diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9277 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 1542,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2653721682847896e-07,
+      "loss": 5.9961,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.530744336569579e-07,
+      "loss": 6.0273,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.796116504854367e-07,
+      "loss": 5.9609,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.061488673139158e-07,
+      "loss": 6.0117,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1326860841423946e-06,
+      "loss": 6.0156,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3592233009708735e-06,
+      "loss": 6.0234,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5857605177993528e-06,
+      "loss": 5.9766,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8122977346278317e-06,
+      "loss": 5.9766,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0388349514563107e-06,
+      "loss": 5.9531,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.265372168284789e-06,
+      "loss": 5.9805,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4919093851132685e-06,
+      "loss": 5.9648,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.718446601941747e-06,
+      "loss": 5.9883,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9449838187702263e-06,
+      "loss": 5.9297,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1715210355987056e-06,
+      "loss": 5.9609,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.398058252427184e-06,
+      "loss": 5.9141,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6245954692556633e-06,
+      "loss": 5.9297,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.851132686084142e-06,
+      "loss": 5.9414,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.0776699029126215e-06,
+      "loss": 5.8984,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3042071197411e-06,
+      "loss": 5.8867,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.530744336569578e-06,
+      "loss": 5.8359,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.757281553398058e-06,
+      "loss": 5.8125,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983818770226537e-06,
+      "loss": 5.7969,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.210355987055016e-06,
+      "loss": 5.8203,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.436893203883494e-06,
+      "loss": 5.8125,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.663430420711973e-06,
+      "loss": 5.7812,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.8899676375404525e-06,
+      "loss": 5.7422,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.116504854368932e-06,
+      "loss": 5.6953,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.343042071197411e-06,
+      "loss": 5.6719,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.569579288025889e-06,
+      "loss": 5.6602,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.796116504854368e-06,
+      "loss": 5.5977,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.022653721682847e-06,
+      "loss": 5.5195,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.249190938511327e-06,
+      "loss": 5.4961,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.475728155339805e-06,
+      "loss": 5.4258,
+      "step": 33
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.702265372168284e-06,
+      "loss": 5.3281,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.928802588996764e-06,
+      "loss": 5.3008,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.155339805825243e-06,
+      "loss": 5.2266,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.38187702265372e-06,
+      "loss": 5.0859,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.6084142394822e-06,
+      "loss": 4.9844,
+      "step": 38
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.83495145631068e-06,
+      "loss": 5.0039,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.061488673139157e-06,
+      "loss": 4.7812,
+      "step": 40
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.288025889967638e-06,
+      "loss": 4.7578,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.514563106796115e-06,
+      "loss": 4.7695,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.741100323624595e-06,
+      "loss": 4.8906,
+      "step": 43
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967637540453074e-06,
+      "loss": 4.7578,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0194174757281552e-05,
+      "loss": 4.582,
+      "step": 45
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0420711974110033e-05,
+      "loss": 4.3047,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.064724919093851e-05,
+      "loss": 4.582,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0873786407766988e-05,
+      "loss": 4.3242,
+      "step": 48
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1100323624595469e-05,
+      "loss": 4.0664,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1326860841423946e-05,
+      "loss": 4.1875,
+      "step": 50
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1553398058252427e-05,
+      "loss": 3.9863,
+      "step": 51
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1779935275080905e-05,
+      "loss": 4.0781,
+      "step": 52
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2006472491909383e-05,
+      "loss": 3.9512,
+      "step": 53
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2233009708737864e-05,
+      "loss": 3.9141,
+      "step": 54
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2459546925566341e-05,
+      "loss": 3.6641,
+      "step": 55
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2686084142394822e-05,
+      "loss": 3.1035,
+      "step": 56
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.29126213592233e-05,
+      "loss": 3.7227,
+      "step": 57
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3139158576051777e-05,
+      "loss": 3.2754,
+      "step": 58
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3365695792880258e-05,
+      "loss": 3.6465,
+      "step": 59
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3592233009708736e-05,
+      "loss": 3.1211,
+      "step": 60
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3818770226537215e-05,
+      "loss": 3.4707,
+      "step": 61
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4045307443365695e-05,
+      "loss": 3.1699,
+      "step": 62
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4271844660194172e-05,
+      "loss": 3.7422,
+      "step": 63
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4498381877022653e-05,
+      "loss": 2.7637,
+      "step": 64
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.472491909385113e-05,
+      "loss": 3.2949,
+      "step": 65
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.495145631067961e-05,
+      "loss": 2.543,
+      "step": 66
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.517799352750809e-05,
+      "loss": 2.0547,
+      "step": 67
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.540453074433657e-05,
+      "loss": 2.2539,
+      "step": 68
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5631067961165046e-05,
+      "loss": 2.7383,
+      "step": 69
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5857605177993527e-05,
+      "loss": 2.2031,
+      "step": 70
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6084142394822005e-05,
+      "loss": 2.873,
+      "step": 71
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6310679611650486e-05,
+      "loss": 2.2285,
+      "step": 72
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6537216828478964e-05,
+      "loss": 2.5234,
+      "step": 73
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.676375404530744e-05,
+      "loss": 1.9932,
+      "step": 74
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6990291262135922e-05,
+      "loss": 1.793,
+      "step": 75
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.72168284789644e-05,
+      "loss": 1.9678,
+      "step": 76
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.744336569579288e-05,
+      "loss": 2.1387,
+      "step": 77
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.766990291262136e-05,
+      "loss": 2.0762,
+      "step": 78
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.789644012944984e-05,
+      "loss": 1.7725,
+      "step": 79
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8122977346278314e-05,
+      "loss": 1.7256,
+      "step": 80
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8349514563106795e-05,
+      "loss": 1.9189,
+      "step": 81
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8576051779935276e-05,
+      "loss": 2.3809,
+      "step": 82
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880258899676375e-05,
+      "loss": 2.0996,
+      "step": 83
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.902912621359223e-05,
+      "loss": 1.6475,
+      "step": 84
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9255663430420712e-05,
+      "loss": 2.9414,
+      "step": 85
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.948220064724919e-05,
+      "loss": 2.2305,
+      "step": 86
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9708737864077667e-05,
+      "loss": 2.3281,
+      "step": 87
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9935275080906148e-05,
+      "loss": 2.2109,
+      "step": 88
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.016181229773463e-05,
+      "loss": 1.7871,
+      "step": 89
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0388349514563103e-05,
+      "loss": 1.6367,
+      "step": 90
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0614886731391584e-05,
+      "loss": 1.9697,
+      "step": 91
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0841423948220065e-05,
+      "loss": 1.7139,
+      "step": 92
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.106796116504854e-05,
+      "loss": 2.2852,
+      "step": 93
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.129449838187702e-05,
+      "loss": 1.6562,
+      "step": 94
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.15210355987055e-05,
+      "loss": 1.9287,
+      "step": 95
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1747572815533976e-05,
+      "loss": 1.126,
+      "step": 96
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1974110032362457e-05,
+      "loss": 1.7559,
+      "step": 97
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.2200647249190938e-05,
+      "loss": 1.9482,
+      "step": 98
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.242718446601942e-05,
+      "loss": 1.5293,
+      "step": 99
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.2653721682847893e-05,
+      "loss": 2.0527,
+      "step": 100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2880258899676374e-05,
+      "loss": 1.6289,
+      "step": 101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.3106796116504855e-05,
+      "loss": 1.6133,
+      "step": 102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.333333333333333e-05,
+      "loss": 1.7148,
+      "step": 103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.355987055016181e-05,
+      "loss": 1.4717,
+      "step": 104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.378640776699029e-05,
+      "loss": 1.8027,
+      "step": 105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.4012944983818765e-05,
+      "loss": 2.1504,
+      "step": 106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.4239482200647246e-05,
+      "loss": 1.665,
+      "step": 107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.4466019417475727e-05,
+      "loss": 1.4902,
+      "step": 108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.46925566343042e-05,
+      "loss": 1.6582,
+      "step": 109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.4919093851132682e-05,
+      "loss": 1.6836,
+      "step": 110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.5145631067961163e-05,
+      "loss": 1.25,
+      "step": 111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.5372168284789644e-05,
+      "loss": 1.5781,
+      "step": 112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.559870550161812e-05,
+      "loss": 1.4697,
+      "step": 113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.58252427184466e-05,
+      "loss": 1.6738,
+      "step": 114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.605177993527508e-05,
+      "loss": 1.8496,
+      "step": 115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6278317152103555e-05,
+      "loss": 1.0762,
+      "step": 116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6504854368932036e-05,
+      "loss": 1.8564,
+      "step": 117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6731391585760517e-05,
+      "loss": 1.3311,
+      "step": 118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6957928802588994e-05,
+      "loss": 1.7188,
+      "step": 119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7184466019417472e-05,
+      "loss": 1.6914,
+      "step": 120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7411003236245953e-05,
+      "loss": 1.3633,
+      "step": 121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.763754045307443e-05,
+      "loss": 1.2354,
+      "step": 122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7864077669902908e-05,
+      "loss": 1.3633,
+      "step": 123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.809061488673139e-05,
+      "loss": 1.9648,
+      "step": 124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.831715210355987e-05,
+      "loss": 1.8584,
+      "step": 125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8543689320388344e-05,
+      "loss": 2.3164,
+      "step": 126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8770226537216825e-05,
+      "loss": 1.4102,
+      "step": 127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8996763754045306e-05,
+      "loss": 1.6836,
+      "step": 128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9223300970873784e-05,
+      "loss": 2.4863,
+      "step": 129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.944983818770226e-05,
+      "loss": 1.1318,
+      "step": 130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9676375404530743e-05,
+      "loss": 1.748,
+      "step": 131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.990291262135922e-05,
+      "loss": 1.0938,
+      "step": 132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.0129449838187698e-05,
+      "loss": 1.2949,
+      "step": 133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.035598705501618e-05,
+      "loss": 0.9761,
+      "step": 134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.058252427184465e-05,
+      "loss": 1.5547,
+      "step": 135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.080906148867314e-05,
+      "loss": 1.0957,
+      "step": 136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.1035598705501615e-05,
+      "loss": 1.0596,
+      "step": 137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.126213592233009e-05,
+      "loss": 1.627,
+      "step": 138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.148867313915857e-05,
+      "loss": 1.3721,
+      "step": 139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.1715210355987055e-05,
+      "loss": 1.8223,
+      "step": 140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.194174757281553e-05,
+      "loss": 1.2832,
+      "step": 141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.216828478964401e-05,
+      "loss": 1.8086,
+      "step": 142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.239482200647249e-05,
+      "loss": 1.6064,
+      "step": 143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.262135922330097e-05,
+      "loss": 1.2197,
+      "step": 144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.284789644012944e-05,
+      "loss": 1.3887,
+      "step": 145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.307443365695793e-05,
+      "loss": 1.2725,
+      "step": 146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3300970873786405e-05,
+      "loss": 0.9053,
+      "step": 147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.352750809061488e-05,
+      "loss": 1.4629,
+      "step": 148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.375404530744336e-05,
+      "loss": 1.2461,
+      "step": 149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3980582524271844e-05,
+      "loss": 1.4775,
+      "step": 150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.420711974110032e-05,
+      "loss": 1.4043,
+      "step": 151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.44336569579288e-05,
+      "loss": 1.2822,
+      "step": 152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.466019417475728e-05,
+      "loss": 1.0049,
+      "step": 153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.488673139158576e-05,
+      "loss": 1.3574,
+      "step": 154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.511326860841424e-05,
+      "loss": 0.7646,
+      "step": 155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.533980582524272e-05,
+      "loss": 1.3389,
+      "step": 156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5566343042071194e-05,
+      "loss": 1.0557,
+      "step": 157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.579288025889968e-05,
+      "loss": 0.9287,
+      "step": 158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.601941747572815e-05,
+      "loss": 1.0508,
+      "step": 159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.624595469255663e-05,
+      "loss": 1.3984,
+      "step": 160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.647249190938511e-05,
+      "loss": 1.3623,
+      "step": 161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.669902912621359e-05,
+      "loss": 1.3555,
+      "step": 162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.692556634304207e-05,
+      "loss": 1.5928,
+      "step": 163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.715210355987055e-05,
+      "loss": 1.3867,
+      "step": 164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.737864077669902e-05,
+      "loss": 0.9873,
+      "step": 165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.76051779935275e-05,
+      "loss": 1.1855,
+      "step": 166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.7831715210355984e-05,
+      "loss": 1.3008,
+      "step": 167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.805825242718446e-05,
+      "loss": 0.9136,
+      "step": 168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8284789644012946e-05,
+      "loss": 1.3359,
+      "step": 169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8511326860841424e-05,
+      "loss": 1.208,
+      "step": 170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.87378640776699e-05,
+      "loss": 1.5762,
+      "step": 171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.896440129449838e-05,
+      "loss": 1.8926,
+      "step": 172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9190938511326856e-05,
+      "loss": 1.4688,
+      "step": 173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9417475728155334e-05,
+      "loss": 0.9941,
+      "step": 174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.964401294498382e-05,
+      "loss": 1.6846,
+      "step": 175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9870550161812296e-05,
+      "loss": 1.5684,
+      "step": 176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.0097087378640774e-05,
+      "loss": 1.208,
+      "step": 177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.032362459546926e-05,
+      "loss": 1.5352,
+      "step": 178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.055016181229773e-05,
+      "loss": 1.0684,
+      "step": 179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.0776699029126206e-05,
+      "loss": 1.0049,
+      "step": 180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.100323624595469e-05,
+      "loss": 1.4014,
+      "step": 181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.122977346278317e-05,
+      "loss": 1.1924,
+      "step": 182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.1456310679611646e-05,
+      "loss": 1.4521,
+      "step": 183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.168284789644013e-05,
+      "loss": 0.7993,
+      "step": 184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.19093851132686e-05,
+      "loss": 1.0498,
+      "step": 185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.213592233009708e-05,
+      "loss": 0.9438,
+      "step": 186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.236245954692556e-05,
+      "loss": 1.625,
+      "step": 187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.258899676375404e-05,
+      "loss": 1.2715,
+      "step": 188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.281553398058252e-05,
+      "loss": 0.9678,
+      "step": 189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3042071197411e-05,
+      "loss": 1.1182,
+      "step": 190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.326860841423948e-05,
+      "loss": 1.2285,
+      "step": 191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.349514563106795e-05,
+      "loss": 0.5293,
+      "step": 192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3721682847896436e-05,
+      "loss": 1.1416,
+      "step": 193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.394822006472491e-05,
+      "loss": 1.1992,
+      "step": 194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.41747572815534e-05,
+      "loss": 1.543,
+      "step": 195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.4401294498381875e-05,
+      "loss": 1.6172,
+      "step": 196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.462783171521035e-05,
+      "loss": 1.3574,
+      "step": 197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.485436893203884e-05,
+      "loss": 1.5391,
+      "step": 198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.508090614886731e-05,
+      "loss": 1.4824,
+      "step": 199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5307443365695786e-05,
+      "loss": 0.9688,
+      "step": 200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.553398058252427e-05,
+      "loss": 1.125,
+      "step": 201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.576051779935275e-05,
+      "loss": 1.1611,
+      "step": 202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5987055016181225e-05,
+      "loss": 1.2461,
+      "step": 203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.621359223300971e-05,
+      "loss": 1.0439,
+      "step": 204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.644012944983818e-05,
+      "loss": 0.8789,
+      "step": 205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.666666666666666e-05,
+      "loss": 1.8975,
+      "step": 206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.689320388349514e-05,
+      "loss": 1.3994,
+      "step": 207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.711974110032362e-05,
+      "loss": 1.1035,
+      "step": 208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.73462783171521e-05,
+      "loss": 1.1377,
+      "step": 209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.757281553398058e-05,
+      "loss": 1.8164,
+      "step": 210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.779935275080906e-05,
+      "loss": 0.9883,
+      "step": 211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.802588996763753e-05,
+      "loss": 1.7783,
+      "step": 212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8252427184466015e-05,
+      "loss": 1.0312,
+      "step": 213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.847896440129449e-05,
+      "loss": 0.605,
+      "step": 214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.870550161812297e-05,
+      "loss": 1.0938,
+      "step": 215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8932038834951454e-05,
+      "loss": 1.2627,
+      "step": 216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.915857605177993e-05,
+      "loss": 1.2236,
+      "step": 217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.93851132686084e-05,
+      "loss": 1.2236,
+      "step": 218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.961165048543689e-05,
+      "loss": 1.1504,
+      "step": 219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9838187702265365e-05,
+      "loss": 1.667,
+      "step": 220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.006472491909385e-05,
+      "loss": 1.3867,
+      "step": 221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.029126213592233e-05,
+      "loss": 0.6914,
+      "step": 222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.0517799352750804e-05,
+      "loss": 0.7217,
+      "step": 223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.074433656957929e-05,
+      "loss": 0.9858,
+      "step": 224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.0970873786407766e-05,
+      "loss": 1.0264,
+      "step": 225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.119741100323624e-05,
+      "loss": 1.4688,
+      "step": 226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.142394822006472e-05,
+      "loss": 1.1348,
+      "step": 227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.16504854368932e-05,
+      "loss": 0.8359,
+      "step": 228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.187702265372168e-05,
+      "loss": 1.4746,
+      "step": 229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.210355987055016e-05,
+      "loss": 1.4648,
+      "step": 230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.233009708737864e-05,
+      "loss": 0.9995,
+      "step": 231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.255663430420711e-05,
+      "loss": 1.2109,
+      "step": 232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.2783171521035594e-05,
+      "loss": 1.0557,
+      "step": 233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.300970873786407e-05,
+      "loss": 0.896,
+      "step": 234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.323624595469255e-05,
+      "loss": 0.5439,
+      "step": 235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.3462783171521034e-05,
+      "loss": 1.4521,
+      "step": 236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.368932038834951e-05,
+      "loss": 1.2764,
+      "step": 237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.391585760517799e-05,
+      "loss": 1.1455,
+      "step": 238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.4142394822006467e-05,
+      "loss": 1.6074,
+      "step": 239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4368932038834944e-05,
+      "loss": 1.1514,
+      "step": 240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.459546925566342e-05,
+      "loss": 1.6689,
+      "step": 241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4822006472491906e-05,
+      "loss": 1.2422,
+      "step": 242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5048543689320384e-05,
+      "loss": 1.3496,
+      "step": 243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.527508090614886e-05,
+      "loss": 0.7944,
+      "step": 244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5501618122977346e-05,
+      "loss": 1.1084,
+      "step": 245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5728155339805817e-05,
+      "loss": 1.0918,
+      "step": 246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.59546925566343e-05,
+      "loss": 1.4404,
+      "step": 247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.618122977346278e-05,
+      "loss": 0.9751,
+      "step": 248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6407766990291256e-05,
+      "loss": 1.1484,
+      "step": 249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.663430420711974e-05,
+      "loss": 0.5786,
+      "step": 250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.686084142394822e-05,
+      "loss": 1.1953,
+      "step": 251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.708737864077669e-05,
+      "loss": 1.1045,
+      "step": 252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.731391585760517e-05,
+      "loss": 0.9697,
+      "step": 253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.754045307443365e-05,
+      "loss": 1.0615,
+      "step": 254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776699029126213e-05,
+      "loss": 1.126,
+      "step": 255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.799352750809061e-05,
+      "loss": 0.4761,
+      "step": 256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.822006472491909e-05,
+      "loss": 1.0898,
+      "step": 257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.844660194174757e-05,
+      "loss": 0.8623,
+      "step": 258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8673139158576046e-05,
+      "loss": 1.2393,
+      "step": 259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.889967637540452e-05,
+      "loss": 0.7666,
+      "step": 260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9126213592233e-05,
+      "loss": 0.5508,
+      "step": 261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9352750809061485e-05,
+      "loss": 0.8848,
+      "step": 262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.957928802588996e-05,
+      "loss": 0.7822,
+      "step": 263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.980582524271844e-05,
+      "loss": 1.6299,
+      "step": 264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.0032362459546925e-05,
+      "loss": 0.7979,
+      "step": 265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.0258899676375396e-05,
+      "loss": 1.1875,
+      "step": 266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.0485436893203873e-05,
+      "loss": 1.085,
+      "step": 267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.071197411003236e-05,
+      "loss": 1.5332,
+      "step": 268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.0938511326860835e-05,
+      "loss": 1.5684,
+      "step": 269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.11650485436893e-05,
+      "loss": 0.999,
+      "step": 270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.139158576051779e-05,
+      "loss": 1.1016,
+      "step": 271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.161812297734627e-05,
+      "loss": 0.9336,
+      "step": 272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.184466019417476e-05,
+      "loss": 0.6851,
+      "step": 273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.207119741100323e-05,
+      "loss": 0.5234,
+      "step": 274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.229773462783171e-05,
+      "loss": 2.0,
+      "step": 275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.252427184466019e-05,
+      "loss": 0.978,
+      "step": 276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.275080906148867e-05,
+      "loss": 1.3438,
+      "step": 277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.297734627831714e-05,
+      "loss": 1.2803,
+      "step": 278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.320388349514563e-05,
+      "loss": 0.9878,
+      "step": 279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.343042071197411e-05,
+      "loss": 0.6392,
+      "step": 280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.365695792880258e-05,
+      "loss": 0.8438,
+      "step": 281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.388349514563106e-05,
+      "loss": 0.8643,
+      "step": 282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.411003236245954e-05,
+      "loss": 1.2822,
+      "step": 283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.433656957928802e-05,
+      "loss": 0.8174,
+      "step": 284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.45631067961165e-05,
+      "loss": 1.1475,
+      "step": 285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.478964401294498e-05,
+      "loss": 0.9644,
+      "step": 286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.501618122977346e-05,
+      "loss": 0.5596,
+      "step": 287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.524271844660194e-05,
+      "loss": 0.873,
+      "step": 288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.546925566343041e-05,
+      "loss": 1.1631,
+      "step": 289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.569579288025889e-05,
+      "loss": 1.5566,
+      "step": 290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.592233009708737e-05,
+      "loss": 0.8647,
+      "step": 291
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.614886731391585e-05,
+      "loss": 1.626,
+      "step": 292
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.637540453074433e-05,
+      "loss": 0.9136,
+      "step": 293
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.660194174757281e-05,
+      "loss": 1.1504,
+      "step": 294
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.68284789644013e-05,
+      "loss": 1.0781,
+      "step": 295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.705501618122976e-05,
+      "loss": 0.668,
+      "step": 296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.728155339805825e-05,
+      "loss": 1.207,
+      "step": 297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.750809061488672e-05,
+      "loss": 1.0244,
+      "step": 298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.77346278317152e-05,
+      "loss": 0.8579,
+      "step": 299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.796116504854369e-05,
+      "loss": 1.2969,
+      "step": 300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.818770226537216e-05,
+      "loss": 1.0801,
+      "step": 301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.841423948220064e-05,
+      "loss": 0.9541,
+      "step": 302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.864077669902911e-05,
+      "loss": 1.0059,
+      "step": 303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.88673139158576e-05,
+      "loss": 0.9102,
+      "step": 304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.909385113268608e-05,
+      "loss": 0.7891,
+      "step": 305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.932038834951455e-05,
+      "loss": 0.9858,
+      "step": 306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.954692556634304e-05,
+      "loss": 1.2471,
+      "step": 307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.977346278317152e-05,
+      "loss": 0.9438,
+      "step": 308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7e-05,
+      "loss": 1.0449,
+      "step": 309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.994322789943227e-05,
+      "loss": 0.7666,
+      "step": 310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988645579886455e-05,
+      "loss": 1.9082,
+      "step": 311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982968369829683e-05,
+      "loss": 0.8555,
+      "step": 312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97729115977291e-05,
+      "loss": 1.2793,
+      "step": 313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971613949716139e-05,
+      "loss": 0.6646,
+      "step": 314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965936739659366e-05,
+      "loss": 0.8003,
+      "step": 315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960259529602595e-05,
+      "loss": 1.1543,
+      "step": 316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.954582319545822e-05,
+      "loss": 0.9565,
+      "step": 317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.948905109489051e-05,
+      "loss": 1.5156,
+      "step": 318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.943227899432278e-05,
+      "loss": 1.6465,
+      "step": 319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.937550689375507e-05,
+      "loss": 1.3076,
+      "step": 320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.931873479318734e-05,
+      "loss": 0.8125,
+      "step": 321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.926196269261962e-05,
+      "loss": 0.7104,
+      "step": 322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.92051905920519e-05,
+      "loss": 1.3789,
+      "step": 323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.914841849148418e-05,
+      "loss": 0.9189,
+      "step": 324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.909164639091645e-05,
+      "loss": 0.8179,
+      "step": 325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.903487429034874e-05,
+      "loss": 1.8174,
+      "step": 326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.897810218978101e-05,
+      "loss": 0.769,
+      "step": 327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.89213300892133e-05,
+      "loss": 1.6465,
+      "step": 328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.886455798864557e-05,
+      "loss": 0.8848,
+      "step": 329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.880778588807786e-05,
+      "loss": 1.2383,
+      "step": 330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.875101378751013e-05,
+      "loss": 1.209,
+      "step": 331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.869424168694241e-05,
+      "loss": 0.9526,
+      "step": 332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.86374695863747e-05,
+      "loss": 0.6997,
+      "step": 333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.858069748580697e-05,
+      "loss": 1.3457,
+      "step": 334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.852392538523924e-05,
+      "loss": 0.8306,
+      "step": 335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.846715328467153e-05,
+      "loss": 1.21,
+      "step": 336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.84103811841038e-05,
+      "loss": 1.6484,
+      "step": 337
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.835360908353608e-05,
+      "loss": 1.0049,
+      "step": 338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.829683698296836e-05,
+      "loss": 0.9609,
+      "step": 339
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.824006488240064e-05,
+      "loss": 1.5117,
+      "step": 340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.818329278183292e-05,
+      "loss": 1.1641,
+      "step": 341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.81265206812652e-05,
+      "loss": 1.0146,
+      "step": 342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.806974858069749e-05,
+      "loss": 0.5044,
+      "step": 343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.801297648012976e-05,
+      "loss": 0.5928,
+      "step": 344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.795620437956205e-05,
+      "loss": 0.7925,
+      "step": 345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.789943227899432e-05,
+      "loss": 0.8438,
+      "step": 346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.784266017842659e-05,
+      "loss": 0.853,
+      "step": 347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.778588807785888e-05,
+      "loss": 0.5464,
+      "step": 348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.772911597729115e-05,
+      "loss": 1.1338,
+      "step": 349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.767234387672343e-05,
+      "loss": 1.0312,
+      "step": 350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.761557177615571e-05,
+      "loss": 0.9209,
+      "step": 351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.755879967558799e-05,
+      "loss": 0.8164,
+      "step": 352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.750202757502026e-05,
+      "loss": 1.0479,
+      "step": 353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.744525547445255e-05,
+      "loss": 1.4258,
+      "step": 354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.738848337388482e-05,
+      "loss": 0.6206,
+      "step": 355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.733171127331711e-05,
+      "loss": 0.7578,
+      "step": 356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.727493917274938e-05,
+      "loss": 0.9341,
+      "step": 357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.721816707218167e-05,
+      "loss": 1.1143,
+      "step": 358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.716139497161394e-05,
+      "loss": 1.0234,
+      "step": 359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.710462287104623e-05,
+      "loss": 1.3389,
+      "step": 360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.70478507704785e-05,
+      "loss": 1.752,
+      "step": 361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.699107866991078e-05,
+      "loss": 1.4131,
+      "step": 362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.693430656934307e-05,
+      "loss": 0.7134,
+      "step": 363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.687753446877534e-05,
+      "loss": 1.1045,
+      "step": 364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.682076236820761e-05,
+      "loss": 0.6382,
+      "step": 365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.67639902676399e-05,
+      "loss": 0.6548,
+      "step": 366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.670721816707217e-05,
+      "loss": 0.9341,
+      "step": 367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.665044606650446e-05,
+      "loss": 0.9302,
+      "step": 368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.659367396593673e-05,
+      "loss": 1.0996,
+      "step": 369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.653690186536902e-05,
+      "loss": 1.1855,
+      "step": 370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.64801297648013e-05,
+      "loss": 0.7583,
+      "step": 371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.642335766423357e-05,
+      "loss": 0.7979,
+      "step": 372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.636658556366586e-05,
+      "loss": 1.2578,
+      "step": 373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.630981346309813e-05,
+      "loss": 0.7188,
+      "step": 374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.62530413625304e-05,
+      "loss": 1.3262,
+      "step": 375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.619626926196269e-05,
+      "loss": 0.728,
+      "step": 376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.613949716139496e-05,
+      "loss": 0.957,
+      "step": 377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.608272506082724e-05,
+      "loss": 1.0469,
+      "step": 378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.602595296025952e-05,
+      "loss": 0.6592,
+      "step": 379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.59691808596918e-05,
+      "loss": 0.999,
+      "step": 380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.591240875912408e-05,
+      "loss": 0.9458,
+      "step": 381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.585563665855636e-05,
+      "loss": 0.8193,
+      "step": 382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.579886455798865e-05,
+      "loss": 1.4395,
+      "step": 383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.574209245742092e-05,
+      "loss": 1.1357,
+      "step": 384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.56853203568532e-05,
+      "loss": 0.6948,
+      "step": 385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.562854825628548e-05,
+      "loss": 1.4521,
+      "step": 386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.557177615571775e-05,
+      "loss": 2.0938,
+      "step": 387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.551500405515004e-05,
+      "loss": 0.8203,
+      "step": 388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.545823195458231e-05,
+      "loss": 0.918,
+      "step": 389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.540145985401459e-05,
+      "loss": 0.666,
+      "step": 390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.534468775344687e-05,
+      "loss": 0.8672,
+      "step": 391
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.528791565287915e-05,
+      "loss": 0.6846,
+      "step": 392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.523114355231142e-05,
+      "loss": 0.7441,
+      "step": 393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.517437145174371e-05,
+      "loss": 0.6484,
+      "step": 394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.511759935117598e-05,
+      "loss": 1.6113,
+      "step": 395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.506082725060827e-05,
+      "loss": 0.9854,
+      "step": 396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.500405515004054e-05,
+      "loss": 0.9419,
+      "step": 397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.494728304947283e-05,
+      "loss": 0.6738,
+      "step": 398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.48905109489051e-05,
+      "loss": 0.813,
+      "step": 399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.483373884833739e-05,
+      "loss": 1.2549,
+      "step": 400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.477696674776966e-05,
+      "loss": 0.5796,
+      "step": 401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.472019464720194e-05,
+      "loss": 0.6401,
+      "step": 402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.466342254663421e-05,
+      "loss": 0.6421,
+      "step": 403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.46066504460665e-05,
+      "loss": 0.8018,
+      "step": 404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.454987834549877e-05,
+      "loss": 1.9502,
+      "step": 405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.449310624493106e-05,
+      "loss": 1.3848,
+      "step": 406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.443633414436333e-05,
+      "loss": 0.8154,
+      "step": 407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.437956204379562e-05,
+      "loss": 0.8955,
+      "step": 408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.43227899432279e-05,
+      "loss": 0.9185,
+      "step": 409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.426601784266018e-05,
+      "loss": 0.7637,
+      "step": 410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.420924574209245e-05,
+      "loss": 1.5586,
+      "step": 411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.415247364152473e-05,
+      "loss": 1.1768,
+      "step": 412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.409570154095702e-05,
+      "loss": 1.0439,
+      "step": 413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.403892944038929e-05,
+      "loss": 0.7505,
+      "step": 414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.398215733982156e-05,
+      "loss": 0.5596,
+      "step": 415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.392538523925385e-05,
+      "loss": 1.1162,
+      "step": 416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.386861313868612e-05,
+      "loss": 1.2373,
+      "step": 417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.38118410381184e-05,
+      "loss": 1.1953,
+      "step": 418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.375506893755068e-05,
+      "loss": 0.9351,
+      "step": 419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.369829683698296e-05,
+      "loss": 0.7739,
+      "step": 420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.364152473641524e-05,
+      "loss": 1.2188,
+      "step": 421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.358475263584752e-05,
+      "loss": 0.5005,
+      "step": 422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.35279805352798e-05,
+      "loss": 0.8481,
+      "step": 423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.347120843471208e-05,
+      "loss": 1.125,
+      "step": 424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.341443633414437e-05,
+      "loss": 0.5742,
+      "step": 425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.335766423357664e-05,
+      "loss": 0.8838,
+      "step": 426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.330089213300891e-05,
+      "loss": 1.1221,
+      "step": 427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.32441200324412e-05,
+      "loss": 1.1895,
+      "step": 428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.318734793187347e-05,
+      "loss": 1.0889,
+      "step": 429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.313057583130575e-05,
+      "loss": 0.5264,
+      "step": 430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.307380373073803e-05,
+      "loss": 1.124,
+      "step": 431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.301703163017031e-05,
+      "loss": 1.2236,
+      "step": 432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.296025952960258e-05,
+      "loss": 1.1484,
+      "step": 433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.290348742903487e-05,
+      "loss": 1.4531,
+      "step": 434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.284671532846714e-05,
+      "loss": 0.7983,
+      "step": 435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.278994322789943e-05,
+      "loss": 1.2295,
+      "step": 436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.27331711273317e-05,
+      "loss": 0.9453,
+      "step": 437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.267639902676399e-05,
+      "loss": 1.1094,
+      "step": 438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.261962692619626e-05,
+      "loss": 0.7842,
+      "step": 439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.256285482562854e-05,
+      "loss": 0.8726,
+      "step": 440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.250608272506082e-05,
+      "loss": 0.8071,
+      "step": 441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.24493106244931e-05,
+      "loss": 1.0039,
+      "step": 442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.239253852392537e-05,
+      "loss": 0.8965,
+      "step": 443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.233576642335766e-05,
+      "loss": 1.5293,
+      "step": 444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.227899432278993e-05,
+      "loss": 1.6602,
+      "step": 445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.222222222222222e-05,
+      "loss": 1.1553,
+      "step": 446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.21654501216545e-05,
+      "loss": 0.9912,
+      "step": 447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.210867802108678e-05,
+      "loss": 0.6924,
+      "step": 448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.205190592051905e-05,
+      "loss": 1.1436,
+      "step": 449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.199513381995134e-05,
+      "loss": 0.9307,
+      "step": 450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.193836171938361e-05,
+      "loss": 1.2578,
+      "step": 451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.188158961881589e-05,
+      "loss": 1.0254,
+      "step": 452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.182481751824818e-05,
+      "loss": 1.1309,
+      "step": 453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.176804541768045e-05,
+      "loss": 1.0244,
+      "step": 454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.171127331711272e-05,
+      "loss": 0.6079,
+      "step": 455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.165450121654501e-05,
+      "loss": 0.5811,
+      "step": 456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.159772911597728e-05,
+      "loss": 0.7163,
+      "step": 457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.154095701540956e-05,
+      "loss": 0.6978,
+      "step": 458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.148418491484184e-05,
+      "loss": 0.7197,
+      "step": 459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.142741281427412e-05,
+      "loss": 0.9082,
+      "step": 460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.13706407137064e-05,
+      "loss": 0.5596,
+      "step": 461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.131386861313868e-05,
+      "loss": 0.5796,
+      "step": 462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.125709651257097e-05,
+      "loss": 1.249,
+      "step": 463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.120032441200324e-05,
+      "loss": 0.6138,
+      "step": 464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.114355231143553e-05,
+      "loss": 0.7173,
+      "step": 465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.10867802108678e-05,
+      "loss": 1.2246,
+      "step": 466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.1030008110300073e-05,
+      "loss": 1.167,
+      "step": 467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.097323600973236e-05,
+      "loss": 0.604,
+      "step": 468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.0916463909164634e-05,
+      "loss": 0.5674,
+      "step": 469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.085969180859691e-05,
+      "loss": 0.6787,
+      "step": 470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.0802919708029195e-05,
+      "loss": 0.7441,
+      "step": 471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.074614760746147e-05,
+      "loss": 1.1162,
+      "step": 472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.068937550689375e-05,
+      "loss": 0.7705,
+      "step": 473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.063260340632603e-05,
+      "loss": 0.8652,
+      "step": 474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.057583130575831e-05,
+      "loss": 1.2236,
+      "step": 475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.051905920519058e-05,
+      "loss": 1.5527,
+      "step": 476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.046228710462287e-05,
+      "loss": 1.1719,
+      "step": 477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.0405515004055144e-05,
+      "loss": 0.9263,
+      "step": 478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.0348742903487424e-05,
+      "loss": 1.04,
+      "step": 479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.02919708029197e-05,
+      "loss": 0.6138,
+      "step": 480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.0235198702351985e-05,
+      "loss": 1.252,
+      "step": 481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.017842660178426e-05,
+      "loss": 0.9878,
+      "step": 482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.012165450121654e-05,
+      "loss": 1.1973,
+      "step": 483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.006488240064882e-05,
+      "loss": 0.9111,
+      "step": 484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.00081103000811e-05,
+      "loss": 1.0986,
+      "step": 485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.995133819951337e-05,
+      "loss": 0.835,
+      "step": 486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.989456609894566e-05,
+      "loss": 0.7715,
+      "step": 487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9837793998377934e-05,
+      "loss": 0.5693,
+      "step": 488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.978102189781021e-05,
+      "loss": 0.6743,
+      "step": 489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9724249797242494e-05,
+      "loss": 0.7642,
+      "step": 490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.966747769667477e-05,
+      "loss": 0.8359,
+      "step": 491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.961070559610705e-05,
+      "loss": 0.8711,
+      "step": 492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.955393349553933e-05,
+      "loss": 0.6592,
+      "step": 493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.949716139497161e-05,
+      "loss": 1.0557,
+      "step": 494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.944038929440388e-05,
+      "loss": 1.5254,
+      "step": 495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.938361719383617e-05,
+      "loss": 0.4392,
+      "step": 496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9326845093268443e-05,
+      "loss": 0.916,
+      "step": 497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9270072992700724e-05,
+      "loss": 0.9312,
+      "step": 498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9213300892133004e-05,
+      "loss": 0.8643,
+      "step": 499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9156528791565285e-05,
+      "loss": 0.8589,
+      "step": 500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.909975669099756e-05,
+      "loss": 1.0107,
+      "step": 501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9042984590429845e-05,
+      "loss": 0.5156,
+      "step": 502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.898621248986212e-05,
+      "loss": 1.5537,
+      "step": 503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.89294403892944e-05,
+      "loss": 1.2002,
+      "step": 504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.887266828872668e-05,
+      "loss": 1.2891,
+      "step": 505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.881589618815896e-05,
+      "loss": 1.1338,
+      "step": 506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8759124087591233e-05,
+      "loss": 0.9946,
+      "step": 507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.870235198702352e-05,
+      "loss": 0.8799,
+      "step": 508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8645579886455794e-05,
+      "loss": 0.9248,
+      "step": 509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.858880778588807e-05,
+      "loss": 0.8345,
+      "step": 510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8532035685320355e-05,
+      "loss": 0.7388,
+      "step": 511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.847526358475263e-05,
+      "loss": 0.7773,
+      "step": 512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.841849148418491e-05,
+      "loss": 1.0996,
+      "step": 513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8361719383617196e-05,
+      "loss": 0.4233,
+      "step": 514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.830494728304947e-05,
+      "loss": 0.6987,
+      "step": 515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.824817518248174e-05,
+      "loss": 0.7031,
+      "step": 516
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8191403081914023e-05,
+      "loss": 1.0371,
+      "step": 517
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8134630981346304e-05,
+      "loss": 0.4688,
+      "step": 518
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8077858880778584e-05,
+      "loss": 0.9785,
+      "step": 519
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.802108678021086e-05,
+      "loss": 1.0127,
+      "step": 520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.7964314679643145e-05,
+      "loss": 0.4585,
+      "step": 521
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.790754257907542e-05,
+      "loss": 0.7402,
+      "step": 522
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.78507704785077e-05,
+      "loss": 0.8706,
+      "step": 523
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.779399837793998e-05,
+      "loss": 1.0312,
+      "step": 524
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.773722627737226e-05,
+      "loss": 0.7119,
+      "step": 525
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.768045417680453e-05,
+      "loss": 0.6558,
+      "step": 526
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.762368207623682e-05,
+      "loss": 0.4546,
+      "step": 527
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.7566909975669094e-05,
+      "loss": 0.7456,
+      "step": 528
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.7510137875101374e-05,
+      "loss": 1.2949,
+      "step": 529
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.7453365774533655e-05,
+      "loss": 0.7246,
+      "step": 530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.7396593673965935e-05,
+      "loss": 0.7959,
+      "step": 531
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.733982157339821e-05,
+      "loss": 1.5547,
+      "step": 532
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.7283049472830496e-05,
+      "loss": 1.3662,
+      "step": 533
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.722627737226277e-05,
+      "loss": 0.7383,
+      "step": 534
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.716950527169504e-05,
+      "loss": 0.5659,
+      "step": 535
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.711273317112733e-05,
+      "loss": 0.813,
+      "step": 536
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.7055961070559603e-05,
+      "loss": 0.5952,
+      "step": 537
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.6999188969991884e-05,
+      "loss": 0.7021,
+      "step": 538
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.6942416869424164e-05,
+      "loss": 1.4424,
+      "step": 539
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.6885644768856445e-05,
+      "loss": 0.6577,
+      "step": 540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.682887266828872e-05,
+      "loss": 0.6777,
+      "step": 541
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.6772100567721005e-05,
+      "loss": 0.3254,
+      "step": 542
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.671532846715328e-05,
+      "loss": 0.7036,
+      "step": 543
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.665855636658556e-05,
+      "loss": 0.6216,
+      "step": 544
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.660178426601784e-05,
+      "loss": 0.3704,
+      "step": 545
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.654501216545012e-05,
+      "loss": 1.3623,
+      "step": 546
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.6488240064882394e-05,
+      "loss": 0.5874,
+      "step": 547
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.643146796431468e-05,
+      "loss": 0.8247,
+      "step": 548
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6374695863746954e-05,
+      "loss": 0.7695,
+      "step": 549
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6317923763179235e-05,
+      "loss": 0.7334,
+      "step": 550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6261151662611515e-05,
+      "loss": 0.3455,
+      "step": 551
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.6204379562043795e-05,
+      "loss": 0.7368,
+      "step": 552
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.614760746147607e-05,
+      "loss": 0.6914,
+      "step": 553
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.609083536090834e-05,
+      "loss": 0.6787,
+      "step": 554
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.603406326034063e-05,
+      "loss": 0.7505,
+      "step": 555
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.59772911597729e-05,
+      "loss": 0.583,
+      "step": 556
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.5920519059205184e-05,
+      "loss": 0.666,
+      "step": 557
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.5863746958637464e-05,
+      "loss": 0.5659,
+      "step": 558
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.5806974858069744e-05,
+      "loss": 1.3018,
+      "step": 559
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.575020275750202e-05,
+      "loss": 1.0088,
+      "step": 560
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.5693430656934305e-05,
+      "loss": 0.2649,
+      "step": 561
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.563665855636658e-05,
+      "loss": 0.7944,
+      "step": 562
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.557988645579886e-05,
+      "loss": 0.5659,
+      "step": 563
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.552311435523114e-05,
+      "loss": 0.6167,
+      "step": 564
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.546634225466342e-05,
+      "loss": 0.5488,
+      "step": 565
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.540957015409569e-05,
+      "loss": 0.2141,
+      "step": 566
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.535279805352798e-05,
+      "loss": 1.0156,
+      "step": 567
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5296025952960254e-05,
+      "loss": 0.314,
+      "step": 568
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5239253852392534e-05,
+      "loss": 1.252,
+      "step": 569
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5182481751824815e-05,
+      "loss": 1.1738,
+      "step": 570
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5125709651257095e-05,
+      "loss": 0.303,
+      "step": 571
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.506893755068937e-05,
+      "loss": 0.8926,
+      "step": 572
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5012165450121656e-05,
+      "loss": 0.8857,
+      "step": 573
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.495539334955393e-05,
+      "loss": 0.5151,
+      "step": 574
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.48986212489862e-05,
+      "loss": 0.647,
+      "step": 575
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.484184914841849e-05,
+      "loss": 0.4602,
+      "step": 576
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.4785077047850764e-05,
+      "loss": 0.6899,
+      "step": 577
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.4728304947283044e-05,
+      "loss": 0.7246,
+      "step": 578
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.4671532846715324e-05,
+      "loss": 0.5537,
+      "step": 579
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.4614760746147605e-05,
+      "loss": 0.4607,
+      "step": 580
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.455798864557988e-05,
+      "loss": 0.8325,
+      "step": 581
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.4501216545012165e-05,
+      "loss": 0.6743,
+      "step": 582
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.444444444444444e-05,
+      "loss": 0.9932,
+      "step": 583
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.438767234387672e-05,
+      "loss": 0.6094,
+      "step": 584
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.4330900243309e-05,
+      "loss": 0.8311,
+      "step": 585
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.427412814274128e-05,
+      "loss": 0.2861,
+      "step": 586
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.4217356042173554e-05,
+      "loss": 0.4551,
+      "step": 587
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.416058394160584e-05,
+      "loss": 0.7676,
+      "step": 588
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.4103811841038114e-05,
+      "loss": 0.5674,
+      "step": 589
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.4047039740470395e-05,
+      "loss": 0.9771,
+      "step": 590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.399026763990267e-05,
+      "loss": 1.2178,
+      "step": 591
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.3933495539334955e-05,
+      "loss": 0.4553,
+      "step": 592
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.387672343876723e-05,
+      "loss": 0.4836,
+      "step": 593
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.38199513381995e-05,
+      "loss": 1.502,
+      "step": 594
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.376317923763179e-05,
+      "loss": 0.5693,
+      "step": 595
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.370640713706406e-05,
+      "loss": 0.9121,
+      "step": 596
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.3649635036496344e-05,
+      "loss": 1.2461,
+      "step": 597
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.3592862935928624e-05,
+      "loss": 0.3044,
+      "step": 598
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.3536090835360904e-05,
+      "loss": 0.7896,
+      "step": 599
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.347931873479318e-05,
+      "loss": 0.9272,
+      "step": 600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.3422546634225465e-05,
+      "loss": 0.8188,
+      "step": 601
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.336577453365774e-05,
+      "loss": 0.8926,
+      "step": 602
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.330900243309002e-05,
+      "loss": 1.1768,
+      "step": 603
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.32522303325223e-05,
+      "loss": 0.7588,
+      "step": 604
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.319545823195458e-05,
+      "loss": 0.9399,
+      "step": 605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.313868613138685e-05,
+      "loss": 0.6055,
+      "step": 606
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.308191403081914e-05,
+      "loss": 0.8447,
+      "step": 607
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.3025141930251414e-05,
+      "loss": 0.5488,
+      "step": 608
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.2968369829683694e-05,
+      "loss": 0.7842,
+      "step": 609
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.2911597729115975e-05,
+      "loss": 0.7739,
+      "step": 610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.2854825628548255e-05,
+      "loss": 1.0391,
+      "step": 611
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.279805352798053e-05,
+      "loss": 0.812,
+      "step": 612
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.2741281427412816e-05,
+      "loss": 0.9531,
+      "step": 613
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.268450932684509e-05,
+      "loss": 1.0264,
+      "step": 614
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.262773722627736e-05,
+      "loss": 0.7812,
+      "step": 615
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.257096512570965e-05,
+      "loss": 1.5,
+      "step": 616
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.2514193025141924e-05,
+      "loss": 0.4814,
+      "step": 617
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.2457420924574204e-05,
+      "loss": 0.4253,
+      "step": 618
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.2400648824006484e-05,
+      "loss": 0.7495,
+      "step": 619
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.2343876723438765e-05,
+      "loss": 0.4854,
+      "step": 620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.228710462287104e-05,
+      "loss": 0.6201,
+      "step": 621
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.2230332522303325e-05,
+      "loss": 0.4961,
+      "step": 622
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.21735604217356e-05,
+      "loss": 0.4758,
+      "step": 623
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.211678832116788e-05,
+      "loss": 0.5234,
+      "step": 624
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.206001622060016e-05,
+      "loss": 0.8047,
+      "step": 625
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.200324412003244e-05,
+      "loss": 0.3557,
+      "step": 626
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.1946472019464714e-05,
+      "loss": 0.6372,
+      "step": 627
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.1889699918896994e-05,
+      "loss": 0.7012,
+      "step": 628
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.1832927818329274e-05,
+      "loss": 0.6177,
+      "step": 629
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.1776155717761555e-05,
+      "loss": 0.8203,
+      "step": 630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.171938361719383e-05,
+      "loss": 0.2954,
+      "step": 631
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.1662611516626115e-05,
+      "loss": 0.4924,
+      "step": 632
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.160583941605839e-05,
+      "loss": 0.3064,
+      "step": 633
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.154906731549066e-05,
+      "loss": 0.4688,
+      "step": 634
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.149229521492295e-05,
+      "loss": 1.1973,
+      "step": 635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.143552311435522e-05,
+      "loss": 0.3574,
+      "step": 636
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.1378751013787504e-05,
+      "loss": 0.8818,
+      "step": 637
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.1321978913219784e-05,
+      "loss": 0.5815,
+      "step": 638
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.1265206812652064e-05,
+      "loss": 0.9336,
+      "step": 639
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.120843471208434e-05,
+      "loss": 0.7153,
+      "step": 640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.1151662611516625e-05,
+      "loss": 0.8955,
+      "step": 641
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.10948905109489e-05,
+      "loss": 0.7441,
+      "step": 642
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.103811841038118e-05,
+      "loss": 0.6123,
+      "step": 643
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.098134630981346e-05,
+      "loss": 0.4082,
+      "step": 644
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.092457420924574e-05,
+      "loss": 0.7158,
+      "step": 645
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.086780210867801e-05,
+      "loss": 0.7539,
+      "step": 646
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.08110300081103e-05,
+      "loss": 0.8965,
+      "step": 647
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0754257907542574e-05,
+      "loss": 0.522,
+      "step": 648
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0697485806974854e-05,
+      "loss": 0.4617,
+      "step": 649
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0640713706407135e-05,
+      "loss": 0.4121,
+      "step": 650
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.0583941605839415e-05,
+      "loss": 0.4231,
+      "step": 651
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.052716950527169e-05,
+      "loss": 0.7451,
+      "step": 652
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.0470397404703976e-05,
+      "loss": 1.0732,
+      "step": 653
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.041362530413625e-05,
+      "loss": 0.9907,
+      "step": 654
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.035685320356852e-05,
+      "loss": 0.4873,
+      "step": 655
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.030008110300081e-05,
+      "loss": 0.5576,
+      "step": 656
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.0243309002433084e-05,
+      "loss": 0.6743,
+      "step": 657
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.0186536901865364e-05,
+      "loss": 2.2871,
+      "step": 658
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.0129764801297644e-05,
+      "loss": 0.3877,
+      "step": 659
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.0072992700729925e-05,
+      "loss": 1.2285,
+      "step": 660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.00162206001622e-05,
+      "loss": 0.54,
+      "step": 661
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9959448499594485e-05,
+      "loss": 0.7734,
+      "step": 662
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.990267639902676e-05,
+      "loss": 0.3262,
+      "step": 663
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.984590429845904e-05,
+      "loss": 0.8691,
+      "step": 664
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.978913219789131e-05,
+      "loss": 0.9824,
+      "step": 665
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.97323600973236e-05,
+      "loss": 0.5977,
+      "step": 666
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9675587996755874e-05,
+      "loss": 0.3306,
+      "step": 667
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9618815896188154e-05,
+      "loss": 0.6807,
+      "step": 668
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9562043795620434e-05,
+      "loss": 0.522,
+      "step": 669
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9505271695052715e-05,
+      "loss": 0.469,
+      "step": 670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.944849959448499e-05,
+      "loss": 0.9087,
+      "step": 671
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.9391727493917275e-05,
+      "loss": 0.8984,
+      "step": 672
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.933495539334955e-05,
+      "loss": 0.8662,
+      "step": 673
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.927818329278182e-05,
+      "loss": 1.2021,
+      "step": 674
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.922141119221411e-05,
+      "loss": 1.0156,
+      "step": 675
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.916463909164638e-05,
+      "loss": 1.0498,
+      "step": 676
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.9107866991078664e-05,
+      "loss": 0.9355,
+      "step": 677
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.9051094890510944e-05,
+      "loss": 1.4971,
+      "step": 678
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.8994322789943224e-05,
+      "loss": 0.542,
+      "step": 679
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.89375506893755e-05,
+      "loss": 0.7451,
+      "step": 680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.8880778588807785e-05,
+      "loss": 0.7549,
+      "step": 681
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.882400648824006e-05,
+      "loss": 1.1846,
+      "step": 682
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.876723438767234e-05,
+      "loss": 1.3223,
+      "step": 683
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.871046228710462e-05,
+      "loss": 0.9224,
+      "step": 684
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.86536901865369e-05,
+      "loss": 0.9736,
+      "step": 685
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.859691808596917e-05,
+      "loss": 0.9375,
+      "step": 686
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.854014598540146e-05,
+      "loss": 0.7939,
+      "step": 687
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.8483373884833734e-05,
+      "loss": 0.425,
+      "step": 688
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.8426601784266014e-05,
+      "loss": 0.9941,
+      "step": 689
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.8369829683698295e-05,
+      "loss": 0.3867,
+      "step": 690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.8313057583130575e-05,
+      "loss": 0.8789,
+      "step": 691
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.825628548256285e-05,
+      "loss": 0.856,
+      "step": 692
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.8199513381995136e-05,
+      "loss": 0.6328,
+      "step": 693
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.814274128142741e-05,
+      "loss": 0.3904,
+      "step": 694
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.808596918085968e-05,
+      "loss": 0.9131,
+      "step": 695
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.802919708029197e-05,
+      "loss": 1.7617,
+      "step": 696
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7972424979724244e-05,
+      "loss": 0.7188,
+      "step": 697
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7915652879156524e-05,
+      "loss": 0.7603,
+      "step": 698
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7858880778588804e-05,
+      "loss": 0.4224,
+      "step": 699
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.7802108678021085e-05,
+      "loss": 0.4788,
+      "step": 700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.774533657745336e-05,
+      "loss": 0.3206,
+      "step": 701
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.768856447688564e-05,
+      "loss": 0.8936,
+      "step": 702
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.763179237631792e-05,
+      "loss": 1.0488,
+      "step": 703
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.75750202757502e-05,
+      "loss": 0.4692,
+      "step": 704
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.751824817518247e-05,
+      "loss": 1.1621,
+      "step": 705
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.746147607461476e-05,
+      "loss": 0.6011,
+      "step": 706
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7404703974047034e-05,
+      "loss": 0.5776,
+      "step": 707
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7347931873479314e-05,
+      "loss": 0.7192,
+      "step": 708
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7291159772911594e-05,
+      "loss": 0.6357,
+      "step": 709
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7234387672343875e-05,
+      "loss": 0.9116,
+      "step": 710
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.717761557177615e-05,
+      "loss": 0.8164,
+      "step": 711
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.7120843471208436e-05,
+      "loss": 0.5967,
+      "step": 712
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.706407137064071e-05,
+      "loss": 0.978,
+      "step": 713
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.700729927007298e-05,
+      "loss": 0.6914,
+      "step": 714
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.695052716950527e-05,
+      "loss": 0.5815,
+      "step": 715
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6893755068937543e-05,
+      "loss": 0.9937,
+      "step": 716
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.6836982968369824e-05,
+      "loss": 0.6064,
+      "step": 717
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6780210867802104e-05,
+      "loss": 0.7002,
+      "step": 718
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6723438767234384e-05,
+      "loss": 0.4333,
+      "step": 719
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.666666666666666e-05,
+      "loss": 1.2129,
+      "step": 720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.6609894566098945e-05,
+      "loss": 0.6694,
+      "step": 721
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.655312246553122e-05,
+      "loss": 0.6294,
+      "step": 722
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.64963503649635e-05,
+      "loss": 0.6826,
+      "step": 723
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.643957826439578e-05,
+      "loss": 1.0996,
+      "step": 724
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.638280616382806e-05,
+      "loss": 0.5732,
+      "step": 725
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.6326034063260333e-05,
+      "loss": 0.6401,
+      "step": 726
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.626926196269262e-05,
+      "loss": 0.8794,
+      "step": 727
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6212489862124894e-05,
+      "loss": 0.4543,
+      "step": 728
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6155717761557174e-05,
+      "loss": 0.8379,
+      "step": 729
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6098945660989455e-05,
+      "loss": 0.5986,
+      "step": 730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.6042173560421735e-05,
+      "loss": 0.6763,
+      "step": 731
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.598540145985401e-05,
+      "loss": 0.5933,
+      "step": 732
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5928629359286296e-05,
+      "loss": 1.1729,
+      "step": 733
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.587185725871857e-05,
+      "loss": 0.4932,
+      "step": 734
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.581508515815084e-05,
+      "loss": 0.4761,
+      "step": 735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.575831305758313e-05,
+      "loss": 0.6289,
+      "step": 736
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5701540957015404e-05,
+      "loss": 1.4434,
+      "step": 737
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5644768856447684e-05,
+      "loss": 0.6553,
+      "step": 738
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.558799675587996e-05,
+      "loss": 0.7256,
+      "step": 739
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.5531224655312245e-05,
+      "loss": 0.8916,
+      "step": 740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.547445255474452e-05,
+      "loss": 0.4094,
+      "step": 741
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.54176804541768e-05,
+      "loss": 0.6704,
+      "step": 742
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.536090835360908e-05,
+      "loss": 0.9302,
+      "step": 743
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.530413625304136e-05,
+      "loss": 0.6665,
+      "step": 744
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.524736415247363e-05,
+      "loss": 0.4407,
+      "step": 745
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.519059205190592e-05,
+      "loss": 0.624,
+      "step": 746
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5133819951338194e-05,
+      "loss": 0.2291,
+      "step": 747
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5077047850770474e-05,
+      "loss": 0.9922,
+      "step": 748
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5020275750202755e-05,
+      "loss": 0.9346,
+      "step": 749
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.4963503649635035e-05,
+      "loss": 0.9111,
+      "step": 750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.490673154906731e-05,
+      "loss": 0.7959,
+      "step": 751
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.4849959448499596e-05,
+      "loss": 1.416,
+      "step": 752
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.479318734793187e-05,
+      "loss": 0.7896,
+      "step": 753
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.473641524736414e-05,
+      "loss": 0.3201,
+      "step": 754
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.467964314679643e-05,
+      "loss": 1.1387,
+      "step": 755
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.4622871046228703e-05,
+      "loss": 1.4805,
+      "step": 756
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.4566098945660984e-05,
+      "loss": 0.5156,
+      "step": 757
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.4509326845093264e-05,
+      "loss": 0.6445,
+      "step": 758
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.4452554744525545e-05,
+      "loss": 0.4756,
+      "step": 759
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.439578264395782e-05,
+      "loss": 0.4639,
+      "step": 760
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.4339010543390105e-05,
+      "loss": 0.6973,
+      "step": 761
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.428223844282238e-05,
+      "loss": 0.6152,
+      "step": 762
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.422546634225466e-05,
+      "loss": 1.7656,
+      "step": 763
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.416869424168694e-05,
+      "loss": 0.9326,
+      "step": 764
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.411192214111922e-05,
+      "loss": 0.3223,
+      "step": 765
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.4055150040551493e-05,
+      "loss": 1.0098,
+      "step": 766
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.399837793998378e-05,
+      "loss": 0.4707,
+      "step": 767
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.3941605839416054e-05,
+      "loss": 0.5889,
+      "step": 768
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.3884833738848335e-05,
+      "loss": 1.167,
+      "step": 769
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.3828061638280615e-05,
+      "loss": 0.9785,
+      "step": 770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.3771289537712895e-05,
+      "loss": 0.5278,
+      "step": 771
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.371451743714517e-05,
+      "loss": 0.6733,
+      "step": 772
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.3657745336577456e-05,
+      "loss": 0.77,
+      "step": 773
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.360097323600973e-05,
+      "loss": 0.6948,
+      "step": 774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.3544201135442e-05,
+      "loss": 0.5112,
+      "step": 775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.348742903487429e-05,
+      "loss": 0.6782,
+      "step": 776
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.3430656934306564e-05,
+      "loss": 0.8408,
+      "step": 777
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.3373884833738844e-05,
+      "loss": 0.8418,
+      "step": 778
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.331711273317112e-05,
+      "loss": 1.1699,
+      "step": 779
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3260340632603405e-05,
+      "loss": 0.2642,
+      "step": 780
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.320356853203568e-05,
+      "loss": 0.9331,
+      "step": 781
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.314679643146796e-05,
+      "loss": 0.4326,
+      "step": 782
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.309002433090024e-05,
+      "loss": 0.5601,
+      "step": 783
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.303325223033252e-05,
+      "loss": 1.208,
+      "step": 784
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.297648012976479e-05,
+      "loss": 0.7935,
+      "step": 785
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.291970802919708e-05,
+      "loss": 0.8384,
+      "step": 786
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2862935928629354e-05,
+      "loss": 1.0117,
+      "step": 787
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2806163828061634e-05,
+      "loss": 0.9062,
+      "step": 788
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2749391727493915e-05,
+      "loss": 0.9189,
+      "step": 789
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2692619626926195e-05,
+      "loss": 0.9731,
+      "step": 790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.263584752635847e-05,
+      "loss": 0.4832,
+      "step": 791
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2579075425790756e-05,
+      "loss": 0.6538,
+      "step": 792
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.252230332522303e-05,
+      "loss": 0.791,
+      "step": 793
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.24655312246553e-05,
+      "loss": 0.9517,
+      "step": 794
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.240875912408759e-05,
+      "loss": 1.2822,
+      "step": 795
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2351987023519864e-05,
+      "loss": 1.3242,
+      "step": 796
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2295214922952144e-05,
+      "loss": 0.5776,
+      "step": 797
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2238442822384424e-05,
+      "loss": 0.8545,
+      "step": 798
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2181670721816705e-05,
+      "loss": 0.7012,
+      "step": 799
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.212489862124898e-05,
+      "loss": 1.1816,
+      "step": 800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2068126520681265e-05,
+      "loss": 0.6606,
+      "step": 801
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.201135442011354e-05,
+      "loss": 0.1469,
+      "step": 802
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.195458231954582e-05,
+      "loss": 0.6958,
+      "step": 803
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.18978102189781e-05,
+      "loss": 0.9199,
+      "step": 804
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.184103811841038e-05,
+      "loss": 0.5605,
+      "step": 805
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.1784266017842654e-05,
+      "loss": 0.8633,
+      "step": 806
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.172749391727494e-05,
+      "loss": 0.7412,
+      "step": 807
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.1670721816707214e-05,
+      "loss": 1.3379,
+      "step": 808
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.1613949716139495e-05,
+      "loss": 0.3428,
+      "step": 809
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.1557177615571775e-05,
+      "loss": 0.8501,
+      "step": 810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.1500405515004055e-05,
+      "loss": 1.1455,
+      "step": 811
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.144363341443633e-05,
+      "loss": 0.5195,
+      "step": 812
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.1386861313868616e-05,
+      "loss": 0.5435,
+      "step": 813
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.133008921330089e-05,
+      "loss": 0.9927,
+      "step": 814
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.127331711273316e-05,
+      "loss": 0.8477,
+      "step": 815
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1216545012165444e-05,
+      "loss": 1.041,
+      "step": 816
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1159772911597724e-05,
+      "loss": 1.0352,
+      "step": 817
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1103000811030004e-05,
+      "loss": 0.8618,
+      "step": 818
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.104622871046228e-05,
+      "loss": 0.5752,
+      "step": 819
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0989456609894565e-05,
+      "loss": 0.2349,
+      "step": 820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.093268450932684e-05,
+      "loss": 0.7603,
+      "step": 821
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.087591240875912e-05,
+      "loss": 1.0488,
+      "step": 822
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.08191403081914e-05,
+      "loss": 0.3303,
+      "step": 823
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.076236820762368e-05,
+      "loss": 0.467,
+      "step": 824
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.070559610705595e-05,
+      "loss": 0.5454,
+      "step": 825
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.064882400648824e-05,
+      "loss": 1.1592,
+      "step": 826
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.0592051905920514e-05,
+      "loss": 0.8486,
+      "step": 827
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.0535279805352794e-05,
+      "loss": 1.3613,
+      "step": 828
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.0478507704785075e-05,
+      "loss": 0.7183,
+      "step": 829
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.0421735604217355e-05,
+      "loss": 0.7847,
+      "step": 830
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.036496350364963e-05,
+      "loss": 0.6357,
+      "step": 831
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.0308191403081916e-05,
+      "loss": 0.5742,
+      "step": 832
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.025141930251419e-05,
+      "loss": 0.7642,
+      "step": 833
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.019464720194646e-05,
+      "loss": 0.6035,
+      "step": 834
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.013787510137875e-05,
+      "loss": 0.9321,
+      "step": 835
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.0081103000811024e-05,
+      "loss": 1.1025,
+      "step": 836
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.0024330900243304e-05,
+      "loss": 0.7021,
+      "step": 837
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9967558799675584e-05,
+      "loss": 0.4368,
+      "step": 838
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9910786699107865e-05,
+      "loss": 0.7178,
+      "step": 839
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.985401459854014e-05,
+      "loss": 0.7056,
+      "step": 840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9797242497972425e-05,
+      "loss": 0.7983,
+      "step": 841
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.97404703974047e-05,
+      "loss": 0.2686,
+      "step": 842
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.968369829683698e-05,
+      "loss": 1.1807,
+      "step": 843
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.962692619626926e-05,
+      "loss": 0.3396,
+      "step": 844
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.957015409570154e-05,
+      "loss": 0.8848,
+      "step": 845
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9513381995133814e-05,
+      "loss": 0.5806,
+      "step": 846
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.94566098945661e-05,
+      "loss": 0.9692,
+      "step": 847
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9399837793998374e-05,
+      "loss": 1.0947,
+      "step": 848
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9343065693430655e-05,
+      "loss": 1.0781,
+      "step": 849
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9286293592862935e-05,
+      "loss": 0.5312,
+      "step": 850
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9229521492295215e-05,
+      "loss": 0.855,
+      "step": 851
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.917274939172749e-05,
+      "loss": 0.428,
+      "step": 852
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.911597729115976e-05,
+      "loss": 0.8184,
+      "step": 853
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.905920519059205e-05,
+      "loss": 0.8101,
+      "step": 854
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.900243309002432e-05,
+      "loss": 0.8364,
+      "step": 855
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8945660989456604e-05,
+      "loss": 0.8145,
+      "step": 856
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8888888888888884e-05,
+      "loss": 0.3484,
+      "step": 857
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8832116788321164e-05,
+      "loss": 0.8086,
+      "step": 858
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.877534468775344e-05,
+      "loss": 0.7988,
+      "step": 859
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8718572587185725e-05,
+      "loss": 1.1143,
+      "step": 860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8661800486618e-05,
+      "loss": 0.9492,
+      "step": 861
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.860502838605028e-05,
+      "loss": 1.209,
+      "step": 862
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.854825628548256e-05,
+      "loss": 0.8208,
+      "step": 863
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.849148418491484e-05,
+      "loss": 0.4158,
+      "step": 864
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.843471208434711e-05,
+      "loss": 0.4329,
+      "step": 865
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.83779399837794e-05,
+      "loss": 0.585,
+      "step": 866
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8321167883211674e-05,
+      "loss": 0.5088,
+      "step": 867
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8264395782643954e-05,
+      "loss": 0.5562,
+      "step": 868
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8207623682076235e-05,
+      "loss": 1.0244,
+      "step": 869
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8150851581508515e-05,
+      "loss": 0.6074,
+      "step": 870
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.809407948094079e-05,
+      "loss": 0.6514,
+      "step": 871
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8037307380373076e-05,
+      "loss": 1.0977,
+      "step": 872
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.798053527980535e-05,
+      "loss": 1.0625,
+      "step": 873
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.792376317923762e-05,
+      "loss": 0.8379,
+      "step": 874
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.786699107866991e-05,
+      "loss": 0.8101,
+      "step": 875
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7810218978102184e-05,
+      "loss": 0.75,
+      "step": 876
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7753446877534464e-05,
+      "loss": 0.8555,
+      "step": 877
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7696674776966744e-05,
+      "loss": 0.75,
+      "step": 878
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7639902676399025e-05,
+      "loss": 0.8521,
+      "step": 879
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.75831305758313e-05,
+      "loss": 1.1045,
+      "step": 880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7526358475263585e-05,
+      "loss": 1.3887,
+      "step": 881
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.746958637469586e-05,
+      "loss": 0.2522,
+      "step": 882
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.741281427412814e-05,
+      "loss": 1.3838,
+      "step": 883
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.735604217356042e-05,
+      "loss": 0.9751,
+      "step": 884
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.72992700729927e-05,
+      "loss": 0.4792,
+      "step": 885
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7242497972424974e-05,
+      "loss": 0.4321,
+      "step": 886
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.718572587185726e-05,
+      "loss": 0.9722,
+      "step": 887
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7128953771289534e-05,
+      "loss": 0.562,
+      "step": 888
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7072181670721815e-05,
+      "loss": 0.4463,
+      "step": 889
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.701540957015409e-05,
+      "loss": 1.5498,
+      "step": 890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.6958637469586375e-05,
+      "loss": 0.8501,
+      "step": 891
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.690186536901865e-05,
+      "loss": 0.8066,
+      "step": 892
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.684509326845092e-05,
+      "loss": 0.9126,
+      "step": 893
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.678832116788321e-05,
+      "loss": 0.8511,
+      "step": 894
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.673154906731548e-05,
+      "loss": 0.4231,
+      "step": 895
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.6674776966747764e-05,
+      "loss": 0.4185,
+      "step": 896
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.6618004866180044e-05,
+      "loss": 0.8369,
+      "step": 897
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.6561232765612324e-05,
+      "loss": 0.5913,
+      "step": 898
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.65044606650446e-05,
+      "loss": 0.9609,
+      "step": 899
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.6447688564476885e-05,
+      "loss": 0.4048,
+      "step": 900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.639091646390916e-05,
+      "loss": 0.5698,
+      "step": 901
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.633414436334144e-05,
+      "loss": 1.0156,
+      "step": 902
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.627737226277372e-05,
+      "loss": 0.6201,
+      "step": 903
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6220600162206e-05,
+      "loss": 1.6387,
+      "step": 904
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.616382806163827e-05,
+      "loss": 0.9468,
+      "step": 905
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.610705596107056e-05,
+      "loss": 0.8638,
+      "step": 906
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6050283860502834e-05,
+      "loss": 0.4985,
+      "step": 907
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5993511759935114e-05,
+      "loss": 1.2568,
+      "step": 908
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5936739659367395e-05,
+      "loss": 0.8823,
+      "step": 909
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5879967558799675e-05,
+      "loss": 0.6323,
+      "step": 910
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.582319545823195e-05,
+      "loss": 0.3994,
+      "step": 911
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5766423357664236e-05,
+      "loss": 0.6123,
+      "step": 912
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.570965125709651e-05,
+      "loss": 0.3184,
+      "step": 913
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.565287915652878e-05,
+      "loss": 0.5298,
+      "step": 914
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.559610705596107e-05,
+      "loss": 0.6157,
+      "step": 915
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.5539334955393344e-05,
+      "loss": 0.5684,
+      "step": 916
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.5482562854825624e-05,
+      "loss": 0.7129,
+      "step": 917
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.5425790754257904e-05,
+      "loss": 0.8398,
+      "step": 918
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.5369018653690185e-05,
+      "loss": 0.4697,
+      "step": 919
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.531224655312246e-05,
+      "loss": 1.2695,
+      "step": 920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.5255474452554745e-05,
+      "loss": 0.4897,
+      "step": 921
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.519870235198702e-05,
+      "loss": 0.6719,
+      "step": 922
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.51419302514193e-05,
+      "loss": 0.5254,
+      "step": 923
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.508515815085158e-05,
+      "loss": 0.4199,
+      "step": 924
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.502838605028386e-05,
+      "loss": 0.5068,
+      "step": 925
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.4971613949716134e-05,
+      "loss": 0.4521,
+      "step": 926
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.4914841849148414e-05,
+      "loss": 0.6313,
+      "step": 927
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4858069748580694e-05,
+      "loss": 1.1426,
+      "step": 928
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4801297648012975e-05,
+      "loss": 1.0918,
+      "step": 929
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4744525547445255e-05,
+      "loss": 0.623,
+      "step": 930
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4687753446877536e-05,
+      "loss": 0.7119,
+      "step": 931
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.463098134630981e-05,
+      "loss": 0.7236,
+      "step": 932
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.457420924574209e-05,
+      "loss": 0.7139,
+      "step": 933
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.451743714517437e-05,
+      "loss": 1.2646,
+      "step": 934
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.446066504460665e-05,
+      "loss": 0.7563,
+      "step": 935
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.440389294403893e-05,
+      "loss": 0.5947,
+      "step": 936
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4347120843471204e-05,
+      "loss": 1.3467,
+      "step": 937
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4290348742903484e-05,
+      "loss": 0.6465,
+      "step": 938
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.4233576642335765e-05,
+      "loss": 0.5835,
+      "step": 939
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.417680454176804e-05,
+      "loss": 1.332,
+      "step": 940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.412003244120032e-05,
+      "loss": 0.9106,
+      "step": 941
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.40632603406326e-05,
+      "loss": 1.0898,
+      "step": 942
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.400648824006488e-05,
+      "loss": 0.7256,
+      "step": 943
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.394971613949716e-05,
+      "loss": 0.6187,
+      "step": 944
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.389294403892944e-05,
+      "loss": 0.7061,
+      "step": 945
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3836171938361714e-05,
+      "loss": 0.9189,
+      "step": 946
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3779399837793994e-05,
+      "loss": 0.8999,
+      "step": 947
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3722627737226274e-05,
+      "loss": 0.7769,
+      "step": 948
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.3665855636658555e-05,
+      "loss": 0.627,
+      "step": 949
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.3609083536090835e-05,
+      "loss": 0.7388,
+      "step": 950
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.3552311435523116e-05,
+      "loss": 0.3435,
+      "step": 951
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.349553933495539e-05,
+      "loss": 0.9785,
+      "step": 952
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.343876723438767e-05,
+      "loss": 0.6763,
+      "step": 953
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.338199513381995e-05,
+      "loss": 0.5811,
+      "step": 954
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.332522303325223e-05,
+      "loss": 1.1953,
+      "step": 955
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.326845093268451e-05,
+      "loss": 0.7393,
+      "step": 956
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.3211678832116784e-05,
+      "loss": 0.7554,
+      "step": 957
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.3154906731549064e-05,
+      "loss": 0.5854,
+      "step": 958
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3098134630981345e-05,
+      "loss": 0.1134,
+      "step": 959
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.304136253041362e-05,
+      "loss": 1.1055,
+      "step": 960
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.29845904298459e-05,
+      "loss": 0.7324,
+      "step": 961
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.292781832927818e-05,
+      "loss": 0.6943,
+      "step": 962
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.287104622871046e-05,
+      "loss": 0.4536,
+      "step": 963
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.281427412814274e-05,
+      "loss": 0.8086,
+      "step": 964
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.275750202757502e-05,
+      "loss": 0.5688,
+      "step": 965
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.2700729927007294e-05,
+      "loss": 0.7705,
+      "step": 966
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.2643957826439574e-05,
+      "loss": 0.5151,
+      "step": 967
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.2587185725871854e-05,
+      "loss": 0.897,
+      "step": 968
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.2530413625304135e-05,
+      "loss": 1.124,
+      "step": 969
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.2473641524736415e-05,
+      "loss": 0.5747,
+      "step": 970
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.2416869424168696e-05,
+      "loss": 0.5112,
+      "step": 971
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.236009732360097e-05,
+      "loss": 0.3132,
+      "step": 972
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.230332522303325e-05,
+      "loss": 0.875,
+      "step": 973
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.224655312246553e-05,
+      "loss": 0.5933,
+      "step": 974
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.218978102189781e-05,
+      "loss": 0.7305,
+      "step": 975
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.213300892133009e-05,
+      "loss": 0.9233,
+      "step": 976
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.2076236820762364e-05,
+      "loss": 1.7334,
+      "step": 977
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.2019464720194645e-05,
+      "loss": 0.9136,
+      "step": 978
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1962692619626925e-05,
+      "loss": 0.7993,
+      "step": 979
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.19059205190592e-05,
+      "loss": 1.1094,
+      "step": 980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.184914841849148e-05,
+      "loss": 0.321,
+      "step": 981
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.179237631792376e-05,
+      "loss": 1.1719,
+      "step": 982
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.173560421735604e-05,
+      "loss": 0.6455,
+      "step": 983
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.167883211678832e-05,
+      "loss": 0.4949,
+      "step": 984
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.16220600162206e-05,
+      "loss": 0.856,
+      "step": 985
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.1565287915652874e-05,
+      "loss": 0.5679,
+      "step": 986
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.1508515815085154e-05,
+      "loss": 0.7729,
+      "step": 987
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.1451743714517435e-05,
+      "loss": 0.668,
+      "step": 988
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.1394971613949715e-05,
+      "loss": 1.25,
+      "step": 989
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.1338199513381995e-05,
+      "loss": 0.7476,
+      "step": 990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.128142741281427e-05,
+      "loss": 0.7188,
+      "step": 991
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.122465531224655e-05,
+      "loss": 0.4893,
+      "step": 992
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.116788321167883e-05,
+      "loss": 0.7192,
+      "step": 993
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 0.6304,
+      "step": 994
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.105433901054339e-05,
+      "loss": 0.4111,
+      "step": 995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.099756690997567e-05,
+      "loss": 0.9019,
+      "step": 996
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0940794809407944e-05,
+      "loss": 0.6289,
+      "step": 997
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0884022708840225e-05,
+      "loss": 0.5425,
+      "step": 998
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0827250608272505e-05,
+      "loss": 0.3123,
+      "step": 999
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.077047850770478e-05,
+      "loss": 0.3379,
+      "step": 1000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.071370640713706e-05,
+      "loss": 0.959,
+      "step": 1001
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.065693430656934e-05,
+      "loss": 0.541,
+      "step": 1002
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.060016220600162e-05,
+      "loss": 0.7529,
+      "step": 1003
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.05433901054339e-05,
+      "loss": 0.2869,
+      "step": 1004
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.048661800486618e-05,
+      "loss": 1.0674,
+      "step": 1005
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0429845904298454e-05,
+      "loss": 0.6401,
+      "step": 1006
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0373073803730734e-05,
+      "loss": 0.6665,
+      "step": 1007
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0316301703163015e-05,
+      "loss": 1.0078,
+      "step": 1008
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.025952960259529e-05,
+      "loss": 1.085,
+      "step": 1009
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0202757502027572e-05,
+      "loss": 0.626,
+      "step": 1010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.014598540145985e-05,
+      "loss": 0.9785,
+      "step": 1011
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.008921330089213e-05,
+      "loss": 0.895,
+      "step": 1012
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.003244120032441e-05,
+      "loss": 1.0342,
+      "step": 1013
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.9975669099756687e-05,
+      "loss": 0.8584,
+      "step": 1014
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.9918896999188967e-05,
+      "loss": 0.7554,
+      "step": 1015
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9862124898621247e-05,
+      "loss": 0.9302,
+      "step": 1016
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9805352798053524e-05,
+      "loss": 0.5811,
+      "step": 1017
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9748580697485805e-05,
+      "loss": 1.125,
+      "step": 1018
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9691808596918085e-05,
+      "loss": 1.1416,
+      "step": 1019
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9635036496350362e-05,
+      "loss": 1.1533,
+      "step": 1020
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.9578264395782642e-05,
+      "loss": 0.9238,
+      "step": 1021
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.9521492295214923e-05,
+      "loss": 1.2266,
+      "step": 1022
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.94647201946472e-05,
+      "loss": 0.272,
+      "step": 1023
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.940794809407948e-05,
+      "loss": 0.8979,
+      "step": 1024
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.935117599351176e-05,
+      "loss": 0.835,
+      "step": 1025
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9294403892944034e-05,
+      "loss": 0.4558,
+      "step": 1026
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9237631792376314e-05,
+      "loss": 0.708,
+      "step": 1027
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9180859691808598e-05,
+      "loss": 0.709,
+      "step": 1028
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.912408759124087e-05,
+      "loss": 0.4038,
+      "step": 1029
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9067315490673152e-05,
+      "loss": 0.6201,
+      "step": 1030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.901054339010543e-05,
+      "loss": 0.4866,
+      "step": 1031
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.895377128953771e-05,
+      "loss": 0.1422,
+      "step": 1032
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.889699918896999e-05,
+      "loss": 0.4102,
+      "step": 1033
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.8840227088402267e-05,
+      "loss": 0.3677,
+      "step": 1034
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.8783454987834547e-05,
+      "loss": 0.3108,
+      "step": 1035
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.8726682887266827e-05,
+      "loss": 0.8662,
+      "step": 1036
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.8669910786699104e-05,
+      "loss": 0.9453,
+      "step": 1037
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.8613138686131385e-05,
+      "loss": 0.8242,
+      "step": 1038
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.8556366585563665e-05,
+      "loss": 0.5547,
+      "step": 1039
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.8499594484995942e-05,
+      "loss": 0.2257,
+      "step": 1040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.8442822384428222e-05,
+      "loss": 0.6704,
+      "step": 1041
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.8386050283860503e-05,
+      "loss": 0.5303,
+      "step": 1042
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.832927818329278e-05,
+      "loss": 0.7231,
+      "step": 1043
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.827250608272506e-05,
+      "loss": 0.2352,
+      "step": 1044
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.821573398215734e-05,
+      "loss": 0.4023,
+      "step": 1045
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.8158961881589617e-05,
+      "loss": 0.8174,
+      "step": 1046
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.8102189781021898e-05,
+      "loss": 0.5205,
+      "step": 1047
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.804541768045417e-05,
+      "loss": 0.5718,
+      "step": 1048
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.798864557988645e-05,
+      "loss": 1.1748,
+      "step": 1049
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7931873479318732e-05,
+      "loss": 0.479,
+      "step": 1050
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.787510137875101e-05,
+      "loss": 0.7202,
+      "step": 1051
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.781832927818329e-05,
+      "loss": 0.647,
+      "step": 1052
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.776155717761557e-05,
+      "loss": 0.5723,
+      "step": 1053
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7704785077047847e-05,
+      "loss": 0.5718,
+      "step": 1054
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7648012976480127e-05,
+      "loss": 0.5396,
+      "step": 1055
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7591240875912407e-05,
+      "loss": 0.2593,
+      "step": 1056
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7534468775344684e-05,
+      "loss": 0.5273,
+      "step": 1057
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7477696674776965e-05,
+      "loss": 0.439,
+      "step": 1058
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7420924574209245e-05,
+      "loss": 0.3242,
+      "step": 1059
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7364152473641522e-05,
+      "loss": 0.9819,
+      "step": 1060
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7307380373073802e-05,
+      "loss": 0.4307,
+      "step": 1061
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.7250608272506083e-05,
+      "loss": 0.5859,
+      "step": 1062
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.719383617193836e-05,
+      "loss": 0.5366,
+      "step": 1063
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.713706407137064e-05,
+      "loss": 0.8901,
+      "step": 1064
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.708029197080292e-05,
+      "loss": 1.3379,
+      "step": 1065
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.7023519870235197e-05,
+      "loss": 0.687,
+      "step": 1066
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6966747769667478e-05,
+      "loss": 0.4709,
+      "step": 1067
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.690997566909975e-05,
+      "loss": 0.4983,
+      "step": 1068
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.685320356853203e-05,
+      "loss": 0.2065,
+      "step": 1069
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6796431467964312e-05,
+      "loss": 0.5854,
+      "step": 1070
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.673965936739659e-05,
+      "loss": 0.4934,
+      "step": 1071
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.668288726682887e-05,
+      "loss": 0.4431,
+      "step": 1072
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.662611516626115e-05,
+      "loss": 0.4934,
+      "step": 1073
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6569343065693427e-05,
+      "loss": 1.1514,
+      "step": 1074
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6512570965125707e-05,
+      "loss": 1.3037,
+      "step": 1075
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6455798864557987e-05,
+      "loss": 0.5269,
+      "step": 1076
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6399026763990264e-05,
+      "loss": 0.2634,
+      "step": 1077
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6342254663422545e-05,
+      "loss": 0.3381,
+      "step": 1078
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6285482562854825e-05,
+      "loss": 0.1937,
+      "step": 1079
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6228710462287102e-05,
+      "loss": 0.2969,
+      "step": 1080
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6171938361719382e-05,
+      "loss": 0.6128,
+      "step": 1081
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6115166261151663e-05,
+      "loss": 0.71,
+      "step": 1082
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.605839416058394e-05,
+      "loss": 0.4531,
+      "step": 1083
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.600162206001622e-05,
+      "loss": 0.4827,
+      "step": 1084
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.5944849959448497e-05,
+      "loss": 0.4341,
+      "step": 1085
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.5888077858880777e-05,
+      "loss": 0.5361,
+      "step": 1086
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.5831305758313058e-05,
+      "loss": 0.3362,
+      "step": 1087
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.577453365774533e-05,
+      "loss": 0.3809,
+      "step": 1088
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.571776155717761e-05,
+      "loss": 0.147,
+      "step": 1089
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.5660989456609892e-05,
+      "loss": 1.0244,
+      "step": 1090
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.560421735604217e-05,
+      "loss": 1.2217,
+      "step": 1091
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.554744525547445e-05,
+      "loss": 0.3718,
+      "step": 1092
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.549067315490673e-05,
+      "loss": 0.5864,
+      "step": 1093
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5433901054339007e-05,
+      "loss": 0.5439,
+      "step": 1094
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5377128953771287e-05,
+      "loss": 0.3328,
+      "step": 1095
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5320356853203567e-05,
+      "loss": 0.3213,
+      "step": 1096
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5263584752635844e-05,
+      "loss": 0.3284,
+      "step": 1097
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5206812652068125e-05,
+      "loss": 0.5542,
+      "step": 1098
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5150040551500405e-05,
+      "loss": 1.1582,
+      "step": 1099
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5093268450932682e-05,
+      "loss": 0.4353,
+      "step": 1100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5036496350364962e-05,
+      "loss": 0.4209,
+      "step": 1101
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.4979724249797243e-05,
+      "loss": 0.9199,
+      "step": 1102
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.492295214922952e-05,
+      "loss": 0.5474,
+      "step": 1103
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.48661800486618e-05,
+      "loss": 0.9546,
+      "step": 1104
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4809407948094077e-05,
+      "loss": 0.488,
+      "step": 1105
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4752635847526357e-05,
+      "loss": 0.4436,
+      "step": 1106
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4695863746958638e-05,
+      "loss": 0.5791,
+      "step": 1107
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.463909164639091e-05,
+      "loss": 0.6582,
+      "step": 1108
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.458231954582319e-05,
+      "loss": 0.8828,
+      "step": 1109
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4525547445255472e-05,
+      "loss": 0.2367,
+      "step": 1110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.446877534468775e-05,
+      "loss": 1.0918,
+      "step": 1111
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.441200324412003e-05,
+      "loss": 0.4905,
+      "step": 1112
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.435523114355231e-05,
+      "loss": 0.7236,
+      "step": 1113
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.4298459042984587e-05,
+      "loss": 0.3337,
+      "step": 1114
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.4241686942416867e-05,
+      "loss": 0.4087,
+      "step": 1115
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.4184914841849147e-05,
+      "loss": 0.281,
+      "step": 1116
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.4128142741281424e-05,
+      "loss": 1.0273,
+      "step": 1117
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.4071370640713705e-05,
+      "loss": 0.7656,
+      "step": 1118
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.4014598540145985e-05,
+      "loss": 1.0479,
+      "step": 1119
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3957826439578262e-05,
+      "loss": 0.6377,
+      "step": 1120
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3901054339010542e-05,
+      "loss": 0.363,
+      "step": 1121
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.384428223844282e-05,
+      "loss": 0.4946,
+      "step": 1122
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.37875101378751e-05,
+      "loss": 1.1221,
+      "step": 1123
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.373073803730738e-05,
+      "loss": 0.5117,
+      "step": 1124
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3673965936739657e-05,
+      "loss": 0.5752,
+      "step": 1125
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3617193836171937e-05,
+      "loss": 0.3254,
+      "step": 1126
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3560421735604218e-05,
+      "loss": 0.4553,
+      "step": 1127
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.350364963503649e-05,
+      "loss": 0.385,
+      "step": 1128
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3446877534468772e-05,
+      "loss": 0.3816,
+      "step": 1129
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.3390105433901052e-05,
+      "loss": 0.5527,
+      "step": 1130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.333333333333333e-05,
+      "loss": 0.9141,
+      "step": 1131
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.327656123276561e-05,
+      "loss": 0.5244,
+      "step": 1132
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.321978913219789e-05,
+      "loss": 0.6611,
+      "step": 1133
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3163017031630167e-05,
+      "loss": 0.205,
+      "step": 1134
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3106244931062447e-05,
+      "loss": 0.8145,
+      "step": 1135
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.3049472830494727e-05,
+      "loss": 0.2952,
+      "step": 1136
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2992700729927004e-05,
+      "loss": 0.7324,
+      "step": 1137
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2935928629359285e-05,
+      "loss": 0.915,
+      "step": 1138
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2879156528791565e-05,
+      "loss": 0.6216,
+      "step": 1139
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2822384428223842e-05,
+      "loss": 0.5337,
+      "step": 1140
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2765612327656122e-05,
+      "loss": 0.4937,
+      "step": 1141
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.27088402270884e-05,
+      "loss": 0.5557,
+      "step": 1142
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.265206812652068e-05,
+      "loss": 0.4761,
+      "step": 1143
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.259529602595296e-05,
+      "loss": 0.4863,
+      "step": 1144
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2538523925385237e-05,
+      "loss": 0.3481,
+      "step": 1145
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2481751824817517e-05,
+      "loss": 0.5713,
+      "step": 1146
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2424979724249798e-05,
+      "loss": 0.3926,
+      "step": 1147
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.236820762368207e-05,
+      "loss": 0.4314,
+      "step": 1148
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2311435523114352e-05,
+      "loss": 0.5977,
+      "step": 1149
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2254663422546632e-05,
+      "loss": 0.4341,
+      "step": 1150
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.219789132197891e-05,
+      "loss": 0.8169,
+      "step": 1151
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.214111922141119e-05,
+      "loss": 0.876,
+      "step": 1152
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.208434712084347e-05,
+      "loss": 0.6338,
+      "step": 1153
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2027575020275747e-05,
+      "loss": 0.4614,
+      "step": 1154
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1970802919708027e-05,
+      "loss": 0.54,
+      "step": 1155
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1914030819140307e-05,
+      "loss": 0.2418,
+      "step": 1156
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1857258718572584e-05,
+      "loss": 0.8965,
+      "step": 1157
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1800486618004865e-05,
+      "loss": 0.174,
+      "step": 1158
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1743714517437145e-05,
+      "loss": 0.5371,
+      "step": 1159
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1686942416869422e-05,
+      "loss": 0.7256,
+      "step": 1160
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1630170316301702e-05,
+      "loss": 0.6338,
+      "step": 1161
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.157339821573398e-05,
+      "loss": 0.8237,
+      "step": 1162
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.151662611516626e-05,
+      "loss": 0.7646,
+      "step": 1163
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.145985401459854e-05,
+      "loss": 0.2607,
+      "step": 1164
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1403081914030817e-05,
+      "loss": 0.6035,
+      "step": 1165
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1346309813463097e-05,
+      "loss": 0.3855,
+      "step": 1166
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1289537712895378e-05,
+      "loss": 0.4417,
+      "step": 1167
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.123276561232765e-05,
+      "loss": 0.8057,
+      "step": 1168
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1175993511759932e-05,
+      "loss": 0.4263,
+      "step": 1169
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1119221411192212e-05,
+      "loss": 0.7349,
+      "step": 1170
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.106244931062449e-05,
+      "loss": 0.6836,
+      "step": 1171
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.100567721005677e-05,
+      "loss": 0.3931,
+      "step": 1172
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.094890510948905e-05,
+      "loss": 0.6533,
+      "step": 1173
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.0892133008921327e-05,
+      "loss": 1.002,
+      "step": 1174
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0835360908353607e-05,
+      "loss": 0.5088,
+      "step": 1175
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0778588807785887e-05,
+      "loss": 0.4814,
+      "step": 1176
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0721816707218164e-05,
+      "loss": 1.1133,
+      "step": 1177
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0665044606650445e-05,
+      "loss": 0.6992,
+      "step": 1178
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0608272506082722e-05,
+      "loss": 0.4233,
+      "step": 1179
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0551500405515002e-05,
+      "loss": 0.9634,
+      "step": 1180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0494728304947282e-05,
+      "loss": 0.8037,
+      "step": 1181
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.043795620437956e-05,
+      "loss": 0.2686,
+      "step": 1182
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.038118410381184e-05,
+      "loss": 0.625,
+      "step": 1183
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.032441200324412e-05,
+      "loss": 0.8828,
+      "step": 1184
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0267639902676397e-05,
+      "loss": 0.3313,
+      "step": 1185
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0210867802108678e-05,
+      "loss": 0.6484,
+      "step": 1186
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0154095701540958e-05,
+      "loss": 0.8306,
+      "step": 1187
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.009732360097323e-05,
+      "loss": 0.8872,
+      "step": 1188
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0040551500405512e-05,
+      "loss": 0.5132,
+      "step": 1189
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9983779399837792e-05,
+      "loss": 0.2336,
+      "step": 1190
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.992700729927007e-05,
+      "loss": 0.5596,
+      "step": 1191
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.987023519870235e-05,
+      "loss": 0.5449,
+      "step": 1192
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.981346309813463e-05,
+      "loss": 0.5312,
+      "step": 1193
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9756690997566907e-05,
+      "loss": 0.2983,
+      "step": 1194
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9699918896999187e-05,
+      "loss": 0.2976,
+      "step": 1195
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9643146796431468e-05,
+      "loss": 0.3208,
+      "step": 1196
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9586374695863744e-05,
+      "loss": 0.4143,
+      "step": 1197
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9529602595296025e-05,
+      "loss": 0.875,
+      "step": 1198
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9472830494728302e-05,
+      "loss": 0.6592,
+      "step": 1199
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9416058394160582e-05,
+      "loss": 1.1064,
+      "step": 1200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9359286293592863e-05,
+      "loss": 0.6777,
+      "step": 1201
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.930251419302514e-05,
+      "loss": 0.4482,
+      "step": 1202
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.924574209245742e-05,
+      "loss": 0.8452,
+      "step": 1203
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.91889699918897e-05,
+      "loss": 0.2957,
+      "step": 1204
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9132197891321977e-05,
+      "loss": 0.6172,
+      "step": 1205
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9075425790754258e-05,
+      "loss": 0.5103,
+      "step": 1206
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9018653690186538e-05,
+      "loss": 0.4517,
+      "step": 1207
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.896188158961881e-05,
+      "loss": 0.3567,
+      "step": 1208
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8905109489051092e-05,
+      "loss": 0.6704,
+      "step": 1209
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.8848337388483372e-05,
+      "loss": 0.6201,
+      "step": 1210
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.879156528791565e-05,
+      "loss": 0.4514,
+      "step": 1211
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.873479318734793e-05,
+      "loss": 0.8965,
+      "step": 1212
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.867802108678021e-05,
+      "loss": 0.4563,
+      "step": 1213
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8621248986212487e-05,
+      "loss": 0.4766,
+      "step": 1214
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8564476885644767e-05,
+      "loss": 0.1726,
+      "step": 1215
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8507704785077044e-05,
+      "loss": 0.45,
+      "step": 1216
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8450932684509325e-05,
+      "loss": 0.8271,
+      "step": 1217
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8394160583941605e-05,
+      "loss": 0.6323,
+      "step": 1218
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8337388483373882e-05,
+      "loss": 0.8447,
+      "step": 1219
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8280616382806162e-05,
+      "loss": 0.2271,
+      "step": 1220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8223844282238443e-05,
+      "loss": 0.7334,
+      "step": 1221
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.816707218167072e-05,
+      "loss": 0.5259,
+      "step": 1222
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8110300081103e-05,
+      "loss": 0.8145,
+      "step": 1223
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.805352798053528e-05,
+      "loss": 0.998,
+      "step": 1224
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7996755879967557e-05,
+      "loss": 0.563,
+      "step": 1225
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7939983779399838e-05,
+      "loss": 0.7578,
+      "step": 1226
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7883211678832118e-05,
+      "loss": 0.9736,
+      "step": 1227
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.782643957826439e-05,
+      "loss": 0.5923,
+      "step": 1228
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7769667477696672e-05,
+      "loss": 0.7061,
+      "step": 1229
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7712895377128952e-05,
+      "loss": 0.5552,
+      "step": 1230
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.765612327656123e-05,
+      "loss": 0.5576,
+      "step": 1231
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.759935117599351e-05,
+      "loss": 0.748,
+      "step": 1232
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.754257907542579e-05,
+      "loss": 0.3291,
+      "step": 1233
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7485806974858067e-05,
+      "loss": 0.9028,
+      "step": 1234
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7429034874290347e-05,
+      "loss": 0.5415,
+      "step": 1235
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7372262773722628e-05,
+      "loss": 0.4973,
+      "step": 1236
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7315490673154905e-05,
+      "loss": 0.4338,
+      "step": 1237
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7258718572587185e-05,
+      "loss": 0.4854,
+      "step": 1238
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7201946472019465e-05,
+      "loss": 0.6221,
+      "step": 1239
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7145174371451742e-05,
+      "loss": 0.8428,
+      "step": 1240
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.708840227088402e-05,
+      "loss": 0.8164,
+      "step": 1241
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.70316301703163e-05,
+      "loss": 0.4241,
+      "step": 1242
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.697485806974858e-05,
+      "loss": 0.4199,
+      "step": 1243
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6918085969180857e-05,
+      "loss": 0.4744,
+      "step": 1244
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6861313868613137e-05,
+      "loss": 0.1241,
+      "step": 1245
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6804541768045418e-05,
+      "loss": 0.2839,
+      "step": 1246
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6747769667477695e-05,
+      "loss": 0.6265,
+      "step": 1247
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6690997566909975e-05,
+      "loss": 0.2062,
+      "step": 1248
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6634225466342255e-05,
+      "loss": 0.5576,
+      "step": 1249
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6577453365774532e-05,
+      "loss": 0.3423,
+      "step": 1250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.652068126520681e-05,
+      "loss": 0.9883,
+      "step": 1251
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.646390916463909e-05,
+      "loss": 0.6357,
+      "step": 1252
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.640713706407137e-05,
+      "loss": 0.6514,
+      "step": 1253
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6350364963503647e-05,
+      "loss": 0.292,
+      "step": 1254
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6293592862935927e-05,
+      "loss": 0.4519,
+      "step": 1255
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6236820762368208e-05,
+      "loss": 0.5962,
+      "step": 1256
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6180048661800485e-05,
+      "loss": 0.3174,
+      "step": 1257
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6123276561232765e-05,
+      "loss": 0.5894,
+      "step": 1258
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6066504460665045e-05,
+      "loss": 0.6484,
+      "step": 1259
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6009732360097322e-05,
+      "loss": 0.3057,
+      "step": 1260
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.59529602595296e-05,
+      "loss": 0.4146,
+      "step": 1261
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.589618815896188e-05,
+      "loss": 0.4551,
+      "step": 1262
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.583941605839416e-05,
+      "loss": 0.1633,
+      "step": 1263
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5782643957826437e-05,
+      "loss": 0.4495,
+      "step": 1264
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5725871857258717e-05,
+      "loss": 0.2476,
+      "step": 1265
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5669099756690998e-05,
+      "loss": 0.6172,
+      "step": 1266
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5612327656123275e-05,
+      "loss": 0.4644,
+      "step": 1267
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 0.4917,
+      "step": 1268
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5498783454987835e-05,
+      "loss": 1.1738,
+      "step": 1269
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5442011354420112e-05,
+      "loss": 0.2059,
+      "step": 1270
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.538523925385239e-05,
+      "loss": 0.4333,
+      "step": 1271
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.532846715328467e-05,
+      "loss": 0.2751,
+      "step": 1272
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.527169505271695e-05,
+      "loss": 0.6938,
+      "step": 1273
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5214922952149227e-05,
+      "loss": 0.8027,
+      "step": 1274
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5158150851581507e-05,
+      "loss": 0.5044,
+      "step": 1275
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5101378751013786e-05,
+      "loss": 0.3484,
+      "step": 1276
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5044606650446065e-05,
+      "loss": 0.5415,
+      "step": 1277
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4987834549878343e-05,
+      "loss": 0.4456,
+      "step": 1278
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4931062449310624e-05,
+      "loss": 0.2499,
+      "step": 1279
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4874290348742902e-05,
+      "loss": 0.8345,
+      "step": 1280
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4817518248175181e-05,
+      "loss": 0.5835,
+      "step": 1281
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4760746147607461e-05,
+      "loss": 0.4609,
+      "step": 1282
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.470397404703974e-05,
+      "loss": 0.7793,
+      "step": 1283
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4647201946472017e-05,
+      "loss": 1.123,
+      "step": 1284
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4590429845904299e-05,
+      "loss": 0.7969,
+      "step": 1285
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4533657745336576e-05,
+      "loss": 0.4326,
+      "step": 1286
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4476885644768855e-05,
+      "loss": 0.7173,
+      "step": 1287
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4420113544201133e-05,
+      "loss": 0.8926,
+      "step": 1288
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4363341443633414e-05,
+      "loss": 0.4236,
+      "step": 1289
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4306569343065692e-05,
+      "loss": 0.5669,
+      "step": 1290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4249797242497971e-05,
+      "loss": 0.4136,
+      "step": 1291
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4193025141930251e-05,
+      "loss": 0.8345,
+      "step": 1292
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.413625304136253e-05,
+      "loss": 0.6509,
+      "step": 1293
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4079480940794809e-05,
+      "loss": 0.3733,
+      "step": 1294
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4022708840227086e-05,
+      "loss": 0.3032,
+      "step": 1295
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3965936739659366e-05,
+      "loss": 0.689,
+      "step": 1296
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3909164639091645e-05,
+      "loss": 0.4873,
+      "step": 1297
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3852392538523923e-05,
+      "loss": 0.2715,
+      "step": 1298
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3795620437956204e-05,
+      "loss": 0.4514,
+      "step": 1299
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3738848337388482e-05,
+      "loss": 0.5317,
+      "step": 1300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3682076236820761e-05,
+      "loss": 0.4238,
+      "step": 1301
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3625304136253041e-05,
+      "loss": 0.2284,
+      "step": 1302
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.356853203568532e-05,
+      "loss": 0.8516,
+      "step": 1303
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3511759935117599e-05,
+      "loss": 0.5547,
+      "step": 1304
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3454987834549876e-05,
+      "loss": 1.2266,
+      "step": 1305
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3398215733982156e-05,
+      "loss": 0.2224,
+      "step": 1306
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3341443633414435e-05,
+      "loss": 0.6514,
+      "step": 1307
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3284671532846713e-05,
+      "loss": 0.2062,
+      "step": 1308
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3227899432278994e-05,
+      "loss": 0.4329,
+      "step": 1309
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3171127331711272e-05,
+      "loss": 0.7588,
+      "step": 1310
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3114355231143551e-05,
+      "loss": 0.2451,
+      "step": 1311
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3057583130575831e-05,
+      "loss": 0.689,
+      "step": 1312
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.300081103000811e-05,
+      "loss": 0.4387,
+      "step": 1313
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2944038929440389e-05,
+      "loss": 0.2125,
+      "step": 1314
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2887266828872666e-05,
+      "loss": 0.5562,
+      "step": 1315
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2830494728304946e-05,
+      "loss": 0.5137,
+      "step": 1316
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2773722627737225e-05,
+      "loss": 0.958,
+      "step": 1317
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2716950527169503e-05,
+      "loss": 0.8799,
+      "step": 1318
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2660178426601784e-05,
+      "loss": 0.7983,
+      "step": 1319
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2603406326034062e-05,
+      "loss": 0.4785,
+      "step": 1320
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2546634225466341e-05,
+      "loss": 0.2351,
+      "step": 1321
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2489862124898621e-05,
+      "loss": 0.4695,
+      "step": 1322
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.24330900243309e-05,
+      "loss": 0.604,
+      "step": 1323
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2376317923763179e-05,
+      "loss": 0.6479,
+      "step": 1324
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2319545823195456e-05,
+      "loss": 1.166,
+      "step": 1325
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2262773722627736e-05,
+      "loss": 0.4951,
+      "step": 1326
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2206001622060015e-05,
+      "loss": 0.3142,
+      "step": 1327
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.2149229521492293e-05,
+      "loss": 0.2808,
+      "step": 1328
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2092457420924574e-05,
+      "loss": 0.4324,
+      "step": 1329
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.2035685320356852e-05,
+      "loss": 1.1162,
+      "step": 1330
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.1978913219789131e-05,
+      "loss": 0.9199,
+      "step": 1331
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.192214111922141e-05,
+      "loss": 0.4338,
+      "step": 1332
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.186536901865369e-05,
+      "loss": 0.7217,
+      "step": 1333
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.1808596918085969e-05,
+      "loss": 0.1216,
+      "step": 1334
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.1751824817518246e-05,
+      "loss": 0.5801,
+      "step": 1335
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.1695052716950526e-05,
+      "loss": 0.5137,
+      "step": 1336
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.1638280616382805e-05,
+      "loss": 0.5264,
+      "step": 1337
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.1581508515815083e-05,
+      "loss": 0.5063,
+      "step": 1338
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.1524736415247364e-05,
+      "loss": 0.6367,
+      "step": 1339
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.1467964314679642e-05,
+      "loss": 0.7559,
+      "step": 1340
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.1411192214111921e-05,
+      "loss": 0.6006,
+      "step": 1341
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.13544201135442e-05,
+      "loss": 0.5874,
+      "step": 1342
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.129764801297648e-05,
+      "loss": 0.5439,
+      "step": 1343
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.1240875912408759e-05,
+      "loss": 0.4546,
+      "step": 1344
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.1184103811841036e-05,
+      "loss": 0.4595,
+      "step": 1345
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.1127331711273316e-05,
+      "loss": 0.457,
+      "step": 1346
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.1070559610705595e-05,
+      "loss": 0.4153,
+      "step": 1347
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.1013787510137873e-05,
+      "loss": 0.606,
+      "step": 1348
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.0957015409570154e-05,
+      "loss": 0.2664,
+      "step": 1349
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.0900243309002432e-05,
+      "loss": 0.3906,
+      "step": 1350
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.0843471208434711e-05,
+      "loss": 0.2925,
+      "step": 1351
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.078669910786699e-05,
+      "loss": 0.8638,
+      "step": 1352
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.072992700729927e-05,
+      "loss": 0.4258,
+      "step": 1353
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.0673154906731549e-05,
+      "loss": 0.9253,
+      "step": 1354
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.0616382806163826e-05,
+      "loss": 0.5903,
+      "step": 1355
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.0559610705596106e-05,
+      "loss": 0.3267,
+      "step": 1356
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.0502838605028385e-05,
+      "loss": 0.6777,
+      "step": 1357
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.0446066504460663e-05,
+      "loss": 0.5342,
+      "step": 1358
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.0389294403892944e-05,
+      "loss": 0.7773,
+      "step": 1359
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.0332522303325222e-05,
+      "loss": 0.3628,
+      "step": 1360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.0275750202757501e-05,
+      "loss": 0.4507,
+      "step": 1361
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.021897810218978e-05,
+      "loss": 0.5913,
+      "step": 1362
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.016220600162206e-05,
+      "loss": 0.1931,
+      "step": 1363
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.0105433901054339e-05,
+      "loss": 0.5566,
+      "step": 1364
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.0048661800486616e-05,
+      "loss": 0.5449,
+      "step": 1365
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.991889699918896e-06,
+      "loss": 0.2639,
+      "step": 1366
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.935117599351175e-06,
+      "loss": 0.4834,
+      "step": 1367
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.878345498783453e-06,
+      "loss": 0.7046,
+      "step": 1368
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.821573398215734e-06,
+      "loss": 0.7881,
+      "step": 1369
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.764801297648012e-06,
+      "loss": 0.6787,
+      "step": 1370
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.708029197080291e-06,
+      "loss": 0.3464,
+      "step": 1371
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.65125709651257e-06,
+      "loss": 0.5957,
+      "step": 1372
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.59448499594485e-06,
+      "loss": 0.5264,
+      "step": 1373
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.537712895377129e-06,
+      "loss": 0.1837,
+      "step": 1374
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.480940794809406e-06,
+      "loss": 0.4675,
+      "step": 1375
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.424168694241686e-06,
+      "loss": 1.3447,
+      "step": 1376
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.367396593673965e-06,
+      "loss": 0.7939,
+      "step": 1377
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.310624493106243e-06,
+      "loss": 0.4553,
+      "step": 1378
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.253852392538522e-06,
+      "loss": 0.5186,
+      "step": 1379
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.197080291970802e-06,
+      "loss": 0.6504,
+      "step": 1380
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.140308191403081e-06,
+      "loss": 0.771,
+      "step": 1381
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.08353609083536e-06,
+      "loss": 0.9619,
+      "step": 1382
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.02676399026764e-06,
+      "loss": 0.7549,
+      "step": 1383
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.969991889699919e-06,
+      "loss": 0.6831,
+      "step": 1384
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.913219789132196e-06,
+      "loss": 0.6719,
+      "step": 1385
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.856447688564476e-06,
+      "loss": 0.2435,
+      "step": 1386
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.799675587996755e-06,
+      "loss": 0.394,
+      "step": 1387
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.742903487429033e-06,
+      "loss": 0.5737,
+      "step": 1388
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.686131386861314e-06,
+      "loss": 0.2098,
+      "step": 1389
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.629359286293592e-06,
+      "loss": 0.8467,
+      "step": 1390
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.572587185725871e-06,
+      "loss": 0.6816,
+      "step": 1391
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.51581508515815e-06,
+      "loss": 0.2177,
+      "step": 1392
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.459042984590428e-06,
+      "loss": 0.6748,
+      "step": 1393
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.402270884022709e-06,
+      "loss": 0.4956,
+      "step": 1394
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 8.345498783454987e-06,
+      "loss": 0.3037,
+      "step": 1395
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.288726682887266e-06,
+      "loss": 0.8867,
+      "step": 1396
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.231954582319545e-06,
+      "loss": 0.8599,
+      "step": 1397
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.175182481751823e-06,
+      "loss": 0.7173,
+      "step": 1398
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.118410381184104e-06,
+      "loss": 0.6348,
+      "step": 1399
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.061638280616382e-06,
+      "loss": 0.416,
+      "step": 1400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.004866180048661e-06,
+      "loss": 0.3101,
+      "step": 1401
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.94809407948094e-06,
+      "loss": 0.6797,
+      "step": 1402
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.891321978913218e-06,
+      "loss": 0.6694,
+      "step": 1403
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.834549878345499e-06,
+      "loss": 0.6025,
+      "step": 1404
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.777777777777777e-06,
+      "loss": 1.0098,
+      "step": 1405
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.721005677210056e-06,
+      "loss": 0.2991,
+      "step": 1406
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.664233576642335e-06,
+      "loss": 0.3538,
+      "step": 1407
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.6074614760746135e-06,
+      "loss": 0.2712,
+      "step": 1408
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.550689375506893e-06,
+      "loss": 0.4722,
+      "step": 1409
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.493917274939172e-06,
+      "loss": 0.7329,
+      "step": 1410
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.437145174371451e-06,
+      "loss": 0.9937,
+      "step": 1411
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.380373073803731e-06,
+      "loss": 0.345,
+      "step": 1412
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.3236009732360085e-06,
+      "loss": 0.4797,
+      "step": 1413
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.266828872668288e-06,
+      "loss": 0.8823,
+      "step": 1414
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.210056772100567e-06,
+      "loss": 0.834,
+      "step": 1415
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.153284671532846e-06,
+      "loss": 0.501,
+      "step": 1416
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.096512570965126e-06,
+      "loss": 0.8916,
+      "step": 1417
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.039740470397404e-06,
+      "loss": 0.5918,
+      "step": 1418
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.982968369829683e-06,
+      "loss": 0.2996,
+      "step": 1419
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.926196269261962e-06,
+      "loss": 0.4622,
+      "step": 1420
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.869424168694241e-06,
+      "loss": 0.4446,
+      "step": 1421
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.812652068126521e-06,
+      "loss": 0.251,
+      "step": 1422
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.755879967558799e-06,
+      "loss": 0.8315,
+      "step": 1423
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.699107866991078e-06,
+      "loss": 0.5166,
+      "step": 1424
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.642335766423357e-06,
+      "loss": 0.6924,
+      "step": 1425
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.585563665855636e-06,
+      "loss": 0.3486,
+      "step": 1426
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.528791565287916e-06,
+      "loss": 0.3926,
+      "step": 1427
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.472019464720194e-06,
+      "loss": 0.9673,
+      "step": 1428
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.415247364152473e-06,
+      "loss": 0.8521,
+      "step": 1429
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.358475263584752e-06,
+      "loss": 0.6758,
+      "step": 1430
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.301703163017031e-06,
+      "loss": 0.6196,
+      "step": 1431
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.244931062449311e-06,
+      "loss": 0.489,
+      "step": 1432
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.188158961881589e-06,
+      "loss": 0.6602,
+      "step": 1433
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.131386861313868e-06,
+      "loss": 0.2769,
+      "step": 1434
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.074614760746147e-06,
+      "loss": 0.2961,
+      "step": 1435
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.017842660178426e-06,
+      "loss": 0.6484,
+      "step": 1436
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.961070559610705e-06,
+      "loss": 1.2812,
+      "step": 1437
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.904298459042984e-06,
+      "loss": 1.0889,
+      "step": 1438
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.847526358475263e-06,
+      "loss": 0.3257,
+      "step": 1439
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.790754257907542e-06,
+      "loss": 0.6099,
+      "step": 1440
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.733982157339821e-06,
+      "loss": 0.6045,
+      "step": 1441
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.6772100567721e-06,
+      "loss": 0.7598,
+      "step": 1442
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.620437956204379e-06,
+      "loss": 0.4126,
+      "step": 1443
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.563665855636658e-06,
+      "loss": 0.2166,
+      "step": 1444
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.506893755068937e-06,
+      "loss": 0.3916,
+      "step": 1445
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.450121654501216e-06,
+      "loss": 0.5898,
+      "step": 1446
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.393349553933495e-06,
+      "loss": 0.7783,
+      "step": 1447
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.336577453365774e-06,
+      "loss": 0.5337,
+      "step": 1448
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.279805352798053e-06,
+      "loss": 0.3035,
+      "step": 1449
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.223033252230332e-06,
+      "loss": 0.5034,
+      "step": 1450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.166261151662611e-06,
+      "loss": 0.3003,
+      "step": 1451
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.10948905109489e-06,
+      "loss": 0.8774,
+      "step": 1452
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.052716950527169e-06,
+      "loss": 0.3027,
+      "step": 1453
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.995944849959448e-06,
+      "loss": 0.7339,
+      "step": 1454
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.939172749391727e-06,
+      "loss": 0.6821,
+      "step": 1455
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.882400648824006e-06,
+      "loss": 0.4023,
+      "step": 1456
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.825628548256285e-06,
+      "loss": 0.386,
+      "step": 1457
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.768856447688564e-06,
+      "loss": 0.6006,
+      "step": 1458
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.712084347120843e-06,
+      "loss": 0.5742,
+      "step": 1459
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.655312246553122e-06,
+      "loss": 0.5044,
+      "step": 1460
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.598540145985401e-06,
+      "loss": 0.353,
+      "step": 1461
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.54176804541768e-06,
+      "loss": 0.7031,
+      "step": 1462
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.484995944849959e-06,
+      "loss": 0.4141,
+      "step": 1463
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.428223844282238e-06,
+      "loss": 0.6055,
+      "step": 1464
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.371451743714517e-06,
+      "loss": 0.2998,
+      "step": 1465
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.314679643146796e-06,
+      "loss": 0.3992,
+      "step": 1466
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.257907542579075e-06,
+      "loss": 0.3438,
+      "step": 1467
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.201135442011354e-06,
+      "loss": 0.3262,
+      "step": 1468
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.144363341443633e-06,
+      "loss": 0.4243,
+      "step": 1469
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.087591240875912e-06,
+      "loss": 0.1632,
+      "step": 1470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.030819140308191e-06,
+      "loss": 0.2795,
+      "step": 1471
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.97404703974047e-06,
+      "loss": 0.4995,
+      "step": 1472
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.917274939172749e-06,
+      "loss": 0.3352,
+      "step": 1473
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.860502838605028e-06,
+      "loss": 0.3313,
+      "step": 1474
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.8037307380373067e-06,
+      "loss": 0.4207,
+      "step": 1475
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.746958637469586e-06,
+      "loss": 0.2896,
+      "step": 1476
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.6901865369018653e-06,
+      "loss": 0.5596,
+      "step": 1477
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.633414436334144e-06,
+      "loss": 0.3398,
+      "step": 1478
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.576642335766423e-06,
+      "loss": 0.2605,
+      "step": 1479
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.519870235198702e-06,
+      "loss": 0.2009,
+      "step": 1480
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.463098134630981e-06,
+      "loss": 1.0205,
+      "step": 1481
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4063260340632603e-06,
+      "loss": 0.9443,
+      "step": 1482
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.349553933495539e-06,
+      "loss": 0.7095,
+      "step": 1483
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.292781832927818e-06,
+      "loss": 0.3589,
+      "step": 1484
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.236009732360097e-06,
+      "loss": 0.582,
+      "step": 1485
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.179237631792376e-06,
+      "loss": 0.304,
+      "step": 1486
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.1224655312246553e-06,
+      "loss": 0.4224,
+      "step": 1487
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.065693430656934e-06,
+      "loss": 0.498,
+      "step": 1488
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.008921330089213e-06,
+      "loss": 0.4141,
+      "step": 1489
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.952149229521492e-06,
+      "loss": 0.5723,
+      "step": 1490
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.895377128953771e-06,
+      "loss": 0.5054,
+      "step": 1491
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.83860502838605e-06,
+      "loss": 0.6367,
+      "step": 1492
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.781832927818329e-06,
+      "loss": 0.2021,
+      "step": 1493
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.725060827250608e-06,
+      "loss": 0.6387,
+      "step": 1494
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.668288726682887e-06,
+      "loss": 0.6621,
+      "step": 1495
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.611516626115166e-06,
+      "loss": 0.9473,
+      "step": 1496
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.554744525547445e-06,
+      "loss": 0.7485,
+      "step": 1497
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.497972424979724e-06,
+      "loss": 0.5908,
+      "step": 1498
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.441200324412003e-06,
+      "loss": 0.1987,
+      "step": 1499
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.384428223844282e-06,
+      "loss": 0.7959,
+      "step": 1500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.327656123276561e-06,
+      "loss": 0.8477,
+      "step": 1501
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.27088402270884e-06,
+      "loss": 0.4907,
+      "step": 1502
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.214111922141119e-06,
+      "loss": 0.2498,
+      "step": 1503
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.157339821573398e-06,
+      "loss": 0.6602,
+      "step": 1504
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.100567721005677e-06,
+      "loss": 0.5947,
+      "step": 1505
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.043795620437956e-06,
+      "loss": 0.5444,
+      "step": 1506
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.987023519870235e-06,
+      "loss": 0.4663,
+      "step": 1507
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.930251419302514e-06,
+      "loss": 0.4692,
+      "step": 1508
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.873479318734793e-06,
+      "loss": 0.3188,
+      "step": 1509
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.816707218167072e-06,
+      "loss": 0.438,
+      "step": 1510
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.759935117599351e-06,
+      "loss": 0.7119,
+      "step": 1511
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7031630170316302e-06,
+      "loss": 0.5,
+      "step": 1512
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.646390916463909e-06,
+      "loss": 0.687,
+      "step": 1513
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.589618815896188e-06,
+      "loss": 0.2094,
+      "step": 1514
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.532846715328467e-06,
+      "loss": 0.71,
+      "step": 1515
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.476074614760746e-06,
+      "loss": 0.6973,
+      "step": 1516
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.419302514193025e-06,
+      "loss": 0.3813,
+      "step": 1517
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.362530413625304e-06,
+      "loss": 0.54,
+      "step": 1518
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.305758313057583e-06,
+      "loss": 0.3499,
+      "step": 1519
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.248986212489862e-06,
+      "loss": 0.4475,
+      "step": 1520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.192214111922141e-06,
+      "loss": 0.6743,
+      "step": 1521
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.13544201135442e-06,
+      "loss": 0.3364,
+      "step": 1522
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.078669910786699e-06,
+      "loss": 0.6958,
+      "step": 1523
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.021897810218978e-06,
+      "loss": 0.5132,
+      "step": 1524
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.65125709651257e-07,
+      "loss": 0.2908,
+      "step": 1525
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.08353609083536e-07,
+      "loss": 0.353,
+      "step": 1526
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.515815085158151e-07,
+      "loss": 0.49,
+      "step": 1527
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.94809407948094e-07,
+      "loss": 0.4644,
+      "step": 1528
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.38037307380373e-07,
+      "loss": 0.647,
+      "step": 1529
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.81265206812652e-07,
+      "loss": 0.4165,
+      "step": 1530
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.24493106244931e-07,
+      "loss": 0.3989,
+      "step": 1531
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6772100567721e-07,
+      "loss": 0.3667,
+      "step": 1532
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.10948905109489e-07,
+      "loss": 0.448,
+      "step": 1533
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.54176804541768e-07,
+      "loss": 0.6055,
+      "step": 1534
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.97404703974047e-07,
+      "loss": 0.2896,
+      "step": 1535
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.40632603406326e-07,
+      "loss": 0.7144,
+      "step": 1536
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.83860502838605e-07,
+      "loss": 0.2627,
+      "step": 1537
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.27088402270884e-07,
+      "loss": 0.7861,
+      "step": 1538
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.70316301703163e-07,
+      "loss": 0.3096,
+      "step": 1539
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.13544201135442e-07,
+      "loss": 0.375,
+      "step": 1540
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6772100567721e-08,
+      "loss": 0.3643,
+      "step": 1541
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 1.0693,
+      "step": 1542
+    },
+    {
+      "epoch": 3.0,
+      "step": 1542,
+      "total_flos": 4.951076031122375e+18,
+      "train_loss": 0.9981908340553056,
+      "train_runtime": 284.2177,
+      "train_samples_per_second": 1391.43,
+      "train_steps_per_second": 5.425
+    }
+  ],
+  "max_steps": 1542,
+  "num_train_epochs": 3,
+  "total_flos": 4.951076031122375e+18,
+  "trial_name": null,
+  "trial_params": null
+}