diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6235 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 1035,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 5.9375,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 5.9727,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 5.8477,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 5.875,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.884615384615385e-06,
+      "loss": 5.8477,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 5.7422,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 5.918,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 5.7539,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.192307692307692e-06,
+      "loss": 5.7891,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.76923076923077e-06,
+      "loss": 5.6055,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 5.6523,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 5.6758,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.5e-06,
+      "loss": 5.3789,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 5.3594,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.653846153846153e-06,
+      "loss": 5.0781,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 5.0273,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 4.9609,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0384615384615384e-05,
+      "loss": 4.6562,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0961538461538462e-05,
+      "loss": 4.8047,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 4.7734,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2115384615384615e-05,
+      "loss": 4.3438,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 4.4531,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3269230769230769e-05,
+      "loss": 4.8125,
+      "step": 23
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 4.2539,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 4.4336,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5e-05,
+      "loss": 3.9082,
+      "step": 26
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.557692307692308e-05,
+      "loss": 3.9883,
+      "step": 27
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 4.293,
+      "step": 28
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.673076923076923e-05,
+      "loss": 3.9648,
+      "step": 29
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7307692307692306e-05,
+      "loss": 3.5488,
+      "step": 30
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7884615384615384e-05,
+      "loss": 3.9082,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 4.2891,
+      "step": 32
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.903846153846154e-05,
+      "loss": 3.3262,
+      "step": 33
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 2.9336,
+      "step": 34
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0192307692307694e-05,
+      "loss": 2.709,
+      "step": 35
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.076923076923077e-05,
+      "loss": 3.2617,
+      "step": 36
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1346153846153846e-05,
+      "loss": 2.2539,
+      "step": 37
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1923076923076924e-05,
+      "loss": 2.6367,
+      "step": 38
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.25e-05,
+      "loss": 2.4766,
+      "step": 39
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 2.7012,
+      "step": 40
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3653846153846153e-05,
+      "loss": 1.6719,
+      "step": 41
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.423076923076923e-05,
+      "loss": 2.0742,
+      "step": 42
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.4807692307692305e-05,
+      "loss": 1.9346,
+      "step": 43
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5384615384615386e-05,
+      "loss": 1.8262,
+      "step": 44
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5961538461538464e-05,
+      "loss": 2.4004,
+      "step": 45
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6538461538461538e-05,
+      "loss": 2.252,
+      "step": 46
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7115384615384616e-05,
+      "loss": 1.9297,
+      "step": 47
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 3.1191,
+      "step": 48
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8269230769230768e-05,
+      "loss": 2.5312,
+      "step": 49
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.884615384615385e-05,
+      "loss": 1.4062,
+      "step": 50
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9423076923076923e-05,
+      "loss": 1.6602,
+      "step": 51
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3e-05,
+      "loss": 2.168,
+      "step": 52
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.057692307692308e-05,
+      "loss": 2.4062,
+      "step": 53
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.115384615384616e-05,
+      "loss": 2.0215,
+      "step": 54
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1730769230769234e-05,
+      "loss": 1.752,
+      "step": 55
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.230769230769231e-05,
+      "loss": 1.8105,
+      "step": 56
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.288461538461539e-05,
+      "loss": 1.4023,
+      "step": 57
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.346153846153846e-05,
+      "loss": 1.5723,
+      "step": 58
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.403846153846154e-05,
+      "loss": 2.3945,
+      "step": 59
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.461538461538461e-05,
+      "loss": 1.1475,
+      "step": 60
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.519230769230769e-05,
+      "loss": 1.2188,
+      "step": 61
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.576923076923077e-05,
+      "loss": 1.4512,
+      "step": 62
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.634615384615384e-05,
+      "loss": 1.5293,
+      "step": 63
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.692307692307693e-05,
+      "loss": 1.8613,
+      "step": 64
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.9375,
+      "step": 65
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.807692307692308e-05,
+      "loss": 2.5215,
+      "step": 66
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.865384615384616e-05,
+      "loss": 1.6035,
+      "step": 67
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.923076923076923e-05,
+      "loss": 1.8506,
+      "step": 68
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.980769230769231e-05,
+      "loss": 1.4258,
+      "step": 69
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.038461538461539e-05,
+      "loss": 1.1777,
+      "step": 70
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.096153846153846e-05,
+      "loss": 2.127,
+      "step": 71
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.153846153846154e-05,
+      "loss": 2.2969,
+      "step": 72
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.211538461538461e-05,
+      "loss": 1.6338,
+      "step": 73
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.269230769230769e-05,
+      "loss": 1.9004,
+      "step": 74
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.3269230769230766e-05,
+      "loss": 1.5615,
+      "step": 75
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.384615384615385e-05,
+      "loss": 2.1914,
+      "step": 76
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.442307692307693e-05,
+      "loss": 1.5537,
+      "step": 77
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5e-05,
+      "loss": 0.98,
+      "step": 78
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.557692307692308e-05,
+      "loss": 1.8662,
+      "step": 79
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 1.4648,
+      "step": 80
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.673076923076923e-05,
+      "loss": 0.8438,
+      "step": 81
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.730769230769231e-05,
+      "loss": 2.1777,
+      "step": 82
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.788461538461539e-05,
+      "loss": 1.8066,
+      "step": 83
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.846153846153846e-05,
+      "loss": 0.9365,
+      "step": 84
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9038461538461536e-05,
+      "loss": 1.8799,
+      "step": 85
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.961538461538461e-05,
+      "loss": 2.0762,
+      "step": 86
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.01923076923077e-05,
+      "loss": 1.6475,
+      "step": 87
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.076923076923077e-05,
+      "loss": 1.291,
+      "step": 88
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.134615384615385e-05,
+      "loss": 1.1172,
+      "step": 89
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.192307692307693e-05,
+      "loss": 1.5508,
+      "step": 90
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.25e-05,
+      "loss": 1.4297,
+      "step": 91
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.3076923076923076e-05,
+      "loss": 1.1201,
+      "step": 92
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.365384615384616e-05,
+      "loss": 1.5566,
+      "step": 93
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.423076923076923e-05,
+      "loss": 1.3662,
+      "step": 94
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.4807692307692306e-05,
+      "loss": 0.4919,
+      "step": 95
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.538461538461539e-05,
+      "loss": 1.7178,
+      "step": 96
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.596153846153846e-05,
+      "loss": 1.8799,
+      "step": 97
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.6538461538461536e-05,
+      "loss": 1.291,
+      "step": 98
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.711538461538462e-05,
+      "loss": 1.6123,
+      "step": 99
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.76923076923077e-05,
+      "loss": 0.8276,
+      "step": 100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.826923076923077e-05,
+      "loss": 1.7314,
+      "step": 101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.8846153846153846e-05,
+      "loss": 1.7471,
+      "step": 102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.942307692307693e-05,
+      "loss": 0.7832,
+      "step": 103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6e-05,
+      "loss": 0.6196,
+      "step": 104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.993555316863588e-05,
+      "loss": 1.041,
+      "step": 105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.987110633727175e-05,
+      "loss": 2.1406,
+      "step": 106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9806659505907625e-05,
+      "loss": 1.7354,
+      "step": 107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.974221267454351e-05,
+      "loss": 1.0576,
+      "step": 108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.967776584317938e-05,
+      "loss": 2.1211,
+      "step": 109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9613319011815255e-05,
+      "loss": 0.9023,
+      "step": 110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9548872180451125e-05,
+      "loss": 3.0996,
+      "step": 111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9484425349087e-05,
+      "loss": 0.9351,
+      "step": 112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.941997851772288e-05,
+      "loss": 1.3848,
+      "step": 113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9355531686358755e-05,
+      "loss": 2.2637,
+      "step": 114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.929108485499463e-05,
+      "loss": 1.5088,
+      "step": 115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.922663802363051e-05,
+      "loss": 1.9033,
+      "step": 116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.916219119226638e-05,
+      "loss": 1.3701,
+      "step": 117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9097744360902255e-05,
+      "loss": 1.3613,
+      "step": 118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.903329752953813e-05,
+      "loss": 1.418,
+      "step": 119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.896885069817401e-05,
+      "loss": 1.0449,
+      "step": 120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8904403866809885e-05,
+      "loss": 0.811,
+      "step": 121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.883995703544576e-05,
+      "loss": 1.3867,
+      "step": 122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.877551020408163e-05,
+      "loss": 1.3105,
+      "step": 123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.871106337271751e-05,
+      "loss": 1.8262,
+      "step": 124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.864661654135338e-05,
+      "loss": 0.9292,
+      "step": 125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.858216970998926e-05,
+      "loss": 1.0,
+      "step": 126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.851772287862514e-05,
+      "loss": 1.6064,
+      "step": 127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.845327604726101e-05,
+      "loss": 1.2256,
+      "step": 128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8388829215896886e-05,
+      "loss": 0.7412,
+      "step": 129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.832438238453276e-05,
+      "loss": 0.7402,
+      "step": 130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.825993555316863e-05,
+      "loss": 1.2227,
+      "step": 131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8195488721804516e-05,
+      "loss": 1.2969,
+      "step": 132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.813104189044039e-05,
+      "loss": 1.7881,
+      "step": 133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.806659505907626e-05,
+      "loss": 1.3223,
+      "step": 134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.800214822771214e-05,
+      "loss": 0.8848,
+      "step": 135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.7937701396348016e-05,
+      "loss": 0.8428,
+      "step": 136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.7873254564983886e-05,
+      "loss": 0.8032,
+      "step": 137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.780880773361977e-05,
+      "loss": 1.1318,
+      "step": 138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.774436090225564e-05,
+      "loss": 2.0195,
+      "step": 139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.7679914070891516e-05,
+      "loss": 1.4062,
+      "step": 140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.761546723952739e-05,
+      "loss": 0.9395,
+      "step": 141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.755102040816326e-05,
+      "loss": 1.5361,
+      "step": 142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.748657357679914e-05,
+      "loss": 1.4014,
+      "step": 143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.742212674543502e-05,
+      "loss": 0.6055,
+      "step": 144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.735767991407089e-05,
+      "loss": 1.3877,
+      "step": 145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.729323308270677e-05,
+      "loss": 0.6528,
+      "step": 146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7228786251342646e-05,
+      "loss": 0.7373,
+      "step": 147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7164339419978516e-05,
+      "loss": 0.8423,
+      "step": 148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.709989258861439e-05,
+      "loss": 1.3916,
+      "step": 149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7035445757250276e-05,
+      "loss": 0.9614,
+      "step": 150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.6970998925886146e-05,
+      "loss": 1.1592,
+      "step": 151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.690655209452202e-05,
+      "loss": 0.8711,
+      "step": 152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.684210526315789e-05,
+      "loss": 1.3662,
+      "step": 153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.677765843179377e-05,
+      "loss": 0.4143,
+      "step": 154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.6713211600429646e-05,
+      "loss": 1.1768,
+      "step": 155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.6648764769065516e-05,
+      "loss": 1.291,
+      "step": 156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.65843179377014e-05,
+      "loss": 1.0244,
+      "step": 157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6519871106337277e-05,
+      "loss": 1.0244,
+      "step": 158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6455424274973147e-05,
+      "loss": 1.0889,
+      "step": 159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.639097744360902e-05,
+      "loss": 1.4307,
+      "step": 160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.63265306122449e-05,
+      "loss": 2.4375,
+      "step": 161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.626208378088077e-05,
+      "loss": 0.5796,
+      "step": 162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.619763694951665e-05,
+      "loss": 0.6992,
+      "step": 163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.613319011815253e-05,
+      "loss": 1.3711,
+      "step": 164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.60687432867884e-05,
+      "loss": 0.7881,
+      "step": 165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.600429645542428e-05,
+      "loss": 2.0098,
+      "step": 166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.593984962406015e-05,
+      "loss": 0.9419,
+      "step": 167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5875402792696023e-05,
+      "loss": 1.5947,
+      "step": 168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.581095596133191e-05,
+      "loss": 1.7793,
+      "step": 169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.574650912996778e-05,
+      "loss": 0.7998,
+      "step": 170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5682062298603654e-05,
+      "loss": 1.2207,
+      "step": 171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.561761546723953e-05,
+      "loss": 1.0303,
+      "step": 172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.55531686358754e-05,
+      "loss": 1.2314,
+      "step": 173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.548872180451128e-05,
+      "loss": 0.6343,
+      "step": 174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.542427497314716e-05,
+      "loss": 0.981,
+      "step": 175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.535982814178303e-05,
+      "loss": 1.9785,
+      "step": 176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.529538131041891e-05,
+      "loss": 1.0107,
+      "step": 177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5230934479054784e-05,
+      "loss": 0.6196,
+      "step": 178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5166487647690654e-05,
+      "loss": 1.375,
+      "step": 179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.510204081632653e-05,
+      "loss": 1.1621,
+      "step": 180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.503759398496241e-05,
+      "loss": 1.2227,
+      "step": 181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4973147153598284e-05,
+      "loss": 1.2305,
+      "step": 182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.490870032223416e-05,
+      "loss": 0.3601,
+      "step": 183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.484425349087003e-05,
+      "loss": 0.6133,
+      "step": 184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.477980665950591e-05,
+      "loss": 0.5396,
+      "step": 185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.4715359828141784e-05,
+      "loss": 0.9004,
+      "step": 186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.465091299677766e-05,
+      "loss": 1.3242,
+      "step": 187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.458646616541354e-05,
+      "loss": 0.812,
+      "step": 188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.4522019334049414e-05,
+      "loss": 1.8721,
+      "step": 189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.4457572502685284e-05,
+      "loss": 0.6016,
+      "step": 190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.439312567132116e-05,
+      "loss": 0.6221,
+      "step": 191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.432867883995704e-05,
+      "loss": 1.2168,
+      "step": 192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.4264232008592914e-05,
+      "loss": 1.8311,
+      "step": 193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.419978517722879e-05,
+      "loss": 1.2402,
+      "step": 194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.413533834586466e-05,
+      "loss": 1.415,
+      "step": 195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.407089151450054e-05,
+      "loss": 1.042,
+      "step": 196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.4006444683136414e-05,
+      "loss": 0.6216,
+      "step": 197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.3941997851772284e-05,
+      "loss": 0.7007,
+      "step": 198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.387755102040817e-05,
+      "loss": 0.5439,
+      "step": 199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.3813104189044045e-05,
+      "loss": 1.8574,
+      "step": 200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.3748657357679914e-05,
+      "loss": 1.1797,
+      "step": 201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.368421052631579e-05,
+      "loss": 1.1494,
+      "step": 202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.361976369495167e-05,
+      "loss": 1.1846,
+      "step": 203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.355531686358754e-05,
+      "loss": 0.9653,
+      "step": 204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.3490870032223415e-05,
+      "loss": 1.2695,
+      "step": 205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.34264232008593e-05,
+      "loss": 1.083,
+      "step": 206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.336197636949517e-05,
+      "loss": 2.4629,
+      "step": 207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.3297529538131045e-05,
+      "loss": 1.0176,
+      "step": 208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.3233082706766915e-05,
+      "loss": 1.6035,
+      "step": 209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.316863587540279e-05,
+      "loss": 0.9111,
+      "step": 210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.310418904403867e-05,
+      "loss": 0.6709,
+      "step": 211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.3039742212674545e-05,
+      "loss": 0.792,
+      "step": 212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.297529538131042e-05,
+      "loss": 0.3862,
+      "step": 213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.29108485499463e-05,
+      "loss": 0.3044,
+      "step": 214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.284640171858217e-05,
+      "loss": 0.7002,
+      "step": 215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.2781954887218045e-05,
+      "loss": 0.4397,
+      "step": 216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.271750805585392e-05,
+      "loss": 0.9385,
+      "step": 217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.26530612244898e-05,
+      "loss": 0.7261,
+      "step": 218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.2588614393125675e-05,
+      "loss": 0.853,
+      "step": 219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.252416756176155e-05,
+      "loss": 0.9805,
+      "step": 220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.245972073039742e-05,
+      "loss": 0.4788,
+      "step": 221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.23952738990333e-05,
+      "loss": 1.4219,
+      "step": 222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.233082706766917e-05,
+      "loss": 1.7207,
+      "step": 223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.226638023630505e-05,
+      "loss": 1.7988,
+      "step": 224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.220193340494093e-05,
+      "loss": 0.8208,
+      "step": 225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.21374865735768e-05,
+      "loss": 1.1367,
+      "step": 226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2073039742212675e-05,
+      "loss": 0.6289,
+      "step": 227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.200859291084855e-05,
+      "loss": 1.2578,
+      "step": 228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.194414607948442e-05,
+      "loss": 0.9995,
+      "step": 229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1879699248120305e-05,
+      "loss": 1.6768,
+      "step": 230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181525241675618e-05,
+      "loss": 1.3379,
+      "step": 231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175080558539205e-05,
+      "loss": 1.3545,
+      "step": 232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.168635875402793e-05,
+      "loss": 0.9443,
+      "step": 233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1621911922663806e-05,
+      "loss": 1.3398,
+      "step": 234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1557465091299675e-05,
+      "loss": 1.4756,
+      "step": 235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.149301825993556e-05,
+      "loss": 0.7642,
+      "step": 236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.142857142857143e-05,
+      "loss": 1.7676,
+      "step": 237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1364124597207306e-05,
+      "loss": 1.2217,
+      "step": 238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.129967776584318e-05,
+      "loss": 0.7295,
+      "step": 239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.123523093447905e-05,
+      "loss": 0.5938,
+      "step": 240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.117078410311493e-05,
+      "loss": 0.212,
+      "step": 241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.110633727175081e-05,
+      "loss": 1.0762,
+      "step": 242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.104189044038668e-05,
+      "loss": 0.9961,
+      "step": 243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.097744360902256e-05,
+      "loss": 0.7412,
+      "step": 244
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0912996777658436e-05,
+      "loss": 1.1523,
+      "step": 245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.0848549946294306e-05,
+      "loss": 1.7646,
+      "step": 246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.078410311493018e-05,
+      "loss": 0.4404,
+      "step": 247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.071965628356606e-05,
+      "loss": 1.2734,
+      "step": 248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.0655209452201936e-05,
+      "loss": 0.21,
+      "step": 249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.059076262083781e-05,
+      "loss": 0.7417,
+      "step": 250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.052631578947368e-05,
+      "loss": 0.8511,
+      "step": 251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.046186895810956e-05,
+      "loss": 1.2627,
+      "step": 252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0397422126745436e-05,
+      "loss": 0.4741,
+      "step": 253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0332975295381306e-05,
+      "loss": 0.5249,
+      "step": 254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.026852846401719e-05,
+      "loss": 0.2087,
+      "step": 255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0204081632653066e-05,
+      "loss": 1.501,
+      "step": 256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0139634801288936e-05,
+      "loss": 0.5781,
+      "step": 257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.007518796992481e-05,
+      "loss": 0.5508,
+      "step": 258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.001074113856069e-05,
+      "loss": 0.1661,
+      "step": 259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.994629430719656e-05,
+      "loss": 1.1377,
+      "step": 260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.988184747583244e-05,
+      "loss": 1.2031,
+      "step": 261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.981740064446831e-05,
+      "loss": 0.791,
+      "step": 262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.975295381310419e-05,
+      "loss": 1.9609,
+      "step": 263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9688506981740066e-05,
+      "loss": 2.9258,
+      "step": 264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9624060150375936e-05,
+      "loss": 0.7656,
+      "step": 265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.955961331901181e-05,
+      "loss": 1.4219,
+      "step": 266
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9495166487647697e-05,
+      "loss": 1.3262,
+      "step": 267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9430719656283567e-05,
+      "loss": 0.8892,
+      "step": 268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.936627282491944e-05,
+      "loss": 1.1738,
+      "step": 269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.930182599355532e-05,
+      "loss": 1.0635,
+      "step": 270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.923737916219119e-05,
+      "loss": 1.4307,
+      "step": 271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.917293233082707e-05,
+      "loss": 1.1562,
+      "step": 272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.910848549946295e-05,
+      "loss": 1.2158,
+      "step": 273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.904403866809882e-05,
+      "loss": 0.9302,
+      "step": 274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.89795918367347e-05,
+      "loss": 1.5469,
+      "step": 275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.891514500537057e-05,
+      "loss": 1.6465,
+      "step": 276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8850698174006443e-05,
+      "loss": 1.1904,
+      "step": 277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.878625134264232e-05,
+      "loss": 1.7793,
+      "step": 278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.87218045112782e-05,
+      "loss": 0.7744,
+      "step": 279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8657357679914074e-05,
+      "loss": 0.3062,
+      "step": 280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.859291084854995e-05,
+      "loss": 1.5322,
+      "step": 281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.852846401718582e-05,
+      "loss": 1.9072,
+      "step": 282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.84640171858217e-05,
+      "loss": 1.5801,
+      "step": 283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.8399570354457574e-05,
+      "loss": 0.9766,
+      "step": 284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.833512352309345e-05,
+      "loss": 0.8052,
+      "step": 285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.827067669172933e-05,
+      "loss": 1.1963,
+      "step": 286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8206229860365204e-05,
+      "loss": 1.0469,
+      "step": 287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8141783029001074e-05,
+      "loss": 0.4531,
+      "step": 288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.807733619763695e-05,
+      "loss": 0.5645,
+      "step": 289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.801288936627282e-05,
+      "loss": 1.5752,
+      "step": 290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7948442534908704e-05,
+      "loss": 0.3367,
+      "step": 291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.788399570354458e-05,
+      "loss": 0.8164,
+      "step": 292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.781954887218045e-05,
+      "loss": 0.6523,
+      "step": 293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.775510204081633e-05,
+      "loss": 1.0527,
+      "step": 294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7690655209452204e-05,
+      "loss": 1.1582,
+      "step": 295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7626208378088074e-05,
+      "loss": 0.853,
+      "step": 296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.756176154672396e-05,
+      "loss": 1.2627,
+      "step": 297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7497314715359834e-05,
+      "loss": 0.4578,
+      "step": 298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7432867883995704e-05,
+      "loss": 1.1797,
+      "step": 299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736842105263158e-05,
+      "loss": 2.209,
+      "step": 300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730397422126746e-05,
+      "loss": 0.7031,
+      "step": 301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.723952738990333e-05,
+      "loss": 0.7183,
+      "step": 302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7175080558539204e-05,
+      "loss": 1.2344,
+      "step": 303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.711063372717508e-05,
+      "loss": 0.9141,
+      "step": 304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.704618689581096e-05,
+      "loss": 0.6128,
+      "step": 305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6981740064446834e-05,
+      "loss": 0.3938,
+      "step": 306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6917293233082704e-05,
+      "loss": 0.522,
+      "step": 307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.685284640171858e-05,
+      "loss": 0.6768,
+      "step": 308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.678839957035446e-05,
+      "loss": 0.8828,
+      "step": 309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6723952738990334e-05,
+      "loss": 1.3838,
+      "step": 310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.665950590762621e-05,
+      "loss": 0.6724,
+      "step": 311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.659505907626209e-05,
+      "loss": 1.3271,
+      "step": 312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.653061224489796e-05,
+      "loss": 1.1445,
+      "step": 313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6466165413533835e-05,
+      "loss": 0.7754,
+      "step": 314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.640171858216971e-05,
+      "loss": 1.7266,
+      "step": 315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.633727175080559e-05,
+      "loss": 0.9507,
+      "step": 316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6272824919441465e-05,
+      "loss": 0.4897,
+      "step": 317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6208378088077335e-05,
+      "loss": 1.0723,
+      "step": 318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.614393125671321e-05,
+      "loss": 1.7969,
+      "step": 319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.607948442534909e-05,
+      "loss": 1.0342,
+      "step": 320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.601503759398496e-05,
+      "loss": 0.6543,
+      "step": 321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.595059076262084e-05,
+      "loss": 1.0791,
+      "step": 322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.588614393125672e-05,
+      "loss": 1.0928,
+      "step": 323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.582169709989259e-05,
+      "loss": 1.6904,
+      "step": 324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5757250268528465e-05,
+      "loss": 1.0898,
+      "step": 325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.569280343716434e-05,
+      "loss": 0.8174,
+      "step": 326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.562835660580021e-05,
+      "loss": 0.5107,
+      "step": 327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5563909774436095e-05,
+      "loss": 0.6909,
+      "step": 328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.549946294307197e-05,
+      "loss": 0.9688,
+      "step": 329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.543501611170784e-05,
+      "loss": 1.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.537056928034372e-05,
+      "loss": 0.4773,
+      "step": 331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.530612244897959e-05,
+      "loss": 1.3174,
+      "step": 332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5241675617615465e-05,
+      "loss": 1.3652,
+      "step": 333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517722878625135e-05,
+      "loss": 1.2168,
+      "step": 334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.511278195488722e-05,
+      "loss": 1.5312,
+      "step": 335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5048335123523095e-05,
+      "loss": 0.7524,
+      "step": 336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.498388829215897e-05,
+      "loss": 1.25,
+      "step": 337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.491944146079484e-05,
+      "loss": 0.9595,
+      "step": 338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.485499462943072e-05,
+      "loss": 1.083,
+      "step": 339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.47905477980666e-05,
+      "loss": 0.8052,
+      "step": 340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.472610096670247e-05,
+      "loss": 0.9404,
+      "step": 341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.466165413533835e-05,
+      "loss": 0.5723,
+      "step": 342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4597207303974226e-05,
+      "loss": 0.1422,
+      "step": 343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4532760472610095e-05,
+      "loss": 0.6899,
+      "step": 344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.446831364124597e-05,
+      "loss": 0.5137,
+      "step": 345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.440386680988185e-05,
+      "loss": 0.5693,
+      "step": 346
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4339419978517726e-05,
+      "loss": 1.0957,
+      "step": 347
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.42749731471536e-05,
+      "loss": 0.7964,
+      "step": 348
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.421052631578947e-05,
+      "loss": 0.7773,
+      "step": 349
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.414607948442535e-05,
+      "loss": 0.5635,
+      "step": 350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4081632653061226e-05,
+      "loss": 0.2981,
+      "step": 351
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4017185821697096e-05,
+      "loss": 0.2563,
+      "step": 352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.395273899033298e-05,
+      "loss": 0.5273,
+      "step": 353
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3888292158968856e-05,
+      "loss": 0.7793,
+      "step": 354
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3823845327604726e-05,
+      "loss": 0.5918,
+      "step": 355
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.37593984962406e-05,
+      "loss": 0.1747,
+      "step": 356
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.369495166487648e-05,
+      "loss": 0.8096,
+      "step": 357
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.363050483351235e-05,
+      "loss": 0.7832,
+      "step": 358
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.356605800214823e-05,
+      "loss": 0.874,
+      "step": 359
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.35016111707841e-05,
+      "loss": 0.7764,
+      "step": 360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.343716433941998e-05,
+      "loss": 0.6079,
+      "step": 361
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3372717508055856e-05,
+      "loss": 0.7559,
+      "step": 362
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3308270676691726e-05,
+      "loss": 0.3838,
+      "step": 363
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.32438238453276e-05,
+      "loss": 0.7886,
+      "step": 364
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3179377013963486e-05,
+      "loss": 0.7612,
+      "step": 365
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3114930182599356e-05,
+      "loss": 1.0811,
+      "step": 366
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.305048335123523e-05,
+      "loss": 1.0557,
+      "step": 367
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.298603651987111e-05,
+      "loss": 0.8599,
+      "step": 368
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.292158968850698e-05,
+      "loss": 0.6172,
+      "step": 369
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.8408,
+      "step": 370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.279269602577873e-05,
+      "loss": 1.1182,
+      "step": 371
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.272824919441461e-05,
+      "loss": 0.8789,
+      "step": 372
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.2663802363050486e-05,
+      "loss": 0.543,
+      "step": 373
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.2599355531686356e-05,
+      "loss": 1.2139,
+      "step": 374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.253490870032223e-05,
+      "loss": 0.5264,
+      "step": 375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.247046186895811e-05,
+      "loss": 0.6206,
+      "step": 376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.2406015037593987e-05,
+      "loss": 0.6782,
+      "step": 377
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.234156820622986e-05,
+      "loss": 0.6387,
+      "step": 378
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.227712137486574e-05,
+      "loss": 0.834,
+      "step": 379
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.221267454350161e-05,
+      "loss": 0.873,
+      "step": 380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.214822771213749e-05,
+      "loss": 0.3003,
+      "step": 381
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.208378088077336e-05,
+      "loss": 0.6514,
+      "step": 382
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.201933404940924e-05,
+      "loss": 1.4746,
+      "step": 383
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.195488721804512e-05,
+      "loss": 0.5039,
+      "step": 384
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.189044038668099e-05,
+      "loss": 1.627,
+      "step": 385
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.1825993555316863e-05,
+      "loss": 0.8374,
+      "step": 386
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.176154672395274e-05,
+      "loss": 0.769,
+      "step": 387
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.169709989258861e-05,
+      "loss": 0.4521,
+      "step": 388
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.1632653061224494e-05,
+      "loss": 1.3594,
+      "step": 389
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.156820622986037e-05,
+      "loss": 0.6465,
+      "step": 390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.150375939849624e-05,
+      "loss": 0.8965,
+      "step": 391
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.143931256713212e-05,
+      "loss": 0.8516,
+      "step": 392
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.1374865735767994e-05,
+      "loss": 0.5928,
+      "step": 393
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.1310418904403864e-05,
+      "loss": 0.6201,
+      "step": 394
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.124597207303975e-05,
+      "loss": 1.373,
+      "step": 395
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.1181525241675624e-05,
+      "loss": 0.3723,
+      "step": 396
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.1117078410311494e-05,
+      "loss": 0.3945,
+      "step": 397
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.105263157894737e-05,
+      "loss": 0.5054,
+      "step": 398
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.098818474758324e-05,
+      "loss": 1.4258,
+      "step": 399
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.092373791621912e-05,
+      "loss": 0.627,
+      "step": 400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.0859291084854994e-05,
+      "loss": 0.687,
+      "step": 401
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.079484425349087e-05,
+      "loss": 1.1934,
+      "step": 402
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.073039742212675e-05,
+      "loss": 0.3296,
+      "step": 403
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.0665950590762624e-05,
+      "loss": 0.6445,
+      "step": 404
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.0601503759398494e-05,
+      "loss": 0.3169,
+      "step": 405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.053705692803437e-05,
+      "loss": 0.7422,
+      "step": 406
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.047261009667025e-05,
+      "loss": 1.2725,
+      "step": 407
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.0408163265306124e-05,
+      "loss": 0.5264,
+      "step": 408
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.0343716433942e-05,
+      "loss": 0.9751,
+      "step": 409
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.027926960257788e-05,
+      "loss": 1.1699,
+      "step": 410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.021482277121375e-05,
+      "loss": 0.417,
+      "step": 411
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.0150375939849624e-05,
+      "loss": 0.5767,
+      "step": 412
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.0085929108485494e-05,
+      "loss": 0.4243,
+      "step": 413
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.002148227712138e-05,
+      "loss": 1.4678,
+      "step": 414
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9957035445757254e-05,
+      "loss": 1.1025,
+      "step": 415
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9892588614393124e-05,
+      "loss": 1.2783,
+      "step": 416
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9828141783029e-05,
+      "loss": 0.3447,
+      "step": 417
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.976369495166488e-05,
+      "loss": 0.4824,
+      "step": 418
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.969924812030075e-05,
+      "loss": 1.3311,
+      "step": 419
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.963480128893663e-05,
+      "loss": 0.5762,
+      "step": 420
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.957035445757251e-05,
+      "loss": 0.5205,
+      "step": 421
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.950590762620838e-05,
+      "loss": 0.4446,
+      "step": 422
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9441460794844255e-05,
+      "loss": 0.2427,
+      "step": 423
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.937701396348013e-05,
+      "loss": 0.7974,
+      "step": 424
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9312567132116e-05,
+      "loss": 0.3008,
+      "step": 425
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9248120300751885e-05,
+      "loss": 1.2344,
+      "step": 426
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9183673469387755e-05,
+      "loss": 0.6748,
+      "step": 427
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.911922663802363e-05,
+      "loss": 0.3877,
+      "step": 428
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.905477980665951e-05,
+      "loss": 1.0928,
+      "step": 429
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.899033297529538e-05,
+      "loss": 1.876,
+      "step": 430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8925886143931255e-05,
+      "loss": 0.9224,
+      "step": 431
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.886143931256714e-05,
+      "loss": 1.3135,
+      "step": 432
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.879699248120301e-05,
+      "loss": 0.8394,
+      "step": 433
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8732545649838885e-05,
+      "loss": 0.8125,
+      "step": 434
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.866809881847476e-05,
+      "loss": 0.6909,
+      "step": 435
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.860365198711063e-05,
+      "loss": 1.6338,
+      "step": 436
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.853920515574651e-05,
+      "loss": 0.3625,
+      "step": 437
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.847475832438239e-05,
+      "loss": 1.2891,
+      "step": 438
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.841031149301826e-05,
+      "loss": 1.0586,
+      "step": 439
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.834586466165414e-05,
+      "loss": 0.6055,
+      "step": 440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.828141783029001e-05,
+      "loss": 0.7695,
+      "step": 441
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.8216970998925885e-05,
+      "loss": 0.4434,
+      "step": 442
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.815252416756176e-05,
+      "loss": 0.8013,
+      "step": 443
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.808807733619764e-05,
+      "loss": 0.5269,
+      "step": 444
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8023630504833515e-05,
+      "loss": 0.6255,
+      "step": 445
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.795918367346939e-05,
+      "loss": 1.5127,
+      "step": 446
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.789473684210526e-05,
+      "loss": 0.7939,
+      "step": 447
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.783029001074114e-05,
+      "loss": 0.6274,
+      "step": 448
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7765843179377015e-05,
+      "loss": 0.8921,
+      "step": 449
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7701396348012885e-05,
+      "loss": 0.7173,
+      "step": 450
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.763694951664877e-05,
+      "loss": 0.8696,
+      "step": 451
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7572502685284646e-05,
+      "loss": 0.3567,
+      "step": 452
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7508055853920516e-05,
+      "loss": 0.2025,
+      "step": 453
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.744360902255639e-05,
+      "loss": 0.1935,
+      "step": 454
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.737916219119226e-05,
+      "loss": 0.9375,
+      "step": 455
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.731471535982814e-05,
+      "loss": 0.7847,
+      "step": 456
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.725026852846402e-05,
+      "loss": 0.2959,
+      "step": 457
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.718582169709989e-05,
+      "loss": 0.873,
+      "step": 458
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.712137486573577e-05,
+      "loss": 0.3376,
+      "step": 459
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7056928034371646e-05,
+      "loss": 0.2786,
+      "step": 460
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6992481203007516e-05,
+      "loss": 0.5781,
+      "step": 461
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.692803437164339e-05,
+      "loss": 1.5342,
+      "step": 462
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6863587540279276e-05,
+      "loss": 1.1494,
+      "step": 463
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6799140708915146e-05,
+      "loss": 0.4146,
+      "step": 464
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.673469387755102e-05,
+      "loss": 0.6514,
+      "step": 465
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.66702470461869e-05,
+      "loss": 0.5601,
+      "step": 466
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.660580021482277e-05,
+      "loss": 0.9951,
+      "step": 467
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6541353383458646e-05,
+      "loss": 0.3831,
+      "step": 468
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.647690655209452e-05,
+      "loss": 0.6987,
+      "step": 469
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.64124597207304e-05,
+      "loss": 0.6621,
+      "step": 470
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6348012889366276e-05,
+      "loss": 0.7173,
+      "step": 471
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6283566058002146e-05,
+      "loss": 1.0811,
+      "step": 472
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.621911922663802e-05,
+      "loss": 1.2783,
+      "step": 473
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.61546723952739e-05,
+      "loss": 1.5293,
+      "step": 474
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6090225563909776e-05,
+      "loss": 1.3809,
+      "step": 475
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.602577873254565e-05,
+      "loss": 1.0098,
+      "step": 476
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.596133190118153e-05,
+      "loss": 0.6865,
+      "step": 477
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.58968850698174e-05,
+      "loss": 2.1152,
+      "step": 478
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.5832438238453276e-05,
+      "loss": 0.7798,
+      "step": 479
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.576799140708915e-05,
+      "loss": 0.9668,
+      "step": 480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.570354457572503e-05,
+      "loss": 0.6685,
+      "step": 481
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5639097744360906e-05,
+      "loss": 0.228,
+      "step": 482
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5574650912996776e-05,
+      "loss": 1.3633,
+      "step": 483
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.551020408163265e-05,
+      "loss": 0.8384,
+      "step": 484
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.544575725026853e-05,
+      "loss": 1.4463,
+      "step": 485
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.53813104189044e-05,
+      "loss": 0.9956,
+      "step": 486
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.531686358754028e-05,
+      "loss": 0.9316,
+      "step": 487
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.525241675617616e-05,
+      "loss": 0.9541,
+      "step": 488
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.518796992481203e-05,
+      "loss": 0.1255,
+      "step": 489
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.512352309344791e-05,
+      "loss": 0.7798,
+      "step": 490
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.505907626208378e-05,
+      "loss": 0.1796,
+      "step": 491
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.499462943071965e-05,
+      "loss": 0.251,
+      "step": 492
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.493018259935554e-05,
+      "loss": 0.9492,
+      "step": 493
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4865735767991414e-05,
+      "loss": 0.6108,
+      "step": 494
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4801288936627283e-05,
+      "loss": 0.5967,
+      "step": 495
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.473684210526316e-05,
+      "loss": 0.8066,
+      "step": 496
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.467239527389903e-05,
+      "loss": 1.2207,
+      "step": 497
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.460794844253491e-05,
+      "loss": 0.8301,
+      "step": 498
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4543501611170784e-05,
+      "loss": 1.124,
+      "step": 499
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.447905477980666e-05,
+      "loss": 0.5278,
+      "step": 500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.441460794844254e-05,
+      "loss": 0.174,
+      "step": 501
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4350161117078414e-05,
+      "loss": 0.5508,
+      "step": 502
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4285714285714284e-05,
+      "loss": 0.9468,
+      "step": 503
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.422126745435016e-05,
+      "loss": 1.4697,
+      "step": 504
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.415682062298604e-05,
+      "loss": 1.375,
+      "step": 505
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4092373791621914e-05,
+      "loss": 1.6553,
+      "step": 506
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.402792696025779e-05,
+      "loss": 0.9907,
+      "step": 507
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.396348012889366e-05,
+      "loss": 0.5063,
+      "step": 508
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.389903329752954e-05,
+      "loss": 1.9912,
+      "step": 509
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3834586466165414e-05,
+      "loss": 0.4668,
+      "step": 510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3770139634801284e-05,
+      "loss": 0.6558,
+      "step": 511
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.370569280343717e-05,
+      "loss": 0.8813,
+      "step": 512
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3641245972073044e-05,
+      "loss": 1.0352,
+      "step": 513
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3576799140708914e-05,
+      "loss": 0.8276,
+      "step": 514
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.351235230934479e-05,
+      "loss": 1.2236,
+      "step": 515
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.344790547798067e-05,
+      "loss": 0.9565,
+      "step": 516
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.338345864661654e-05,
+      "loss": 0.9995,
+      "step": 517
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.331901181525242e-05,
+      "loss": 0.5166,
+      "step": 518
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.32545649838883e-05,
+      "loss": 0.5742,
+      "step": 519
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.319011815252417e-05,
+      "loss": 0.8345,
+      "step": 520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3125671321160044e-05,
+      "loss": 1.2949,
+      "step": 521
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3061224489795914e-05,
+      "loss": 1.1328,
+      "step": 522
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.299677765843179e-05,
+      "loss": 0.7051,
+      "step": 523
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2932330827067674e-05,
+      "loss": 0.8169,
+      "step": 524
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2867883995703544e-05,
+      "loss": 1.0195,
+      "step": 525
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.280343716433942e-05,
+      "loss": 0.8081,
+      "step": 526
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.27389903329753e-05,
+      "loss": 0.7212,
+      "step": 527
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.267454350161117e-05,
+      "loss": 0.7681,
+      "step": 528
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2610096670247044e-05,
+      "loss": 0.7163,
+      "step": 529
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.254564983888293e-05,
+      "loss": 0.9805,
+      "step": 530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.24812030075188e-05,
+      "loss": 0.627,
+      "step": 531
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.2416756176154675e-05,
+      "loss": 1.0391,
+      "step": 532
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.235230934479055e-05,
+      "loss": 1.0283,
+      "step": 533
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.228786251342642e-05,
+      "loss": 2.1133,
+      "step": 534
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.22234156820623e-05,
+      "loss": 1.041,
+      "step": 535
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.2158968850698175e-05,
+      "loss": 0.7285,
+      "step": 536
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.209452201933405e-05,
+      "loss": 0.4043,
+      "step": 537
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.203007518796993e-05,
+      "loss": 0.5142,
+      "step": 538
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.19656283566058e-05,
+      "loss": 1.5195,
+      "step": 539
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.1901181525241675e-05,
+      "loss": 0.2092,
+      "step": 540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.183673469387755e-05,
+      "loss": 0.5186,
+      "step": 541
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.177228786251343e-05,
+      "loss": 0.8418,
+      "step": 542
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.1707841031149305e-05,
+      "loss": 0.5356,
+      "step": 543
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.164339419978518e-05,
+      "loss": 0.647,
+      "step": 544
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 0.6558,
+      "step": 545
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.151450053705693e-05,
+      "loss": 0.8794,
+      "step": 546
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1450053705692805e-05,
+      "loss": 0.7026,
+      "step": 547
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1385606874328675e-05,
+      "loss": 1.7617,
+      "step": 548
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.132116004296456e-05,
+      "loss": 0.7349,
+      "step": 549
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.125671321160043e-05,
+      "loss": 0.3079,
+      "step": 550
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1192266380236305e-05,
+      "loss": 0.4131,
+      "step": 551
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.112781954887218e-05,
+      "loss": 0.4033,
+      "step": 552
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.106337271750805e-05,
+      "loss": 0.665,
+      "step": 553
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.099892588614393e-05,
+      "loss": 0.6807,
+      "step": 554
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.093447905477981e-05,
+      "loss": 0.3914,
+      "step": 555
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.087003222341568e-05,
+      "loss": 0.2849,
+      "step": 556
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.080558539205156e-05,
+      "loss": 0.3276,
+      "step": 557
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0741138560687435e-05,
+      "loss": 0.4121,
+      "step": 558
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0676691729323305e-05,
+      "loss": 0.2688,
+      "step": 559
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.061224489795918e-05,
+      "loss": 0.1829,
+      "step": 560
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0547798066595066e-05,
+      "loss": 0.5039,
+      "step": 561
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0483351235230936e-05,
+      "loss": 0.4885,
+      "step": 562
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0418904403866812e-05,
+      "loss": 0.7773,
+      "step": 563
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0354457572502682e-05,
+      "loss": 0.4307,
+      "step": 564
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.0290010741138562e-05,
+      "loss": 0.8657,
+      "step": 565
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.022556390977444e-05,
+      "loss": 0.4141,
+      "step": 566
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.016111707841031e-05,
+      "loss": 0.2261,
+      "step": 567
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.009667024704619e-05,
+      "loss": 1.0059,
+      "step": 568
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0032223415682066e-05,
+      "loss": 0.6479,
+      "step": 569
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.996777658431794e-05,
+      "loss": 1.0986,
+      "step": 570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9903329752953812e-05,
+      "loss": 0.4487,
+      "step": 571
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.983888292158969e-05,
+      "loss": 0.7275,
+      "step": 572
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9774436090225562e-05,
+      "loss": 0.4023,
+      "step": 573
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.970998925886144e-05,
+      "loss": 0.217,
+      "step": 574
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9645542427497316e-05,
+      "loss": 0.647,
+      "step": 575
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.958109559613319e-05,
+      "loss": 2.2598,
+      "step": 576
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9516648764769066e-05,
+      "loss": 0.7632,
+      "step": 577
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9452201933404943e-05,
+      "loss": 0.2375,
+      "step": 578
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9387755102040816e-05,
+      "loss": 1.0186,
+      "step": 579
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.932330827067669e-05,
+      "loss": 0.3733,
+      "step": 580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.925886143931257e-05,
+      "loss": 0.5693,
+      "step": 581
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9194414607948443e-05,
+      "loss": 0.791,
+      "step": 582
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9129967776584316e-05,
+      "loss": 1.4707,
+      "step": 583
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9065520945220196e-05,
+      "loss": 0.6255,
+      "step": 584
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.900107411385607e-05,
+      "loss": 0.7905,
+      "step": 585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8936627282491943e-05,
+      "loss": 0.1816,
+      "step": 586
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.887218045112782e-05,
+      "loss": 0.5864,
+      "step": 587
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8807733619763696e-05,
+      "loss": 0.5752,
+      "step": 588
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.874328678839957e-05,
+      "loss": 1.4062,
+      "step": 589
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8678839957035446e-05,
+      "loss": 0.3428,
+      "step": 590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8614393125671323e-05,
+      "loss": 0.1882,
+      "step": 591
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8549946294307196e-05,
+      "loss": 0.2407,
+      "step": 592
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8485499462943073e-05,
+      "loss": 0.519,
+      "step": 593
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8421052631578946e-05,
+      "loss": 0.5439,
+      "step": 594
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8356605800214823e-05,
+      "loss": 0.8413,
+      "step": 595
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.82921589688507e-05,
+      "loss": 0.627,
+      "step": 596
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8227712137486573e-05,
+      "loss": 0.2878,
+      "step": 597
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.816326530612245e-05,
+      "loss": 1.4326,
+      "step": 598
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8098818474758327e-05,
+      "loss": 0.7769,
+      "step": 599
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.80343716433942e-05,
+      "loss": 1.1152,
+      "step": 600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7969924812030073e-05,
+      "loss": 0.5122,
+      "step": 601
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.7905477980665953e-05,
+      "loss": 0.4219,
+      "step": 602
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.7841031149301827e-05,
+      "loss": 0.3728,
+      "step": 603
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.77765843179377e-05,
+      "loss": 0.9292,
+      "step": 604
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.771213748657358e-05,
+      "loss": 0.439,
+      "step": 605
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7647690655209454e-05,
+      "loss": 0.4609,
+      "step": 606
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7583243823845327e-05,
+      "loss": 0.5693,
+      "step": 607
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7518796992481204e-05,
+      "loss": 0.6675,
+      "step": 608
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.745435016111708e-05,
+      "loss": 1.1182,
+      "step": 609
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7389903329752954e-05,
+      "loss": 0.5806,
+      "step": 610
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.732545649838883e-05,
+      "loss": 1.4824,
+      "step": 611
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7261009667024707e-05,
+      "loss": 0.8848,
+      "step": 612
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.719656283566058e-05,
+      "loss": 0.6616,
+      "step": 613
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7132116004296457e-05,
+      "loss": 1.1816,
+      "step": 614
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.706766917293233e-05,
+      "loss": 0.7158,
+      "step": 615
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7003222341568207e-05,
+      "loss": 1.0264,
+      "step": 616
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.6938775510204084e-05,
+      "loss": 1.3906,
+      "step": 617
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.6874328678839957e-05,
+      "loss": 0.8315,
+      "step": 618
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.6809881847475834e-05,
+      "loss": 0.325,
+      "step": 619
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6745435016111707e-05,
+      "loss": 0.6333,
+      "step": 620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6680988184747584e-05,
+      "loss": 0.189,
+      "step": 621
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6616541353383457e-05,
+      "loss": 0.5181,
+      "step": 622
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6552094522019334e-05,
+      "loss": 0.7617,
+      "step": 623
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.648764769065521e-05,
+      "loss": 0.6064,
+      "step": 624
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6423200859291084e-05,
+      "loss": 0.4485,
+      "step": 625
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.635875402792696e-05,
+      "loss": 1.4287,
+      "step": 626
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6294307196562838e-05,
+      "loss": 0.5005,
+      "step": 627
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.622986036519871e-05,
+      "loss": 1.1748,
+      "step": 628
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6165413533834584e-05,
+      "loss": 0.6528,
+      "step": 629
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6100966702470464e-05,
+      "loss": 0.8218,
+      "step": 630
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6036519871106338e-05,
+      "loss": 0.667,
+      "step": 631
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.597207303974221e-05,
+      "loss": 0.7935,
+      "step": 632
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.590762620837809e-05,
+      "loss": 0.564,
+      "step": 633
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.5843179377013964e-05,
+      "loss": 0.7573,
+      "step": 634
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.5778732545649838e-05,
+      "loss": 0.6895,
+      "step": 635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 1.2871,
+      "step": 636
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.564983888292159e-05,
+      "loss": 0.2174,
+      "step": 637
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.5585392051557465e-05,
+      "loss": 0.4556,
+      "step": 638
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.552094522019334e-05,
+      "loss": 0.2581,
+      "step": 639
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5456498388829218e-05,
+      "loss": 0.2334,
+      "step": 640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.539205155746509e-05,
+      "loss": 0.9619,
+      "step": 641
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5327604726100968e-05,
+      "loss": 0.8389,
+      "step": 642
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.526315789473684e-05,
+      "loss": 0.4641,
+      "step": 643
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5198711063372718e-05,
+      "loss": 0.6216,
+      "step": 644
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5134264232008595e-05,
+      "loss": 0.8599,
+      "step": 645
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5069817400644468e-05,
+      "loss": 1.0166,
+      "step": 646
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5005370569280345e-05,
+      "loss": 0.5615,
+      "step": 647
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.494092373791622e-05,
+      "loss": 0.2927,
+      "step": 648
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4876476906552095e-05,
+      "loss": 0.2435,
+      "step": 649
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4812030075187968e-05,
+      "loss": 1.791,
+      "step": 650
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4747583243823848e-05,
+      "loss": 0.7222,
+      "step": 651
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.468313641245972e-05,
+      "loss": 0.5127,
+      "step": 652
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4618689581095595e-05,
+      "loss": 1.2412,
+      "step": 653
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4554242749731475e-05,
+      "loss": 0.5151,
+      "step": 654
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 0.6201,
+      "step": 655
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4425349087003222e-05,
+      "loss": 0.3262,
+      "step": 656
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.43609022556391e-05,
+      "loss": 0.9502,
+      "step": 657
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4296455424274975e-05,
+      "loss": 0.8535,
+      "step": 658
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.423200859291085e-05,
+      "loss": 1.6621,
+      "step": 659
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4167561761546725e-05,
+      "loss": 1.0186,
+      "step": 660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4103114930182602e-05,
+      "loss": 1.1162,
+      "step": 661
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4038668098818475e-05,
+      "loss": 0.4033,
+      "step": 662
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3974221267454352e-05,
+      "loss": 1.3906,
+      "step": 663
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3909774436090225e-05,
+      "loss": 0.7471,
+      "step": 664
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3845327604726102e-05,
+      "loss": 1.1973,
+      "step": 665
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.378088077336198e-05,
+      "loss": 0.5869,
+      "step": 666
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3716433941997852e-05,
+      "loss": 0.9888,
+      "step": 667
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.365198711063373e-05,
+      "loss": 1.0742,
+      "step": 668
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3587540279269602e-05,
+      "loss": 0.2125,
+      "step": 669
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.352309344790548e-05,
+      "loss": 1.0732,
+      "step": 670
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3458646616541352e-05,
+      "loss": 0.4409,
+      "step": 671
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.339419978517723e-05,
+      "loss": 0.4382,
+      "step": 672
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3329752953813106e-05,
+      "loss": 0.4102,
+      "step": 673
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.326530612244898e-05,
+      "loss": 0.6001,
+      "step": 674
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3200859291084856e-05,
+      "loss": 0.2225,
+      "step": 675
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3136412459720732e-05,
+      "loss": 0.4841,
+      "step": 676
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3071965628356606e-05,
+      "loss": 0.3896,
+      "step": 677
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.300751879699248e-05,
+      "loss": 0.8467,
+      "step": 678
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.294307196562836e-05,
+      "loss": 0.6938,
+      "step": 679
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2878625134264232e-05,
+      "loss": 0.6113,
+      "step": 680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2814178302900106e-05,
+      "loss": 0.6523,
+      "step": 681
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2749731471535986e-05,
+      "loss": 1.002,
+      "step": 682
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.268528464017186e-05,
+      "loss": 0.1422,
+      "step": 683
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2620837808807733e-05,
+      "loss": 0.4534,
+      "step": 684
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.255639097744361e-05,
+      "loss": 0.2925,
+      "step": 685
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2491944146079486e-05,
+      "loss": 0.5317,
+      "step": 686
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.242749731471536e-05,
+      "loss": 0.8975,
+      "step": 687
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2363050483351236e-05,
+      "loss": 1.1289,
+      "step": 688
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.2298603651987113e-05,
+      "loss": 0.7373,
+      "step": 689
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.2234156820622986e-05,
+      "loss": 0.2634,
+      "step": 690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.2169709989258863e-05,
+      "loss": 0.1131,
+      "step": 691
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2105263157894736e-05,
+      "loss": 0.8384,
+      "step": 692
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2040816326530613e-05,
+      "loss": 0.2357,
+      "step": 693
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.197636949516649e-05,
+      "loss": 0.4929,
+      "step": 694
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.1911922663802363e-05,
+      "loss": 1.3955,
+      "step": 695
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.184747583243824e-05,
+      "loss": 0.3135,
+      "step": 696
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.1783029001074116e-05,
+      "loss": 0.4622,
+      "step": 697
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.171858216970999e-05,
+      "loss": 0.5776,
+      "step": 698
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.1654135338345863e-05,
+      "loss": 1.3086,
+      "step": 699
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.1589688506981743e-05,
+      "loss": 0.0809,
+      "step": 700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.1525241675617616e-05,
+      "loss": 0.192,
+      "step": 701
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.146079484425349e-05,
+      "loss": 0.6079,
+      "step": 702
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.1396348012889367e-05,
+      "loss": 1.252,
+      "step": 703
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.1331901181525243e-05,
+      "loss": 0.8066,
+      "step": 704
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.1267454350161117e-05,
+      "loss": 0.427,
+      "step": 705
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.1203007518796993e-05,
+      "loss": 0.3782,
+      "step": 706
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.113856068743287e-05,
+      "loss": 0.5156,
+      "step": 707
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.1074113856068743e-05,
+      "loss": 0.2861,
+      "step": 708
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.100966702470462e-05,
+      "loss": 0.7266,
+      "step": 709
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0945220193340493e-05,
+      "loss": 0.7905,
+      "step": 710
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.088077336197637e-05,
+      "loss": 0.5312,
+      "step": 711
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.0816326530612247e-05,
+      "loss": 0.3154,
+      "step": 712
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.075187969924812e-05,
+      "loss": 0.2529,
+      "step": 713
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.0687432867883997e-05,
+      "loss": 0.5879,
+      "step": 714
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.0622986036519874e-05,
+      "loss": 1.0176,
+      "step": 715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0558539205155747e-05,
+      "loss": 0.6187,
+      "step": 716
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.049409237379162e-05,
+      "loss": 1.1318,
+      "step": 717
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0429645542427497e-05,
+      "loss": 0.689,
+      "step": 718
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.0365198711063374e-05,
+      "loss": 1.5176,
+      "step": 719
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0300751879699247e-05,
+      "loss": 0.1791,
+      "step": 720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0236305048335124e-05,
+      "loss": 0.6118,
+      "step": 721
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.0171858216971e-05,
+      "loss": 0.7197,
+      "step": 722
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.0107411385606874e-05,
+      "loss": 0.2455,
+      "step": 723
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.0042964554242747e-05,
+      "loss": 0.6484,
+      "step": 724
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9978517722878627e-05,
+      "loss": 0.4783,
+      "step": 725
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.99140708915145e-05,
+      "loss": 0.3025,
+      "step": 726
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9849624060150374e-05,
+      "loss": 0.5659,
+      "step": 727
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9785177228786254e-05,
+      "loss": 0.5,
+      "step": 728
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9720730397422127e-05,
+      "loss": 1.1221,
+      "step": 729
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9656283566058e-05,
+      "loss": 1.127,
+      "step": 730
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 0.8398,
+      "step": 731
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9527389903329754e-05,
+      "loss": 0.6069,
+      "step": 732
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9462943071965627e-05,
+      "loss": 0.6099,
+      "step": 733
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9398496240601504e-05,
+      "loss": 0.3914,
+      "step": 734
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.933404940923738e-05,
+      "loss": 0.478,
+      "step": 735
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9269602577873254e-05,
+      "loss": 0.4048,
+      "step": 736
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.920515574650913e-05,
+      "loss": 0.4072,
+      "step": 737
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9140708915145004e-05,
+      "loss": 0.2915,
+      "step": 738
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.907626208378088e-05,
+      "loss": 0.2668,
+      "step": 739
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9011815252416758e-05,
+      "loss": 0.2759,
+      "step": 740
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.7534,
+      "step": 741
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8882921589688508e-05,
+      "loss": 0.2269,
+      "step": 742
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8818474758324384e-05,
+      "loss": 0.5571,
+      "step": 743
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8754027926960258e-05,
+      "loss": 0.1215,
+      "step": 744
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.868958109559613e-05,
+      "loss": 0.749,
+      "step": 745
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.862513426423201e-05,
+      "loss": 0.4685,
+      "step": 746
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8560687432867885e-05,
+      "loss": 0.6719,
+      "step": 747
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8496240601503758e-05,
+      "loss": 0.3398,
+      "step": 748
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8431793770139638e-05,
+      "loss": 1.0566,
+      "step": 749
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 0.3513,
+      "step": 750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8302900107411385e-05,
+      "loss": 0.7632,
+      "step": 751
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.823845327604726e-05,
+      "loss": 0.1722,
+      "step": 752
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8174006444683138e-05,
+      "loss": 1.0508,
+      "step": 753
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.810955961331901e-05,
+      "loss": 0.394,
+      "step": 754
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8045112781954888e-05,
+      "loss": 1.0664,
+      "step": 755
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7980665950590765e-05,
+      "loss": 0.793,
+      "step": 756
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7916219119226638e-05,
+      "loss": 0.1925,
+      "step": 757
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7851772287862515e-05,
+      "loss": 0.2903,
+      "step": 758
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7787325456498388e-05,
+      "loss": 0.1981,
+      "step": 759
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7722878625134265e-05,
+      "loss": 0.7808,
+      "step": 760
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.765843179377014e-05,
+      "loss": 1.5342,
+      "step": 761
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7593984962406015e-05,
+      "loss": 0.2656,
+      "step": 762
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.752953813104189e-05,
+      "loss": 0.408,
+      "step": 763
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.746509129967777e-05,
+      "loss": 0.4575,
+      "step": 764
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7400644468313642e-05,
+      "loss": 0.5376,
+      "step": 765
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7336197636949515e-05,
+      "loss": 0.6138,
+      "step": 766
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7271750805585392e-05,
+      "loss": 0.6538,
+      "step": 767
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.720730397422127e-05,
+      "loss": 0.3455,
+      "step": 768
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 2.2188,
+      "step": 769
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.707841031149302e-05,
+      "loss": 0.8359,
+      "step": 770
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7013963480128895e-05,
+      "loss": 0.4607,
+      "step": 771
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.694951664876477e-05,
+      "loss": 0.5005,
+      "step": 772
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.6885069817400642e-05,
+      "loss": 0.8154,
+      "step": 773
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.6820622986036522e-05,
+      "loss": 0.561,
+      "step": 774
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.6756176154672395e-05,
+      "loss": 0.7104,
+      "step": 775
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.669172932330827e-05,
+      "loss": 0.1943,
+      "step": 776
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.662728249194415e-05,
+      "loss": 1.1729,
+      "step": 777
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.6562835660580022e-05,
+      "loss": 0.3677,
+      "step": 778
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.6498388829215895e-05,
+      "loss": 0.2639,
+      "step": 779
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.6433941997851772e-05,
+      "loss": 0.6929,
+      "step": 780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.636949516648765e-05,
+      "loss": 0.243,
+      "step": 781
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6305048335123522e-05,
+      "loss": 0.7842,
+      "step": 782
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.62406015037594e-05,
+      "loss": 1.2842,
+      "step": 783
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6176154672395276e-05,
+      "loss": 0.6558,
+      "step": 784
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.611170784103115e-05,
+      "loss": 0.647,
+      "step": 785
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6047261009667026e-05,
+      "loss": 0.3364,
+      "step": 786
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.59828141783029e-05,
+      "loss": 0.45,
+      "step": 787
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 1.127,
+      "step": 788
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5853920515574652e-05,
+      "loss": 0.666,
+      "step": 789
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5789473684210526e-05,
+      "loss": 0.0742,
+      "step": 790
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5725026852846403e-05,
+      "loss": 0.897,
+      "step": 791
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.566058002148228e-05,
+      "loss": 0.0906,
+      "step": 792
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5596133190118153e-05,
+      "loss": 0.79,
+      "step": 793
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5531686358754026e-05,
+      "loss": 0.5742,
+      "step": 794
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5467239527389906e-05,
+      "loss": 0.894,
+      "step": 795
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.540279269602578e-05,
+      "loss": 0.5088,
+      "step": 796
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5338345864661653e-05,
+      "loss": 0.2896,
+      "step": 797
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5273899033297533e-05,
+      "loss": 0.385,
+      "step": 798
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5209452201933406e-05,
+      "loss": 0.7002,
+      "step": 799
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5145005370569281e-05,
+      "loss": 0.2917,
+      "step": 800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5080558539205154e-05,
+      "loss": 0.3679,
+      "step": 801
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5016111707841033e-05,
+      "loss": 0.2607,
+      "step": 802
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4951664876476906e-05,
+      "loss": 0.5879,
+      "step": 803
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4887218045112781e-05,
+      "loss": 0.6631,
+      "step": 804
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4822771213748658e-05,
+      "loss": 0.5054,
+      "step": 805
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4758324382384533e-05,
+      "loss": 0.6641,
+      "step": 806
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4693877551020408e-05,
+      "loss": 0.9521,
+      "step": 807
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4629430719656285e-05,
+      "loss": 1.9043,
+      "step": 808
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4564983888292158e-05,
+      "loss": 1.0449,
+      "step": 809
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4500537056928035e-05,
+      "loss": 0.5547,
+      "step": 810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.443609022556391e-05,
+      "loss": 1.0,
+      "step": 811
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.4371643394199785e-05,
+      "loss": 0.5303,
+      "step": 812
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.4307196562835662e-05,
+      "loss": 0.4465,
+      "step": 813
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.4242749731471537e-05,
+      "loss": 0.6113,
+      "step": 814
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.4178302900107412e-05,
+      "loss": 0.1066,
+      "step": 815
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.4113856068743287e-05,
+      "loss": 0.3657,
+      "step": 816
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.4049409237379163e-05,
+      "loss": 0.8228,
+      "step": 817
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3984962406015037e-05,
+      "loss": 1.3389,
+      "step": 818
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.3920515574650913e-05,
+      "loss": 0.311,
+      "step": 819
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.385606874328679e-05,
+      "loss": 0.4507,
+      "step": 820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.3791621911922663e-05,
+      "loss": 0.1243,
+      "step": 821
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.372717508055854e-05,
+      "loss": 1.3535,
+      "step": 822
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3662728249194415e-05,
+      "loss": 0.3181,
+      "step": 823
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.359828141783029e-05,
+      "loss": 0.5601,
+      "step": 824
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3533834586466165e-05,
+      "loss": 0.4221,
+      "step": 825
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 0.5015,
+      "step": 826
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3404940923737917e-05,
+      "loss": 0.4658,
+      "step": 827
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3340494092373792e-05,
+      "loss": 0.6699,
+      "step": 828
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.3276047261009667e-05,
+      "loss": 0.647,
+      "step": 829
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3211600429645542e-05,
+      "loss": 0.4644,
+      "step": 830
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3147153598281419e-05,
+      "loss": 0.2188,
+      "step": 831
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3082706766917292e-05,
+      "loss": 0.3911,
+      "step": 832
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.3018259935553169e-05,
+      "loss": 0.4988,
+      "step": 833
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.2953813104189046e-05,
+      "loss": 0.2781,
+      "step": 834
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.2889366272824919e-05,
+      "loss": 0.8247,
+      "step": 835
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.2824919441460796e-05,
+      "loss": 1.2646,
+      "step": 836
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.276047261009667e-05,
+      "loss": 1.1211,
+      "step": 837
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.2696025778732546e-05,
+      "loss": 1.5156,
+      "step": 838
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.2452,
+      "step": 839
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.2567132116004297e-05,
+      "loss": 0.5449,
+      "step": 840
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.2502685284640172e-05,
+      "loss": 0.5576,
+      "step": 841
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.2438238453276047e-05,
+      "loss": 0.6338,
+      "step": 842
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.2373791621911924e-05,
+      "loss": 0.3057,
+      "step": 843
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.2309344790547797e-05,
+      "loss": 0.3623,
+      "step": 844
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 0.7378,
+      "step": 845
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.218045112781955e-05,
+      "loss": 0.4917,
+      "step": 846
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.2116004296455424e-05,
+      "loss": 0.8989,
+      "step": 847
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.2051557465091301e-05,
+      "loss": 0.4995,
+      "step": 848
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.1987110633727176e-05,
+      "loss": 0.2791,
+      "step": 849
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.1922663802363051e-05,
+      "loss": 0.6958,
+      "step": 850
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.1858216970998926e-05,
+      "loss": 1.3311,
+      "step": 851
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.1793770139634801e-05,
+      "loss": 1.2793,
+      "step": 852
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.1729323308270676e-05,
+      "loss": 1.166,
+      "step": 853
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.1664876476906553e-05,
+      "loss": 0.8516,
+      "step": 854
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.1600429645542428e-05,
+      "loss": 0.6934,
+      "step": 855
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.1535982814178303e-05,
+      "loss": 0.3757,
+      "step": 856
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.147153598281418e-05,
+      "loss": 0.2754,
+      "step": 857
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1407089151450053e-05,
+      "loss": 0.9756,
+      "step": 858
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.134264232008593e-05,
+      "loss": 0.3345,
+      "step": 859
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1278195488721805e-05,
+      "loss": 0.7622,
+      "step": 860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.121374865735768e-05,
+      "loss": 0.212,
+      "step": 861
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1149301825993556e-05,
+      "loss": 0.4236,
+      "step": 862
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1084854994629431e-05,
+      "loss": 1.582,
+      "step": 863
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 1.0332,
+      "step": 864
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0955961331901181e-05,
+      "loss": 0.9839,
+      "step": 865
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0891514500537058e-05,
+      "loss": 0.8301,
+      "step": 866
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0827067669172932e-05,
+      "loss": 2.4746,
+      "step": 867
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0762620837808808e-05,
+      "loss": 0.9419,
+      "step": 868
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0698174006444683e-05,
+      "loss": 0.5151,
+      "step": 869
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0633727175080558e-05,
+      "loss": 0.584,
+      "step": 870
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0569280343716435e-05,
+      "loss": 0.7236,
+      "step": 871
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.050483351235231e-05,
+      "loss": 0.2401,
+      "step": 872
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0440386680988185e-05,
+      "loss": 1.1973,
+      "step": 873
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.037593984962406e-05,
+      "loss": 0.5747,
+      "step": 874
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0311493018259937e-05,
+      "loss": 0.6992,
+      "step": 875
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.024704618689581e-05,
+      "loss": 0.4014,
+      "step": 876
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0182599355531687e-05,
+      "loss": 0.8003,
+      "step": 877
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0118152524167562e-05,
+      "loss": 0.3843,
+      "step": 878
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0053705692803437e-05,
+      "loss": 0.8589,
+      "step": 879
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.989258861439314e-06,
+      "loss": 0.6914,
+      "step": 880
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.924812030075187e-06,
+      "loss": 0.2391,
+      "step": 881
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.860365198711064e-06,
+      "loss": 0.9551,
+      "step": 882
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 0.7603,
+      "step": 883
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.731471535982814e-06,
+      "loss": 0.4573,
+      "step": 884
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.66702470461869e-06,
+      "loss": 0.7515,
+      "step": 885
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.602577873254565e-06,
+      "loss": 0.3679,
+      "step": 886
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.53813104189044e-06,
+      "loss": 1.2441,
+      "step": 887
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.8926,
+      "step": 888
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.409237379162192e-06,
+      "loss": 0.9185,
+      "step": 889
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.344790547798066e-06,
+      "loss": 0.2751,
+      "step": 890
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.280343716433942e-06,
+      "loss": 0.9556,
+      "step": 891
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.215896885069819e-06,
+      "loss": 0.2913,
+      "step": 892
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.151450053705692e-06,
+      "loss": 0.5596,
+      "step": 893
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.087003222341569e-06,
+      "loss": 1.1748,
+      "step": 894
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.022556390977444e-06,
+      "loss": 0.3442,
+      "step": 895
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.958109559613319e-06,
+      "loss": 0.6567,
+      "step": 896
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.893662728249194e-06,
+      "loss": 1.2559,
+      "step": 897
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.82921589688507e-06,
+      "loss": 0.3486,
+      "step": 898
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.764769065520946e-06,
+      "loss": 0.7559,
+      "step": 899
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.700322234156821e-06,
+      "loss": 0.4709,
+      "step": 900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.635875402792696e-06,
+      "loss": 0.8193,
+      "step": 901
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.7148,
+      "step": 902
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.506981740064448e-06,
+      "loss": 0.5742,
+      "step": 903
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.442534908700321e-06,
+      "loss": 0.7822,
+      "step": 904
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.378088077336198e-06,
+      "loss": 0.7886,
+      "step": 905
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.313641245972074e-06,
+      "loss": 0.3855,
+      "step": 906
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.249194414607948e-06,
+      "loss": 0.9004,
+      "step": 907
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.184747583243824e-06,
+      "loss": 0.6431,
+      "step": 908
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.1203007518797e-06,
+      "loss": 1.0674,
+      "step": 909
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.055853920515575e-06,
+      "loss": 1.1387,
+      "step": 910
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.99140708915145e-06,
+      "loss": 0.5269,
+      "step": 911
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.926960257787326e-06,
+      "loss": 0.2634,
+      "step": 912
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.862513426423201e-06,
+      "loss": 0.3638,
+      "step": 913
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.798066595059076e-06,
+      "loss": 0.2739,
+      "step": 914
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.733619763694953e-06,
+      "loss": 0.448,
+      "step": 915
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.669172932330826e-06,
+      "loss": 1.1904,
+      "step": 916
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.604726100966703e-06,
+      "loss": 0.4287,
+      "step": 917
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.540279269602577e-06,
+      "loss": 0.9272,
+      "step": 918
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.475832438238453e-06,
+      "loss": 1.0234,
+      "step": 919
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.411385606874329e-06,
+      "loss": 0.6655,
+      "step": 920
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.346938775510204e-06,
+      "loss": 0.4514,
+      "step": 921
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.282491944146079e-06,
+      "loss": 0.2781,
+      "step": 922
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.218045112781955e-06,
+      "loss": 0.6792,
+      "step": 923
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.153598281417831e-06,
+      "loss": 0.8042,
+      "step": 924
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.089151450053706e-06,
+      "loss": 0.1783,
+      "step": 925
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.024704618689582e-06,
+      "loss": 0.3884,
+      "step": 926
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.960257787325457e-06,
+      "loss": 0.2115,
+      "step": 927
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.895810955961332e-06,
+      "loss": 0.3872,
+      "step": 928
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.831364124597208e-06,
+      "loss": 0.9307,
+      "step": 929
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.766917293233083e-06,
+      "loss": 0.7832,
+      "step": 930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.7024704618689585e-06,
+      "loss": 0.3894,
+      "step": 931
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.6380236305048335e-06,
+      "loss": 0.2898,
+      "step": 932
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.573576799140709e-06,
+      "loss": 0.0709,
+      "step": 933
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.509129967776584e-06,
+      "loss": 0.165,
+      "step": 934
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.4446831364124594e-06,
+      "loss": 1.2168,
+      "step": 935
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.380236305048335e-06,
+      "loss": 0.2332,
+      "step": 936
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.5137,
+      "step": 937
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.251342642320086e-06,
+      "loss": 0.7983,
+      "step": 938
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.186895810955962e-06,
+      "loss": 1.165,
+      "step": 939
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 0.4495,
+      "step": 940
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.058002148227712e-06,
+      "loss": 1.4053,
+      "step": 941
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.993555316863588e-06,
+      "loss": 0.4312,
+      "step": 942
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.929108485499463e-06,
+      "loss": 0.5498,
+      "step": 943
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.864661654135338e-06,
+      "loss": 0.1597,
+      "step": 944
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.800214822771214e-06,
+      "loss": 0.2886,
+      "step": 945
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.73576799140709e-06,
+      "loss": 0.4109,
+      "step": 946
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.671321160042965e-06,
+      "loss": 0.5054,
+      "step": 947
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.60687432867884e-06,
+      "loss": 0.3936,
+      "step": 948
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.542427497314716e-06,
+      "loss": 0.3992,
+      "step": 949
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.477980665950591e-06,
+      "loss": 0.6304,
+      "step": 950
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.413533834586466e-06,
+      "loss": 0.6699,
+      "step": 951
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.349087003222342e-06,
+      "loss": 0.8867,
+      "step": 952
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.2846401718582175e-06,
+      "loss": 0.6978,
+      "step": 953
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.2201933404940925e-06,
+      "loss": 0.7178,
+      "step": 954
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.155746509129968e-06,
+      "loss": 0.7168,
+      "step": 955
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.091299677765843e-06,
+      "loss": 0.4736,
+      "step": 956
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.0268528464017184e-06,
+      "loss": 0.5347,
+      "step": 957
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.9624060150375935e-06,
+      "loss": 0.6855,
+      "step": 958
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 0.7231,
+      "step": 959
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.833512352309345e-06,
+      "loss": 0.3286,
+      "step": 960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.76906552094522e-06,
+      "loss": 0.2817,
+      "step": 961
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.704618689581096e-06,
+      "loss": 0.5527,
+      "step": 962
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.640171858216971e-06,
+      "loss": 0.2981,
+      "step": 963
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.575725026852846e-06,
+      "loss": 0.4341,
+      "step": 964
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.511278195488722e-06,
+      "loss": 0.7925,
+      "step": 965
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.446831364124597e-06,
+      "loss": 0.1185,
+      "step": 966
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.382384532760473e-06,
+      "loss": 0.1108,
+      "step": 967
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.317937701396348e-06,
+      "loss": 1.0967,
+      "step": 968
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.253490870032224e-06,
+      "loss": 0.2571,
+      "step": 969
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.189044038668099e-06,
+      "loss": 0.7031,
+      "step": 970
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.124597207303974e-06,
+      "loss": 0.9097,
+      "step": 971
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.06015037593985e-06,
+      "loss": 0.3716,
+      "step": 972
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.995703544575725e-06,
+      "loss": 0.3604,
+      "step": 973
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.931256713211601e-06,
+      "loss": 0.5859,
+      "step": 974
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.8668098818474765e-06,
+      "loss": 0.1636,
+      "step": 975
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.8023630504833515e-06,
+      "loss": 0.2457,
+      "step": 976
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7379162191192266e-06,
+      "loss": 0.5952,
+      "step": 977
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.673469387755102e-06,
+      "loss": 0.3359,
+      "step": 978
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6090225563909775e-06,
+      "loss": 0.4526,
+      "step": 979
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.544575725026853e-06,
+      "loss": 0.5771,
+      "step": 980
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.4801288936627283e-06,
+      "loss": 0.2944,
+      "step": 981
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.415682062298604e-06,
+      "loss": 0.6084,
+      "step": 982
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3512352309344792e-06,
+      "loss": 0.6567,
+      "step": 983
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.2867883995703547e-06,
+      "loss": 0.3926,
+      "step": 984
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.2223415682062297e-06,
+      "loss": 0.9121,
+      "step": 985
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.1667,
+      "step": 986
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.093447905477981e-06,
+      "loss": 1.0684,
+      "step": 987
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.029001074113856e-06,
+      "loss": 0.9414,
+      "step": 988
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.9645542427497315e-06,
+      "loss": 0.3645,
+      "step": 989
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.900107411385607e-06,
+      "loss": 1.1328,
+      "step": 990
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.8356605800214824e-06,
+      "loss": 0.5664,
+      "step": 991
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.771213748657358e-06,
+      "loss": 0.311,
+      "step": 992
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.706766917293233e-06,
+      "loss": 0.3245,
+      "step": 993
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.6423200859291087e-06,
+      "loss": 0.5825,
+      "step": 994
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.577873254564984e-06,
+      "loss": 0.1556,
+      "step": 995
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.5134264232008592e-06,
+      "loss": 0.7402,
+      "step": 996
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 0.1271,
+      "step": 997
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.38453276047261e-06,
+      "loss": 0.7485,
+      "step": 998
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.3200859291084856e-06,
+      "loss": 0.3857,
+      "step": 999
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.255639097744361e-06,
+      "loss": 0.5498,
+      "step": 1000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1911922663802365e-06,
+      "loss": 0.8418,
+      "step": 1001
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.126745435016112e-06,
+      "loss": 0.5854,
+      "step": 1002
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.062298603651987e-06,
+      "loss": 0.7441,
+      "step": 1003
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9978517722878624e-06,
+      "loss": 0.5293,
+      "step": 1004
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9334049409237383e-06,
+      "loss": 0.8838,
+      "step": 1005
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8689581095596133e-06,
+      "loss": 0.5181,
+      "step": 1006
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8045112781954887e-06,
+      "loss": 0.8169,
+      "step": 1007
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.7400644468313642e-06,
+      "loss": 0.6387,
+      "step": 1008
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6756176154672396e-06,
+      "loss": 1.4199,
+      "step": 1009
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6111707841031149e-06,
+      "loss": 0.7676,
+      "step": 1010
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5467239527389905e-06,
+      "loss": 0.4863,
+      "step": 1011
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4822771213748658e-06,
+      "loss": 0.5029,
+      "step": 1012
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4178302900107412e-06,
+      "loss": 0.5166,
+      "step": 1013
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3533834586466164e-06,
+      "loss": 0.4446,
+      "step": 1014
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.288936627282492e-06,
+      "loss": 0.4229,
+      "step": 1015
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 0.7046,
+      "step": 1016
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1600429645542428e-06,
+      "loss": 0.1925,
+      "step": 1017
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0955961331901182e-06,
+      "loss": 0.3721,
+      "step": 1018
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0311493018259935e-06,
+      "loss": 0.1495,
+      "step": 1019
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.667024704618691e-07,
+      "loss": 0.4629,
+      "step": 1020
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.022556390977444e-07,
+      "loss": 1.1387,
+      "step": 1021
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.378088077336198e-07,
+      "loss": 1.4092,
+      "step": 1022
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.733619763694953e-07,
+      "loss": 0.4365,
+      "step": 1023
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.089151450053706e-07,
+      "loss": 1.0098,
+      "step": 1024
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.44468313641246e-07,
+      "loss": 0.9849,
+      "step": 1025
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.800214822771214e-07,
+      "loss": 0.4924,
+      "step": 1026
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.155746509129967e-07,
+      "loss": 0.8003,
+      "step": 1027
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.511278195488722e-07,
+      "loss": 0.7852,
+      "step": 1028
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.8668098818474763e-07,
+      "loss": 0.0843,
+      "step": 1029
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.22234156820623e-07,
+      "loss": 0.7656,
+      "step": 1030
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5778732545649837e-07,
+      "loss": 0.2798,
+      "step": 1031
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9334049409237381e-07,
+      "loss": 0.5186,
+      "step": 1032
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2889366272824918e-07,
+      "loss": 0.6855,
+      "step": 1033
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.444683136412459e-08,
+      "loss": 0.3208,
+      "step": 1034
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.5391,
+      "step": 1035
+    },
+    {
+      "epoch": 3.0,
+      "step": 1035,
+      "total_flos": 6.646386111817974e+18,
+      "train_loss": 0.9942761997093901,
+      "train_runtime": 437.6871,
+      "train_samples_per_second": 606.762,
+      "train_steps_per_second": 2.365
+    }
+  ],
+  "max_steps": 1035,
+  "num_train_epochs": 3,
+  "total_flos": 6.646386111817974e+18,
+  "trial_name": null,
+  "trial_params": null
+}