
+      "loss": 1.281,
+      "step": 4295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1889148204015424e-05,
+      "loss": 1.2281,
+      "step": 4296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.188550590116356e-05,
+      "loss": 1.2288,
+      "step": 4297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.18818629391081e-05,
+      "loss": 1.3263,
+      "step": 4298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.187821931799125e-05,
+      "loss": 1.2579,
+      "step": 4299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.187457503795527e-05,
+      "loss": 1.2826,
+      "step": 4300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1870930099142416e-05,
+      "loss": 1.0619,
+      "step": 4301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1867284501695e-05,
+      "loss": 1.1496,
+      "step": 4302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1863638245755335e-05,
+      "loss": 1.137,
+      "step": 4303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.185999133146576e-05,
+      "loss": 1.18,
+      "step": 4304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.185634375896868e-05,
+      "loss": 1.1451,
+      "step": 4305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1852695528406474e-05,
+      "loss": 1.1868,
+      "step": 4306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.184904663992156e-05,
+      "loss": 1.1307,
+      "step": 4307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.18453970936564e-05,
+      "loss": 1.1251,
+      "step": 4308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1841746889753476e-05,
+      "loss": 1.1953,
+      "step": 4309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.183809602835527e-05,
+      "loss": 1.1057,
+      "step": 4310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.183444450960434e-05,
+      "loss": 1.0947,
+      "step": 4311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.183079233364322e-05,
+      "loss": 1.1692,
+      "step": 4312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.182713950061449e-05,
+      "loss": 1.2272,
+      "step": 4313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.182348601066076e-05,
+      "loss": 1.2174,
+      "step": 4314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.181983186392466e-05,
+      "loss": 1.2684,
+      "step": 4315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.181617706054885e-05,
+      "loss": 1.2585,
+      "step": 4316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1812521600676e-05,
+      "loss": 1.1837,
+      "step": 4317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1808865484448835e-05,
+      "loss": 1.2316,
+      "step": 4318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1805208712010074e-05,
+      "loss": 1.2291,
+      "step": 4319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1801551283502485e-05,
+      "loss": 1.0814,
+      "step": 4320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.179789319906885e-05,
+      "loss": 1.2362,
+      "step": 4321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.179423445885197e-05,
+      "loss": 1.1786,
+      "step": 4322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.17905750629947e-05,
+      "loss": 1.1182,
+      "step": 4323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.178691501163988e-05,
+      "loss": 1.2327,
+      "step": 4324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.178325430493041e-05,
+      "loss": 1.2684,
+      "step": 4325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.177959294300921e-05,
+      "loss": 1.1239,
+      "step": 4326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.177593092601919e-05,
+      "loss": 1.2413,
+      "step": 4327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1772268254103344e-05,
+      "loss": 1.0986,
+      "step": 4328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.176860492740464e-05,
+      "loss": 1.1258,
+      "step": 4329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.176494094606611e-05,
+      "loss": 1.1917,
+      "step": 4330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.176127631023079e-05,
+      "loss": 1.2114,
+      "step": 4331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.175761102004174e-05,
+      "loss": 1.2734,
+      "step": 4332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.175394507564205e-05,
+      "loss": 1.3208,
+      "step": 4333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.175027847717484e-05,
+      "loss": 1.3163,
+      "step": 4334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1746611224783247e-05,
+      "loss": 1.275,
+      "step": 4335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.174294331861046e-05,
+      "loss": 1.3616,
+      "step": 4336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.173927475879965e-05,
+      "loss": 1.1837,
+      "step": 4337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.173560554549405e-05,
+      "loss": 1.3041,
+      "step": 4338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.173193567883689e-05,
+      "loss": 1.2728,
+      "step": 4339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.172826515897146e-05,
+      "loss": 1.4251,
+      "step": 4340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.172459398604104e-05,
+      "loss": 1.1923,
+      "step": 4341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.172092216018895e-05,
+      "loss": 1.1352,
+      "step": 4342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1717249681558544e-05,
+      "loss": 1.1745,
+      "step": 4343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1713576550293204e-05,
+      "loss": 1.2016,
+      "step": 4344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.170990276653631e-05,
+      "loss": 1.1817,
+      "step": 4345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.170622833043129e-05,
+      "loss": 1.1421,
+      "step": 4346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.17025532421216e-05,
+      "loss": 1.2068,
+      "step": 4347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.169887750175069e-05,
+      "loss": 1.2303,
+      "step": 4348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.16952011094621e-05,
+      "loss": 1.2523,
+      "step": 4349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1691524065399324e-05,
+      "loss": 1.1542,
+      "step": 4350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.168784636970592e-05,
+      "loss": 1.1641,
+      "step": 4351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.168416802252546e-05,
+      "loss": 1.2221,
+      "step": 4352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.168048902400156e-05,
+      "loss": 1.2059,
+      "step": 4353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1676809374277823e-05,
+      "loss": 1.2771,
+      "step": 4354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.167312907349792e-05,
+      "loss": 1.302,
+      "step": 4355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.166944812180553e-05,
+      "loss": 1.2422,
+      "step": 4356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1665766519344344e-05,
+      "loss": 1.2778,
+      "step": 4357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.16620842662581e-05,
+      "loss": 1.4203,
+      "step": 4358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.165840136269054e-05,
+      "loss": 1.2319,
+      "step": 4359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.165471780878546e-05,
+      "loss": 1.2074,
+      "step": 4360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1651033604686645e-05,
+      "loss": 1.3315,
+      "step": 4361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.164734875053794e-05,
+      "loss": 1.3709,
+      "step": 4362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.164366324648319e-05,
+      "loss": 1.3163,
+      "step": 4363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.163997709266629e-05,
+      "loss": 1.3964,
+      "step": 4364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.163629028923113e-05,
+      "loss": 1.3409,
+      "step": 4365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.163260283632165e-05,
+      "loss": 1.3144,
+      "step": 4366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1628914734081804e-05,
+      "loss": 1.3679,
+      "step": 4367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.162522598265557e-05,
+      "loss": 1.3567,
+      "step": 4368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.162153658218697e-05,
+      "loss": 1.2994,
+      "step": 4369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.161784653282002e-05,
+      "loss": 1.4108,
+      "step": 4370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1614155834698795e-05,
+      "loss": 1.3074,
+      "step": 4371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.161046448796736e-05,
+      "loss": 1.5695,
+      "step": 4372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.160677249276983e-05,
+      "loss": 1.1715,
+      "step": 4373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.160307984925035e-05,
+      "loss": 1.1783,
+      "step": 4374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.159938655755306e-05,
+      "loss": 1.2425,
+      "step": 4375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.159569261782217e-05,
+      "loss": 1.166,
+      "step": 4376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.159199803020186e-05,
+      "loss": 1.0927,
+      "step": 4377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.158830279483639e-05,
+      "loss": 1.1695,
+      "step": 4378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1584606911870004e-05,
+      "loss": 1.1456,
+      "step": 4379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1580910381447004e-05,
+      "loss": 1.1697,
+      "step": 4380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.157721320371168e-05,
+      "loss": 1.1118,
+      "step": 4381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.15735153788084e-05,
+      "loss": 1.2627,
+      "step": 4382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.156981690688149e-05,
+      "loss": 1.3901,
+      "step": 4383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1566117788075356e-05,
+      "loss": 1.3319,
+      "step": 4384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.156241802253441e-05,
+      "loss": 1.3539,
+      "step": 4385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1558717610403084e-05,
+      "loss": 1.5364,
+      "step": 4386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1555016551825846e-05,
+      "loss": 1.527,
+      "step": 4387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.155131484694719e-05,
+      "loss": 1.3641,
+      "step": 4388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.15476124959116e-05,
+      "loss": 1.3642,
+      "step": 4389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.154390949886365e-05,
+      "loss": 1.4762,
+      "step": 4390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.154020585594789e-05,
+      "loss": 1.4054,
+      "step": 4391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.15365015673089e-05,
+      "loss": 1.3653,
+      "step": 4392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1532796633091296e-05,
+      "loss": 1.2906,
+      "step": 4393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1529091053439726e-05,
+      "loss": 1.3157,
+      "step": 4394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.152538482849886e-05,
+      "loss": 1.2931,
+      "step": 4395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.152167795841337e-05,
+      "loss": 1.3369,
+      "step": 4396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.151797044332799e-05,
+      "loss": 1.2855,
+      "step": 4397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.151426228338743e-05,
+      "loss": 1.3628,
+      "step": 4398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.151055347873648e-05,
+      "loss": 1.3506,
+      "step": 4399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.150684402951994e-05,
+      "loss": 1.3288,
+      "step": 4400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.150313393588259e-05,
+      "loss": 1.2434,
+      "step": 4401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.14994231979693e-05,
+      "loss": 1.364,
+      "step": 4402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1495711815924924e-05,
+      "loss": 1.3118,
+      "step": 4403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.149199978989436e-05,
+      "loss": 1.1708,
+      "step": 4404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.148828712002252e-05,
+      "loss": 1.2303,
+      "step": 4405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1484573806454335e-05,
+      "loss": 1.2928,
+      "step": 4406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.148085984933479e-05,
+      "loss": 1.2233,
+      "step": 4407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.147714524880886e-05,
+      "loss": 1.2636,
+      "step": 4408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.147343000502157e-05,
+      "loss": 1.2554,
+      "step": 4409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1469714118117964e-05,
+      "loss": 1.2341,
+      "step": 4410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.14659975882431e-05,
+      "loss": 1.3153,
+      "step": 4411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1462280415542075e-05,
+      "loss": 1.2454,
+      "step": 4412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.145856260016001e-05,
+      "loss": 1.2551,
+      "step": 4413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.145484414224204e-05,
+      "loss": 1.1179,
+      "step": 4414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1451125041933344e-05,
+      "loss": 1.2783,
+      "step": 4415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.14474052993791e-05,
+      "loss": 1.1443,
+      "step": 4416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1443684914724534e-05,
+      "loss": 1.3218,
+      "step": 4417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1439963888114874e-05,
+      "loss": 1.2272,
+      "step": 4418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1436242219695406e-05,
+      "loss": 1.1946,
+      "step": 4419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1432519909611415e-05,
+      "loss": 1.2543,
+      "step": 4420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.142879695800822e-05,
+      "loss": 1.3067,
+      "step": 4421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.142507336503116e-05,
+      "loss": 1.1683,
+      "step": 4422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.142134913082561e-05,
+      "loss": 1.3013,
+      "step": 4423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.141762425553695e-05,
+      "loss": 1.3022,
+      "step": 4424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1413898739310605e-05,
+      "loss": 1.2691,
+      "step": 4425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.141017258229203e-05,
+      "loss": 1.3238,
+      "step": 4426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.140644578462668e-05,
+      "loss": 1.3496,
+      "step": 4427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1402718346460035e-05,
+      "loss": 1.2707,
+      "step": 4428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.139899026793763e-05,
+      "loss": 1.2673,
+      "step": 4429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1395261549205e-05,
+      "loss": 1.2418,
+      "step": 4430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1391532190407734e-05,
+      "loss": 1.3076,
+      "step": 4431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.138780219169139e-05,
+      "loss": 1.3045,
+      "step": 4432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.138407155320161e-05,
+      "loss": 1.3544,
+      "step": 4433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.138034027508403e-05,
+      "loss": 1.1788,
+      "step": 4434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1376608357484314e-05,
+      "loss": 1.2232,
+      "step": 4435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.137287580054816e-05,
+      "loss": 1.2737,
+      "step": 4436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.136914260442128e-05,
+      "loss": 1.1189,
+      "step": 4437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.136540876924943e-05,
+      "loss": 1.2687,
+      "step": 4438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.136167429517836e-05,
+      "loss": 1.3089,
+      "step": 4439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1357939182353874e-05,
+      "loss": 1.1657,
+      "step": 4440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.135420343092178e-05,
+      "loss": 1.201,
+      "step": 4441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.135046704102793e-05,
+      "loss": 1.2729,
+      "step": 4442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.134673001281818e-05,
+      "loss": 1.0849,
+      "step": 4443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.134299234643844e-05,
+      "loss": 1.0007,
+      "step": 4444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.133925404203461e-05,
+      "loss": 1.0246,
+      "step": 4445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.133551509975264e-05,
+      "loss": 1.1249,
+      "step": 4446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.133177551973849e-05,
+      "loss": 1.0617,
+      "step": 4447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.132803530213816e-05,
+      "loss": 1.086,
+      "step": 4448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1324294447097665e-05,
+      "loss": 0.9955,
+      "step": 4449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1320552954763044e-05,
+      "loss": 1.0235,
+      "step": 4450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.131681082528036e-05,
+      "loss": 0.9645,
+      "step": 4451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1313068058795716e-05,
+      "loss": 0.9508,
+      "step": 4452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1309324655455216e-05,
+      "loss": 1.0214,
+      "step": 4453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1305580615405e-05,
+      "loss": 1.1884,
+      "step": 4454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.130183593879125e-05,
+      "loss": 1.191,
+      "step": 4455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.129809062576014e-05,
+      "loss": 1.1751,
+      "step": 4456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1294344676457895e-05,
+      "loss": 1.1474,
+      "step": 4457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1290598091030755e-05,
+      "loss": 1.1893,
+      "step": 4458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.128685086962497e-05,
+      "loss": 1.1934,
+      "step": 4459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.128310301238686e-05,
+      "loss": 1.1273,
+      "step": 4460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.127935451946271e-05,
+      "loss": 1.1629,
+      "step": 4461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1275605390998874e-05,
+      "loss": 1.2197,
+      "step": 4462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1271855627141714e-05,
+      "loss": 1.2622,
+      "step": 4463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1268105228037626e-05,
+      "loss": 1.2095,
+      "step": 4464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.126435419383301e-05,
+      "loss": 1.2808,
+      "step": 4465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.126060252467432e-05,
+      "loss": 1.3205,
+      "step": 4466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.125685022070801e-05,
+      "loss": 1.2901,
+      "step": 4467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1253097282080564e-05,
+      "loss": 1.2995,
+      "step": 4468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.124934370893851e-05,
+      "loss": 1.3014,
+      "step": 4469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.124558950142838e-05,
+      "loss": 1.2389,
+      "step": 4470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.124183465969674e-05,
+      "loss": 1.2687,
+      "step": 4471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1238079183890165e-05,
+      "loss": 1.2074,
+      "step": 4472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.123432307415527e-05,
+      "loss": 1.2317,
+      "step": 4473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1230566330638706e-05,
+      "loss": 1.161,
+      "step": 4474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.122680895348713e-05,
+      "loss": 1.2552,
+      "step": 4475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.122305094284722e-05,
+      "loss": 1.2857,
+      "step": 4476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.121929229886569e-05,
+      "loss": 1.2607,
+      "step": 4477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.121553302168927e-05,
+      "loss": 1.307,
+      "step": 4478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.121177311146474e-05,
+      "loss": 1.2639,
+      "step": 4479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.120801256833887e-05,
+      "loss": 1.2645,
+      "step": 4480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.120425139245848e-05,
+      "loss": 1.2792,
+      "step": 4481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.120048958397039e-05,
+      "loss": 1.1954,
+      "step": 4482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.119672714302147e-05,
+      "loss": 1.3008,
+      "step": 4483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.119296406975861e-05,
+      "loss": 1.24,
+      "step": 4484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1189200364328713e-05,
+      "loss": 1.299,
+      "step": 4485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1185436026878706e-05,
+      "loss": 1.2493,
+      "step": 4486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.118167105755555e-05,
+      "loss": 1.3523,
+      "step": 4487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.117790545650623e-05,
+      "loss": 1.3149,
+      "step": 4488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.117413922387776e-05,
+      "loss": 1.2626,
+      "step": 4489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.117037235981717e-05,
+      "loss": 1.3019,
+      "step": 4490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.11666048644715e-05,
+      "loss": 1.3172,
+      "step": 4491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1162836737987854e-05,
+      "loss": 1.3065,
+      "step": 4492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.115906798051333e-05,
+      "loss": 1.1482,
+      "step": 4493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.115529859219506e-05,
+      "loss": 1.2222,
+      "step": 4494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1151528573180186e-05,
+      "loss": 1.1515,
+      "step": 4495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.114775792361591e-05,
+      "loss": 1.2472,
+      "step": 4496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1143986643649426e-05,
+      "loss": 1.2134,
+      "step": 4497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1140214733427954e-05,
+      "loss": 1.2127,
+      "step": 4498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.113644219309877e-05,
+      "loss": 1.1777,
+      "step": 4499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1132669022809136e-05,
+      "loss": 1.2015,
+      "step": 4500
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.1718709468841553,
+      "eval_runtime": 115.3224,
+      "eval_samples_per_second": 33.133,
+      "eval_steps_per_second": 0.347,
+      "step": 4500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.112889522270635e-05,
+      "loss": 1.203,
+      "step": 4501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1125120792937756e-05,
+      "loss": 1.1934,
+      "step": 4502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1121345733650694e-05,
+      "loss": 1.2789,
+      "step": 4503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.111757004499255e-05,
+      "loss": 1.3329,
+      "step": 4504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.111379372711073e-05,
+      "loss": 1.2149,
+      "step": 4505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.111001678015264e-05,
+      "loss": 1.3011,
+      "step": 4506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.110623920426574e-05,
+      "loss": 1.2689,
+      "step": 4507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.110246099959751e-05,
+      "loss": 1.2309,
+      "step": 4508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1098682166295455e-05,
+      "loss": 1.2535,
+      "step": 4509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.109490270450709e-05,
+      "loss": 1.3408,
+      "step": 4510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.109112261437995e-05,
+      "loss": 1.2356,
+      "step": 4511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1087341896061634e-05,
+      "loss": 1.0869,
+      "step": 4512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.108356054969972e-05,
+      "loss": 1.2109,
+      "step": 4513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1079778575441854e-05,
+      "loss": 1.137,
+      "step": 4514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.107599597343566e-05,
+      "loss": 1.1529,
+      "step": 4515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.107221274382881e-05,
+      "loss": 1.2412,
+      "step": 4516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.106842888676901e-05,
+      "loss": 1.1828,
+      "step": 4517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1064644402403974e-05,
+      "loss": 1.1403,
+      "step": 4518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.106085929088146e-05,
+      "loss": 1.0707,
+      "step": 4519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1057073552349226e-05,
+      "loss": 1.2048,
+      "step": 4520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.105328718695506e-05,
+      "loss": 1.299,
+      "step": 4521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1049500194846784e-05,
+      "loss": 1.1921,
+      "step": 4522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.104571257617225e-05,
+      "loss": 1.2912,
+      "step": 4523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.104192433107932e-05,
+      "loss": 1.2574,
+      "step": 4524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.103813545971589e-05,
+      "loss": 1.3317,
+      "step": 4525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.103434596222986e-05,
+      "loss": 1.2111,
+      "step": 4526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.103055583876918e-05,
+      "loss": 1.2168,
+      "step": 4527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1026765089481824e-05,
+      "loss": 1.2248,
+      "step": 4528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1022973714515764e-05,
+      "loss": 1.2893,
+      "step": 4529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.101918171401904e-05,
+      "loss": 1.2458,
+      "step": 4530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1015389088139664e-05,
+      "loss": 1.2431,
+      "step": 4531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.10115958370257e-05,
+      "loss": 1.3687,
+      "step": 4532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.100780196082525e-05,
+      "loss": 1.4505,
+      "step": 4533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.100400745968642e-05,
+      "loss": 1.5005,
+      "step": 4534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.100021233375735e-05,
+      "loss": 1.4784,
+      "step": 4535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.099641658318619e-05,
+      "loss": 1.5509,
+      "step": 4536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0992620208121124e-05,
+      "loss": 1.4901,
+      "step": 4537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0988823208710374e-05,
+      "loss": 1.5246,
+      "step": 4538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.098502558510217e-05,
+      "loss": 1.4373,
+      "step": 4539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.098122733744475e-05,
+      "loss": 1.4237,
+      "step": 4540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0977428465886425e-05,
+      "loss": 1.5137,
+      "step": 4541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.097362897057548e-05,
+      "loss": 1.554,
+      "step": 4542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.096982885166026e-05,
+      "loss": 1.4972,
+      "step": 4543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0966028109289114e-05,
+      "loss": 1.3485,
+      "step": 4544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0962226743610424e-05,
+      "loss": 1.249,
+      "step": 4545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0958424754772584e-05,
+      "loss": 1.2722,
+      "step": 4546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0954622142924026e-05,
+      "loss": 1.2633,
+      "step": 4547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.095081890821322e-05,
+      "loss": 1.2694,
+      "step": 4548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.094701505078862e-05,
+      "loss": 1.215,
+      "step": 4549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.094321057079874e-05,
+      "loss": 1.2802,
+      "step": 4550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.093940546839209e-05,
+      "loss": 1.2071,
+      "step": 4551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.093559974371725e-05,
+      "loss": 1.18,
+      "step": 4552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.093179339692276e-05,
+      "loss": 1.2696,
+      "step": 4553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.092798642815724e-05,
+      "loss": 1.3535,
+      "step": 4554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.09241788375693e-05,
+      "loss": 1.3098,
+      "step": 4555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0920370625307606e-05,
+      "loss": 1.3152,
+      "step": 4556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.091656179152081e-05,
+      "loss": 1.3597,
+      "step": 4557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0912752336357604e-05,
+      "loss": 1.3657,
+      "step": 4558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.090894225996673e-05,
+      "loss": 1.3833,
+      "step": 4559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0905131562496915e-05,
+      "loss": 1.2876,
+      "step": 4560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0901320244096934e-05,
+      "loss": 1.3195,
+      "step": 4561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.089750830491558e-05,
+      "loss": 1.3185,
+      "step": 4562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.089369574510165e-05,
+      "loss": 1.3265,
+      "step": 4563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0889882564804016e-05,
+      "loss": 1.3529,
+      "step": 4564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.088606876417152e-05,
+      "loss": 1.2729,
+      "step": 4565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0882254343353066e-05,
+      "loss": 1.3106,
+      "step": 4566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0878439302497554e-05,
+      "loss": 1.2265,
+      "step": 4567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.087462364175394e-05,
+      "loss": 1.3339,
+      "step": 4568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.087080736127116e-05,
+      "loss": 1.2017,
+      "step": 4569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.086699046119823e-05,
+      "loss": 1.3484,
+      "step": 4570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.086317294168414e-05,
+      "loss": 1.1932,
+      "step": 4571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.085935480287792e-05,
+      "loss": 1.139,
+      "step": 4572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.085553604492864e-05,
+      "loss": 1.3296,
+      "step": 4573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.085171666798539e-05,
+      "loss": 1.3499,
+      "step": 4574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0847896672197264e-05,
+      "loss": 1.2817,
+      "step": 4575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.08440760577134e-05,
+      "loss": 1.1638,
+      "step": 4576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.084025482468294e-05,
+      "loss": 1.1096,
+      "step": 4577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.083643297325509e-05,
+      "loss": 1.149,
+      "step": 4578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.083261050357903e-05,
+      "loss": 1.143,
+      "step": 4579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.082878741580399e-05,
+      "loss": 1.1328,
+      "step": 4580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0824963710079234e-05,
+      "loss": 1.0904,
+      "step": 4581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.082113938655403e-05,
+      "loss": 1.2335,
+      "step": 4582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0817314445377674e-05,
+      "loss": 1.139,
+      "step": 4583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0813488886699506e-05,
+      "loss": 1.1482,
+      "step": 4584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.080966271066885e-05,
+      "loss": 1.1601,
+      "step": 4585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0805835917435104e-05,
+      "loss": 1.1727,
+      "step": 4586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0802008507147643e-05,
+      "loss": 1.243,
+      "step": 4587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.07981804799559e-05,
+      "loss": 1.222,
+      "step": 4588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0794351836009325e-05,
+      "loss": 1.1778,
+      "step": 4589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.079052257545738e-05,
+      "loss": 1.1977,
+      "step": 4590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.078669269844955e-05,
+      "loss": 1.2671,
+      "step": 4591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0782862205135355e-05,
+      "loss": 1.26,
+      "step": 4592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0779031095664343e-05,
+      "loss": 1.1647,
+      "step": 4593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.077519937018608e-05,
+      "loss": 1.2159,
+      "step": 4594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0771367028850136e-05,
+      "loss": 1.0813,
+      "step": 4595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.076753407180616e-05,
+      "loss": 1.1321,
+      "step": 4596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.076370049920375e-05,
+      "loss": 1.1046,
+      "step": 4597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0759866311192596e-05,
+      "loss": 1.0609,
+      "step": 4598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0756031507922364e-05,
+      "loss": 1.1099,
+      "step": 4599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.075219608954278e-05,
+      "loss": 1.0558,
+      "step": 4600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0748360056203565e-05,
+      "loss": 1.0393,
+      "step": 4601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.074452340805448e-05,
+      "loss": 1.1056,
+      "step": 4602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.07406861452453e-05,
+      "loss": 1.0585,
+      "step": 4603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0736848267925856e-05,
+      "loss": 1.1062,
+      "step": 4604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.073300977624594e-05,
+      "loss": 1.1264,
+      "step": 4605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.072917067035543e-05,
+      "loss": 1.2468,
+      "step": 4606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0725330950404194e-05,
+      "loss": 1.3038,
+      "step": 4607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0721490616542135e-05,
+      "loss": 1.2622,
+      "step": 4608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0717649668919176e-05,
+      "loss": 1.3529,
+      "step": 4609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0713808107685276e-05,
+      "loss": 1.2249,
+      "step": 4610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.07099659329904e-05,
+      "loss": 1.3387,
+      "step": 4611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.070612314498453e-05,
+      "loss": 1.2577,
+      "step": 4612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0702279743817714e-05,
+      "loss": 1.2659,
+      "step": 4613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0698435729639994e-05,
+      "loss": 1.3051,
+      "step": 4614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.069459110260142e-05,
+      "loss": 1.251,
+      "step": 4615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0690745862852104e-05,
+      "loss": 1.0951,
+      "step": 4616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.068690001054214e-05,
+      "loss": 1.1057,
+      "step": 4617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.06830535458217e-05,
+      "loss": 1.0337,
+      "step": 4618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.067920646884092e-05,
+      "loss": 1.1295,
+      "step": 4619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.067535877975e-05,
+      "loss": 0.9874,
+      "step": 4620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.067151047869915e-05,
+      "loss": 1.2736,
+      "step": 4621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.066766156583861e-05,
+      "loss": 1.2584,
+      "step": 4622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.066381204131864e-05,
+      "loss": 1.1613,
+      "step": 4623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.065996190528952e-05,
+      "loss": 1.1217,
+      "step": 4624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0656111157901567e-05,
+      "loss": 1.2631,
+      "step": 4625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.065225979930509e-05,
+      "loss": 1.166,
+      "step": 4626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.064840782965047e-05,
+      "loss": 1.0917,
+      "step": 4627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0644555249088074e-05,
+      "loss": 1.1476,
+      "step": 4628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.064070205776831e-05,
+      "loss": 1.0911,
+      "step": 4629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.06368482558416e-05,
+      "loss": 1.1493,
+      "step": 4630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0632993843458403e-05,
+      "loss": 1.1275,
+      "step": 4631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.062913882076918e-05,
+      "loss": 1.1311,
+      "step": 4632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0625283187924436e-05,
+      "loss": 1.17,
+      "step": 4633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.06214269450747e-05,
+      "loss": 1.0812,
+      "step": 4634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0617570092370515e-05,
+      "loss": 1.1174,
+      "step": 4635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.061371262996245e-05,
+      "loss": 1.1437,
+      "step": 4636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0609854558001094e-05,
+      "loss": 1.1494,
+      "step": 4637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.060599587663707e-05,
+      "loss": 1.2619,
+      "step": 4638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.060213658602102e-05,
+      "loss": 1.2555,
+      "step": 4639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.05982766863036e-05,
+      "loss": 1.3097,
+      "step": 4640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0594416177635524e-05,
+      "loss": 1.2662,
+      "step": 4641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.059055506016748e-05,
+      "loss": 1.1911,
+      "step": 4642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.058669333405021e-05,
+      "loss": 1.2553,
+      "step": 4643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.058283099943447e-05,
+      "loss": 1.1961,
+      "step": 4644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0578968056471054e-05,
+      "loss": 1.2525,
+      "step": 4645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0575104505310774e-05,
+      "loss": 1.2456,
+      "step": 4646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0571240346104443e-05,
+      "loss": 1.1766,
+      "step": 4647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.056737557900293e-05,
+      "loss": 1.2916,
+      "step": 4648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.056351020415712e-05,
+      "loss": 1.2328,
+      "step": 4649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.05596442217179e-05,
+      "loss": 1.2594,
+      "step": 4650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.05557776318362e-05,
+      "loss": 1.209,
+      "step": 4651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0551910434662974e-05,
+      "loss": 1.2393,
+      "step": 4652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.05480426303492e-05,
+      "loss": 1.2842,
+      "step": 4653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0544174219045864e-05,
+      "loss": 1.3426,
+      "step": 4654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.054030520090399e-05,
+      "loss": 1.2611,
+      "step": 4655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0536435576074645e-05,
+      "loss": 1.2968,
+      "step": 4656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0532565344708863e-05,
+      "loss": 1.3162,
+      "step": 4657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.052869450695776e-05,
+      "loss": 1.3289,
+      "step": 4658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0524823062972446e-05,
+      "loss": 1.3152,
+      "step": 4659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.052095101290406e-05,
+      "loss": 1.2615,
+      "step": 4660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.051707835690376e-05,
+      "loss": 1.3086,
+      "step": 4661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.051320509512274e-05,
+      "loss": 1.2245,
+      "step": 4662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.050933122771221e-05,
+      "loss": 1.2442,
+      "step": 4663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.05054567548234e-05,
+      "loss": 1.2827,
+      "step": 4664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.050158167660757e-05,
+      "loss": 1.2407,
+      "step": 4665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0497705993216005e-05,
+      "loss": 1.2982,
+      "step": 4666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.049382970480001e-05,
+      "loss": 1.0987,
+      "step": 4667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.048995281151091e-05,
+      "loss": 1.1396,
+      "step": 4668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.048607531350006e-05,
+      "loss": 1.2211,
+      "step": 4669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.048219721091883e-05,
+      "loss": 1.1408,
+      "step": 4670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.047831850391862e-05,
+      "loss": 1.1189,
+      "step": 4671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.047443919265087e-05,
+      "loss": 1.1022,
+      "step": 4672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0470559277267006e-05,
+      "loss": 1.2808,
+      "step": 4673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0466678757918505e-05,
+      "loss": 1.1798,
+      "step": 4674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.046279763475687e-05,
+      "loss": 1.1761,
+      "step": 4675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.045891590793361e-05,
+      "loss": 1.2396,
+      "step": 4676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0455033577600266e-05,
+      "loss": 1.3847,
+      "step": 4677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0451150643908395e-05,
+      "loss": 1.2842,
+      "step": 4678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0447267107009614e-05,
+      "loss": 1.2627,
+      "step": 4679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.04433829670555e-05,
+      "loss": 1.2707,
+      "step": 4680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.043949822419771e-05,
+      "loss": 1.2463,
+      "step": 4681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0435612878587895e-05,
+      "loss": 1.2783,
+      "step": 4682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0431726930377736e-05,
+      "loss": 1.2154,
+      "step": 4683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0427840379718936e-05,
+      "loss": 1.3252,
+      "step": 4684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0423953226763245e-05,
+      "loss": 1.1967,
+      "step": 4685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.042006547166239e-05,
+      "loss": 1.1925,
+      "step": 4686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.041617711456816e-05,
+      "loss": 1.1014,
+      "step": 4687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.041228815563235e-05,
+      "loss": 1.2528,
+      "step": 4688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.040839859500679e-05,
+      "loss": 1.1981,
+      "step": 4689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.040450843284333e-05,
+      "loss": 1.1598,
+      "step": 4690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.040061766929383e-05,
+      "loss": 1.2529,
+      "step": 4691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.039672630451018e-05,
+      "loss": 1.27,
+      "step": 4692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.039283433864431e-05,
+      "loss": 1.2234,
+      "step": 4693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.038894177184816e-05,
+      "loss": 1.1541,
+      "step": 4694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0385048604273696e-05,
+      "loss": 1.2246,
+      "step": 4695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.038115483607289e-05,
+      "loss": 1.2651,
+      "step": 4696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.037726046739776e-05,
+      "loss": 1.2512,
+      "step": 4697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.037336549840036e-05,
+      "loss": 1.2776,
+      "step": 4698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.036946992923272e-05,
+      "loss": 1.2528,
+      "step": 4699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.036557376004694e-05,
+      "loss": 1.2309,
+      "step": 4700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.036167699099511e-05,
+      "loss": 1.2639,
+      "step": 4701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.035777962222938e-05,
+      "loss": 1.2658,
+      "step": 4702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0353881653901874e-05,
+      "loss": 1.1773,
+      "step": 4703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.034998308616479e-05,
+      "loss": 1.2879,
+      "step": 4704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.034608391917032e-05,
+      "loss": 1.1441,
+      "step": 4705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.034218415307068e-05,
+      "loss": 1.2539,
+      "step": 4706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.033828378801812e-05,
+      "loss": 1.1793,
+      "step": 4707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0334382824164916e-05,
+      "loss": 1.153,
+      "step": 4708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.033048126166334e-05,
+      "loss": 1.2708,
+      "step": 4709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.032657910066573e-05,
+      "loss": 1.0962,
+      "step": 4710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0322676341324415e-05,
+      "loss": 1.1212,
+      "step": 4711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.031877298379175e-05,
+      "loss": 1.2117,
+      "step": 4712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0314869028220134e-05,
+      "loss": 1.3178,
+      "step": 4713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.031096447476197e-05,
+      "loss": 1.2292,
+      "step": 4714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.030705932356969e-05,
+      "loss": 0.9002,
+      "step": 4715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.030315357479575e-05,
+      "loss": 0.9383,
+      "step": 4716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.029924722859262e-05,
+      "loss": 0.9434,
+      "step": 4717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.029534028511282e-05,
+      "loss": 1.1683,
+      "step": 4718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.029143274450886e-05,
+      "loss": 0.9371,
+      "step": 4719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.02875246069333e-05,
+      "loss": 0.8821,
+      "step": 4720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0283615872538704e-05,
+      "loss": 0.9642,
+      "step": 4721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.027970654147767e-05,
+      "loss": 0.942,
+      "step": 4722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.027579661390283e-05,
+      "loss": 0.9098,
+      "step": 4723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0271886089966796e-05,
+      "loss": 0.9038,
+      "step": 4724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.026797496982226e-05,
+      "loss": 0.9836,
+      "step": 4725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0264063253621903e-05,
+      "loss": 0.9261,
+      "step": 4726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.026015094151842e-05,
+      "loss": 1.2221,
+      "step": 4727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0256238033664586e-05,
+      "loss": 1.2649,
+      "step": 4728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.025232453021313e-05,
+      "loss": 1.2232,
+      "step": 4729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.024841043131683e-05,
+      "loss": 1.3572,
+      "step": 4730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0244495737128495e-05,
+      "loss": 1.2105,
+      "step": 4731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.024058044780097e-05,
+      "loss": 1.1695,
+      "step": 4732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.023666456348708e-05,
+      "loss": 1.1941,
+      "step": 4733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0232748084339724e-05,
+      "loss": 1.2229,
+      "step": 4734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.022883101051179e-05,
+      "loss": 1.2435,
+      "step": 4735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.02249133421562e-05,
+      "loss": 1.3033,
+      "step": 4736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0220995079425886e-05,
+      "loss": 1.2311,
+      "step": 4737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.021707622247384e-05,
+      "loss": 1.2196,
+      "step": 4738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0213156771453034e-05,
+      "loss": 1.246,
+      "step": 4739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.020923672651648e-05,
+      "loss": 1.2242,
+      "step": 4740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.020531608781722e-05,
+      "loss": 1.2257,
+      "step": 4741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0201394855508326e-05,
+      "loss": 1.1475,
+      "step": 4742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.019747302974287e-05,
+      "loss": 1.2517,
+      "step": 4743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.019355061067396e-05,
+      "loss": 1.2218,
+      "step": 4744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.018962759845472e-05,
+      "loss": 1.2175,
+      "step": 4745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0185703993238314e-05,
+      "loss": 1.2348,
+      "step": 4746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.018177979517791e-05,
+      "loss": 1.2648,
+      "step": 4747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.017785500442671e-05,
+      "loss": 1.2781,
+      "step": 4748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.017392962113794e-05,
+      "loss": 1.2396,
+      "step": 4749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.017000364546484e-05,
+      "loss": 1.2992,
+      "step": 4750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.016607707756068e-05,
+      "loss": 1.2695,
+      "step": 4751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0162149917578744e-05,
+      "loss": 1.237,
+      "step": 4752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0158222165672365e-05,
+      "loss": 1.2547,
+      "step": 4753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.015429382199487e-05,
+      "loss": 1.3001,
+      "step": 4754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.015036488669961e-05,
+      "loss": 1.2485,
+      "step": 4755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0146435359939985e-05,
+      "loss": 1.3055,
+      "step": 4756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.01425052418694e-05,
+      "loss": 1.2731,
+      "step": 4757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0138574532641284e-05,
+      "loss": 1.202,
+      "step": 4758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.013464323240908e-05,
+      "loss": 1.3223,
+      "step": 4759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.013071134132628e-05,
+      "loss": 1.2405,
+      "step": 4760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.012677885954638e-05,
+      "loss": 1.1969,
+      "step": 4761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0122845787222885e-05,
+      "loss": 1.2479,
+      "step": 4762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.011891212450937e-05,
+      "loss": 1.2526,
+      "step": 4763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.011497787155938e-05,
+      "loss": 1.2355,
+      "step": 4764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.011104302852651e-05,
+      "loss": 1.2691,
+      "step": 4765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.010710759556438e-05,
+      "loss": 1.2732,
+      "step": 4766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.010317157282664e-05,
+      "loss": 1.2386,
+      "step": 4767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0099234960466926e-05,
+      "loss": 1.3,
+      "step": 4768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.009529775863894e-05,
+      "loss": 1.3683,
+      "step": 4769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.009135996749639e-05,
+      "loss": 1.2443,
+      "step": 4770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.008742158719298e-05,
+      "loss": 1.2516,
+      "step": 4771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0083482617882496e-05,
+      "loss": 1.2833,
+      "step": 4772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.00795430597187e-05,
+      "loss": 1.3583,
+      "step": 4773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.007560291285539e-05,
+      "loss": 1.1994,
+      "step": 4774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0071662177446376e-05,
+      "loss": 1.2515,
+      "step": 4775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.006772085364553e-05,
+      "loss": 1.2251,
+      "step": 4776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.00637789416067e-05,
+      "loss": 1.2422,
+      "step": 4777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.005983644148378e-05,
+      "loss": 1.232,
+      "step": 4778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.005589335343068e-05,
+      "loss": 1.3037,
+      "step": 4779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.005194967760135e-05,
+      "loss": 1.2701,
+      "step": 4780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.004800541414973e-05,
+      "loss": 1.2223,
+      "step": 4781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0044060563229824e-05,
+      "loss": 1.1872,
+      "step": 4782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0040115124995626e-05,
+      "loss": 1.2037,
+      "step": 4783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0036169099601164e-05,
+      "loss": 1.1353,
+      "step": 4784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.003222248720049e-05,
+      "loss": 1.1335,
+      "step": 4785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.002827528794768e-05,
+      "loss": 1.1937,
+      "step": 4786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.002432750199683e-05,
+      "loss": 1.2136,
+      "step": 4787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0020379129502064e-05,
+      "loss": 1.2117,
+      "step": 4788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0016430170617515e-05,
+      "loss": 1.1407,
+      "step": 4789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.001248062549735e-05,
+      "loss": 1.1737,
+      "step": 4790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.000853049429577e-05,
+      "loss": 1.1246,
+      "step": 4791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.000457977716698e-05,
+      "loss": 1.1259,
+      "step": 4792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0000628474265214e-05,
+      "loss": 1.2395,
+      "step": 4793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999667658574473e-05,
+      "loss": 1.3137,
+      "step": 4794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9992724111759804e-05,
+      "loss": 1.2756,
+      "step": 4795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9988771052464746e-05,
+      "loss": 1.2951,
+      "step": 4796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.998481740801388e-05,
+      "loss": 1.1746,
+      "step": 4797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9980863178561546e-05,
+      "loss": 1.1893,
+      "step": 4798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.997690836426212e-05,
+      "loss": 1.239,
+      "step": 4799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9972952965270006e-05,
+      "loss": 1.2239,
+      "step": 4800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.996899698173962e-05,
+      "loss": 1.2315,
+      "step": 4801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9965040413825384e-05,
+      "loss": 1.2006,
+      "step": 4802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9961083261681775e-05,
+      "loss": 1.1896,
+      "step": 4803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.995712552546328e-05,
+      "loss": 1.1899,
+      "step": 4804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.995316720532441e-05,
+      "loss": 1.2936,
+      "step": 4805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9949208301419685e-05,
+      "loss": 1.2056,
+      "step": 4806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9945248813903655e-05,
+      "loss": 1.4153,
+      "step": 4807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.994128874293093e-05,
+      "loss": 1.2403,
+      "step": 4808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.993732808865607e-05,
+      "loss": 1.1481,
+      "step": 4809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.993336685123371e-05,
+      "loss": 1.2777,
+      "step": 4810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.992940503081851e-05,
+      "loss": 1.3636,
+      "step": 4811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9925442627565115e-05,
+      "loss": 1.1456,
+      "step": 4812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.992147964162823e-05,
+      "loss": 1.0364,
+      "step": 4813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9917516073162575e-05,
+      "loss": 0.9934,
+      "step": 4814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9913551922322876e-05,
+      "loss": 1.0422,
+      "step": 4815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.990958718926389e-05,
+      "loss": 1.0787,
+      "step": 4816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.99056218741404e-05,
+      "loss": 1.0546,
+      "step": 4817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.990165597710721e-05,
+      "loss": 0.9987,
+      "step": 4818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.989768949831915e-05,
+      "loss": 0.9842,
+      "step": 4819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.989372243793107e-05,
+      "loss": 1.0596,
+      "step": 4820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.988975479609785e-05,
+      "loss": 1.0345,
+      "step": 4821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.988578657297436e-05,
+      "loss": 0.9203,
+      "step": 4822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9881817768715544e-05,
+      "loss": 1.3348,
+      "step": 4823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.987784838347633e-05,
+      "loss": 1.2308,
+      "step": 4824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.987387841741169e-05,
+      "loss": 1.269,
+      "step": 4825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.986990787067659e-05,
+      "loss": 1.3078,
+      "step": 4826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.986593674342607e-05,
+      "loss": 1.2457,
+      "step": 4827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.986196503581513e-05,
+      "loss": 1.2225,
+      "step": 4828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.985799274799885e-05,
+      "loss": 1.3493,
+      "step": 4829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9854019880132285e-05,
+      "loss": 1.2822,
+      "step": 4830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9850046432370555e-05,
+      "loss": 1.2488,
+      "step": 4831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.984607240486875e-05,
+      "loss": 1.2807,
+      "step": 4832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9842097797782055e-05,
+      "loss": 1.3284,
+      "step": 4833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9838122611265617e-05,
+      "loss": 1.2842,
+      "step": 4834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9834146845474614e-05,
+      "loss": 1.2936,
+      "step": 4835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.983017050056428e-05,
+      "loss": 1.3356,
+      "step": 4836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.982619357668983e-05,
+      "loss": 1.2892,
+      "step": 4837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.982221607400653e-05,
+      "loss": 1.2959,
+      "step": 4838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9818237992669684e-05,
+      "loss": 1.3557,
+      "step": 4839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.981425933283456e-05,
+      "loss": 1.3216,
+      "step": 4840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9810280094656494e-05,
+      "loss": 1.1163,
+      "step": 4841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.980630027829083e-05,
+      "loss": 1.1853,
+      "step": 4842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.980231988389295e-05,
+      "loss": 1.304,
+      "step": 4843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9798338911618246e-05,
+      "loss": 1.1877,
+      "step": 4844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9794357361622125e-05,
+      "loss": 1.2758,
+      "step": 4845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.979037523406003e-05,
+      "loss": 1.2641,
+      "step": 4846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.978639252908742e-05,
+      "loss": 1.1637,
+      "step": 4847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.978240924685979e-05,
+      "loss": 1.1681,
+      "step": 4848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.977842538753262e-05,
+      "loss": 1.1726,
+      "step": 4849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.977444095126146e-05,
+      "loss": 1.2755,
+      "step": 4850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.977045593820185e-05,
+      "loss": 1.2025,
+      "step": 4851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.976647034850938e-05,
+      "loss": 1.162,
+      "step": 4852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9762484182339624e-05,
+      "loss": 1.2077,
+      "step": 4853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9758497439848205e-05,
+      "loss": 1.1137,
+      "step": 4854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.975451012119078e-05,
+      "loss": 1.2087,
+      "step": 4855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.975052222652299e-05,
+      "loss": 1.1959,
+      "step": 4856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9746533756000534e-05,
+      "loss": 1.1274,
+      "step": 4857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.974254470977913e-05,
+      "loss": 1.1446,
+      "step": 4858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9738555088014485e-05,
+      "loss": 1.1417,
+      "step": 4859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.973456489086237e-05,
+      "loss": 1.2149,
+      "step": 4860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.973057411847856e-05,
+      "loss": 1.2492,
+      "step": 4861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.972658277101884e-05,
+      "loss": 1.1843,
+      "step": 4862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.972259084863904e-05,
+      "loss": 1.1219,
+      "step": 4863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.971859835149501e-05,
+      "loss": 1.2014,
+      "step": 4864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.971460527974259e-05,
+      "loss": 1.2543,
+      "step": 4865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9710611633537706e-05,
+      "loss": 1.3077,
+      "step": 4866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.970661741303623e-05,
+      "loss": 1.2652,
+      "step": 4867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9702622618394125e-05,
+      "loss": 1.2172,
+      "step": 4868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.969862724976733e-05,
+      "loss": 1.3605,
+      "step": 4869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.969463130731183e-05,
+      "loss": 1.4705,
+      "step": 4870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.969063479118362e-05,
+      "loss": 1.3514,
+      "step": 4871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.968663770153872e-05,
+      "loss": 1.4796,
+      "step": 4872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9682640038533183e-05,
+      "loss": 1.4161,
+      "step": 4873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.967864180232308e-05,
+      "loss": 1.4125,
+      "step": 4874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.967464299306448e-05,
+      "loss": 1.4742,
+      "step": 4875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.967064361091352e-05,
+      "loss": 1.4608,
+      "step": 4876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9666643656026315e-05,
+      "loss": 1.4621,
+      "step": 4877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.966264312855903e-05,
+      "loss": 1.4807,
+      "step": 4878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.965864202866784e-05,
+      "loss": 1.4804,
+      "step": 4879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.965464035650895e-05,
+      "loss": 0.8801,
+      "step": 4880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.965063811223859e-05,
+      "loss": 0.7844,
+      "step": 4881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9646635296012994e-05,
+      "loss": 0.9179,
+      "step": 4882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9642631907988423e-05,
+      "loss": 0.7913,
+      "step": 4883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.96386279483212e-05,
+      "loss": 0.8621,
+      "step": 4884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9634623417167614e-05,
+      "loss": 0.8281,
+      "step": 4885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9630618314684e-05,
+      "loss": 0.9126,
+      "step": 4886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.962661264102672e-05,
+      "loss": 0.8773,
+      "step": 4887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.962260639635216e-05,
+      "loss": 0.7842,
+      "step": 4888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9618599580816703e-05,
+      "loss": 0.7744,
+      "step": 4889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.961459219457681e-05,
+      "loss": 0.8653,
+      "step": 4890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.961058423778889e-05,
+      "loss": 0.8001,
+      "step": 4891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9606575710609414e-05,
+      "loss": 0.8502,
+      "step": 4892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9602566613194906e-05,
+      "loss": 0.8542,
+      "step": 4893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9598556945701854e-05,
+      "loss": 1.3009,
+      "step": 4894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9594546708286803e-05,
+      "loss": 1.2409,
+      "step": 4895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.95905359011063e-05,
+      "loss": 1.2595,
+      "step": 4896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9586524524316935e-05,
+      "loss": 1.2516,
+      "step": 4897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9582512578075316e-05,
+      "loss": 1.2643,
+      "step": 4898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9578500062538056e-05,
+      "loss": 1.2273,
+      "step": 4899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.95744869778618e-05,
+      "loss": 1.2359,
+      "step": 4900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.957047332420324e-05,
+      "loss": 1.3669,
+      "step": 4901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.956645910171903e-05,
+      "loss": 1.1918,
+      "step": 4902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9562444310565924e-05,
+      "loss": 1.0901,
+      "step": 4903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.955842895090063e-05,
+      "loss": 1.1721,
+      "step": 4904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9554413022879916e-05,
+      "loss": 1.0645,
+      "step": 4905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.955039652666056e-05,
+      "loss": 1.0354,
+      "step": 4906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9546379462399366e-05,
+      "loss": 1.0892,
+      "step": 4907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.954236183025316e-05,
+      "loss": 1.1532,
+      "step": 4908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.953834363037879e-05,
+      "loss": 1.0286,
+      "step": 4909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.953432486293312e-05,
+      "loss": 1.0029,
+      "step": 4910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.953030552807304e-05,
+      "loss": 1.1459,
+      "step": 4911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.952628562595547e-05,
+      "loss": 1.0721,
+      "step": 4912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9522265156737345e-05,
+      "loss": 1.2381,
+      "step": 4913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.951824412057561e-05,
+      "loss": 1.2025,
+      "step": 4914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9514222517627264e-05,
+      "loss": 1.2385,
+      "step": 4915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.95102003480493e-05,
+      "loss": 1.2019,
+      "step": 4916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.950617761199873e-05,
+      "loss": 1.2614,
+      "step": 4917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.950215430963261e-05,
+      "loss": 1.1389,
+      "step": 4918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.949813044110802e-05,
+      "loss": 1.1842,
+      "step": 4919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.949410600658204e-05,
+      "loss": 1.23,
+      "step": 4920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.949008100621178e-05,
+      "loss": 1.1616,
+      "step": 4921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.948605544015437e-05,
+      "loss": 1.2293,
+      "step": 4922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.948202930856697e-05,
+      "loss": 1.1663,
+      "step": 4923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.947800261160677e-05,
+      "loss": 1.1656,
+      "step": 4924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.947397534943096e-05,
+      "loss": 1.1978,
+      "step": 4925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.946994752219676e-05,
+      "loss": 1.1861,
+      "step": 4926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.946591913006143e-05,
+      "loss": 1.0623,
+      "step": 4927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.946189017318222e-05,
+      "loss": 1.1026,
+      "step": 4928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9457860651716424e-05,
+      "loss": 1.1597,
+      "step": 4929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9453830565821357e-05,
+      "loss": 1.1767,
+      "step": 4930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.944979991565435e-05,
+      "loss": 1.0849,
+      "step": 4931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.944576870137276e-05,
+      "loss": 1.2712,
+      "step": 4932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9441736923133956e-05,
+      "loss": 1.2817,
+      "step": 4933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.943770458109535e-05,
+      "loss": 1.2354,
+      "step": 4934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9433671675414344e-05,
+      "loss": 1.261,
+      "step": 4935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.94296382062484e-05,
+      "loss": 1.2492,
+      "step": 4936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9425604173754974e-05,
+      "loss": 1.2512,
+      "step": 4937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9421569578091554e-05,
+      "loss": 1.2083,
+      "step": 4938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9417534419415655e-05,
+      "loss": 1.1042,
+      "step": 4939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9413498697884797e-05,
+      "loss": 1.2177,
+      "step": 4940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.940946241365654e-05,
+      "loss": 1.3366,
+      "step": 4941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.940542556688846e-05,
+      "loss": 1.3086,
+      "step": 4942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.940138815773815e-05,
+      "loss": 1.2066,
+      "step": 4943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9397350186363233e-05,
+      "loss": 1.3321,
+      "step": 4944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.939331165292135e-05,
+      "loss": 1.363,
+      "step": 4945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.938927255757015e-05,
+      "loss": 1.2409,
+      "step": 4946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.938523290046734e-05,
+      "loss": 1.4497,
+      "step": 4947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.938119268177061e-05,
+      "loss": 1.3727,
+      "step": 4948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.93771519016377e-05,
+      "loss": 1.4236,
+      "step": 4949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.937311056022634e-05,
+      "loss": 1.3552,
+      "step": 4950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.936906865769434e-05,
+      "loss": 1.2366,
+      "step": 4951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.936502619419945e-05,
+      "loss": 1.1337,
+      "step": 4952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.936098316989952e-05,
+      "loss": 1.2577,
+      "step": 4953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.935693958495238e-05,
+      "loss": 1.1888,
+      "step": 4954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9352895439515876e-05,
+      "loss": 1.1588,
+      "step": 4955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9348850733747896e-05,
+      "loss": 1.2278,
+      "step": 4956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.934480546780635e-05,
+      "loss": 1.2355,
+      "step": 4957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.934075964184917e-05,
+      "loss": 1.2565,
+      "step": 4958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.933671325603429e-05,
+      "loss": 1.204,
+      "step": 4959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.933266631051968e-05,
+      "loss": 1.1781,
+      "step": 4960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.932861880546334e-05,
+      "loss": 1.2492,
+      "step": 4961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9324570741023275e-05,
+      "loss": 1.2849,
+      "step": 4962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.932052211735753e-05,
+      "loss": 1.295,
+      "step": 4963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.931647293462414e-05,
+      "loss": 1.3201,
+      "step": 4964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9312423192981207e-05,
+      "loss": 1.2571,
+      "step": 4965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.930837289258683e-05,
+      "loss": 1.3315,
+      "step": 4966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.93043220335991e-05,
+      "loss": 1.3699,
+      "step": 4967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.93002706161762e-05,
+      "loss": 1.2548,
+      "step": 4968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.929621864047629e-05,
+      "loss": 1.2947,
+      "step": 4969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.929216610665754e-05,
+      "loss": 1.2717,
+      "step": 4970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.928811301487816e-05,
+      "loss": 1.3385,
+      "step": 4971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9284059365296396e-05,
+      "loss": 1.3133,
+      "step": 4972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.928000515807049e-05,
+      "loss": 1.0218,
+      "step": 4973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.927595039335872e-05,
+      "loss": 1.0873,
+      "step": 4974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.927189507131938e-05,
+      "loss": 1.099,
+      "step": 4975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.92678391921108e-05,
+      "loss": 1.0348,
+      "step": 4976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.92637827558913e-05,
+      "loss": 1.0141,
+      "step": 4977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.925972576281925e-05,
+      "loss": 1.1175,
+      "step": 4978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.925566821305304e-05,
+      "loss": 1.092,
+      "step": 4979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.925161010675108e-05,
+      "loss": 1.0322,
+      "step": 4980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.924755144407178e-05,
+      "loss": 1.0825,
+      "step": 4981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9243492225173595e-05,
+      "loss": 1.1412,
+      "step": 4982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9239432450214996e-05,
+      "loss": 1.0878,
+      "step": 4983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.923537211935448e-05,
+      "loss": 1.1414,
+      "step": 4984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9231311232750554e-05,
+      "loss": 1.0901,
+      "step": 4985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.922724979056176e-05,
+      "loss": 1.2483,
+      "step": 4986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9223187792946646e-05,
+      "loss": 1.2873,
+      "step": 4987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9219125240063795e-05,
+      "loss": 1.2601,
+      "step": 4988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.921506213207181e-05,
+      "loss": 1.2332,
+      "step": 4989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.921099846912931e-05,
+      "loss": 1.2197,
+      "step": 4990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9206934251394946e-05,
+      "loss": 1.1963,
+      "step": 4991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.920286947902737e-05,
+      "loss": 1.2106,
+      "step": 4992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9198804152185284e-05,
+      "loss": 1.2266,
+      "step": 4993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9194738271027385e-05,
+      "loss": 1.2164,
+      "step": 4994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.919067183571241e-05,
+      "loss": 1.2867,
+      "step": 4995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.918660484639911e-05,
+      "loss": 1.3557,
+      "step": 4996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.918253730324626e-05,
+      "loss": 1.2142,
+      "step": 4997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.917846920641265e-05,
+      "loss": 1.2109,
+      "step": 4998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.91744005560571e-05,
+      "loss": 1.2663,
+      "step": 4999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.917033135233845e-05,
+      "loss": 1.2606,
+      "step": 5000
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.1719191074371338,
+      "eval_runtime": 114.584,
+      "eval_samples_per_second": 33.347,
+      "eval_steps_per_second": 0.349,
+      "step": 5000
+    }
+  ],
+  "max_steps": 16000,
+  "num_train_epochs": 9223372036854775807,
+  "total_flos": 6.01482923606016e+18,
+  "trial_name": null,
+  "trial_params": null
+}