diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,51019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6063199942934588,
+  "eval_steps": 10000,
+  "global_step": 8500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999758950128e-05,
+      "loss": 1.3765,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999035800506e-05,
+      "loss": 0.8465,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999997830551173e-05,
+      "loss": 2.0242,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999999614320216e-05,
+      "loss": 0.7934,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999999397375352e-05,
+      "loss": 2.0823,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999991322205325e-05,
+      "loss": 0.7788,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999988188557657e-05,
+      "loss": 1.4051,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999984572810622e-05,
+      "loss": 1.0712,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999980474964333e-05,
+      "loss": 0.9605,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999997589501892e-05,
+      "loss": 2.1093,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999970832974533e-05,
+      "loss": 0.7754,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999965288831335e-05,
+      "loss": 0.8897,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999959262589502e-05,
+      "loss": 1.0497,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999995275424923e-05,
+      "loss": 0.9554,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999945763810727e-05,
+      "loss": 1.9197,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999938291274218e-05,
+      "loss": 0.918,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999993033663994e-05,
+      "loss": 1.5311,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999921899908154e-05,
+      "loss": 0.7453,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999991298107913e-05,
+      "loss": 1.4537,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999990358015315e-05,
+      "loss": 0.8122,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999989369713052e-05,
+      "loss": 2.1687,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999988333201156e-05,
+      "loss": 1.4682,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999872484796597e-05,
+      "loss": 0.8202,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999986115548599e-05,
+      "loss": 0.9731,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999984934408009e-05,
+      "loss": 1.7229,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999837050579276e-05,
+      "loss": 1.6751,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999824274983964e-05,
+      "loss": 1.4838,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999981101729454e-05,
+      "loss": 1.1949,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999797277511445e-05,
+      "loss": 1.7139,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999783055635113e-05,
+      "loss": 1.6373,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999768351666005e-05,
+      "loss": 0.8515,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999753165604596e-05,
+      "loss": 1.7342,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999737497451365e-05,
+      "loss": 0.7463,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999972134720683e-05,
+      "loss": 0.8189,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999704714871492e-05,
+      "loss": 1.342,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99996876004459e-05,
+      "loss": 0.9251,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999670003930597e-05,
+      "loss": 0.8174,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999965192532615e-05,
+      "loss": 1.6435,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999633364633145e-05,
+      "loss": 1.7805,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999614321852172e-05,
+      "loss": 1.4517,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999594796983847e-05,
+      "loss": 1.8688,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999957479002879e-05,
+      "loss": 1.9065,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999554300987656e-05,
+      "loss": 1.7128,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999533329861094e-05,
+      "loss": 1.6605,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999511876649783e-05,
+      "loss": 1.7942,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999948994135441e-05,
+      "loss": 1.4698,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999467523975682e-05,
+      "loss": 1.6167,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999444624514315e-05,
+      "loss": 2.1658,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999421242971052e-05,
+      "loss": 1.7078,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999397379346637e-05,
+      "loss": 1.4391,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999373033641844e-05,
+      "loss": 0.5369,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999934820585745e-05,
+      "loss": 1.3267,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999322895994256e-05,
+      "loss": 0.8326,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999297104053072e-05,
+      "loss": 1.3819,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999270830034735e-05,
+      "loss": 0.7997,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999244073940084e-05,
+      "loss": 1.6554,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999216835769977e-05,
+      "loss": 0.2747,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999918911552529e-05,
+      "loss": 0.6978,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999160913206918e-05,
+      "loss": 0.995,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999132228815766e-05,
+      "loss": 0.9695,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999103062352745e-05,
+      "loss": 0.8728,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999073413818812e-05,
+      "loss": 0.5612,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999043283214905e-05,
+      "loss": 0.7937,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999012670542003e-05,
+      "loss": 0.9977,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999898157580108e-05,
+      "loss": 0.7187,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999894999899314e-05,
+      "loss": 1.1189,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9998917940119196e-05,
+      "loss": 0.7396,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9998885399180286e-05,
+      "loss": 0.5447,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9998852376177446e-05,
+      "loss": 0.9083,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999881887111174e-05,
+      "loss": 1.2194,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998784883984246e-05,
+      "loss": 0.7672,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998750414796057e-05,
+      "loss": 1.3149,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999871546354828e-05,
+      "loss": 0.7333,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999868003024204e-05,
+      "loss": 1.566,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998644114878476e-05,
+      "loss": 0.9902,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998607717458736e-05,
+      "loss": 0.6239,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998570837983995e-05,
+      "loss": 0.9882,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999853347645544e-05,
+      "loss": 1.4621,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999849563287427e-05,
+      "loss": 0.9434,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998457307241703e-05,
+      "loss": 0.9577,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998418499558965e-05,
+      "loss": 1.3972,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998379209827306e-05,
+      "loss": 0.7673,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998339438047992e-05,
+      "loss": 1.1077,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998299184222298e-05,
+      "loss": 1.1799,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998258448351523e-05,
+      "loss": 1.2259,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998217230436967e-05,
+      "loss": 0.9639,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999817553047996e-05,
+      "loss": 1.1647,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999813334848184e-05,
+      "loss": 1.3586,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999809068444397e-05,
+      "loss": 0.9549,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999804753836771e-05,
+      "loss": 0.7871,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998003910254458e-05,
+      "loss": 1.3511,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999795980010561e-05,
+      "loss": 0.7557,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999791520792258e-05,
+      "loss": 1.4781,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999787013370681e-05,
+      "loss": 1.196,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997824577459746e-05,
+      "loss": 1.3236,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997778539182847e-05,
+      "loss": 1.8479,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997732018877596e-05,
+      "loss": 1.5171,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999768501654549e-05,
+      "loss": 1.5282,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999763753218804e-05,
+      "loss": 1.5439,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999758956580677e-05,
+      "loss": 1.3938,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997541117403216e-05,
+      "loss": 1.4574,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999749218697895e-05,
+      "loss": 0.477,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997442774535526e-05,
+      "loss": 1.3051,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997392880074547e-05,
+      "loss": 0.2401,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999734250359761e-05,
+      "loss": 0.5354,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997291645106337e-05,
+      "loss": 0.3847,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999724030460236e-05,
+      "loss": 0.5856,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997188482087333e-05,
+      "loss": 0.845,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997136177562913e-05,
+      "loss": 0.4907,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999708339103079e-05,
+      "loss": 0.9107,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997030122492656e-05,
+      "loss": 2.0153,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996976371950224e-05,
+      "loss": 1.7497,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996922139405225e-05,
+      "loss": 1.303,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996867424859394e-05,
+      "loss": 0.5118,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996812228314502e-05,
+      "loss": 2.2299,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999675654977231e-05,
+      "loss": 0.7482,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996700389234618e-05,
+      "loss": 2.1055,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996643746703217e-05,
+      "loss": 1.8221,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996586622179942e-05,
+      "loss": 0.5811,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996529015666622e-05,
+      "loss": 1.9689,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999647092716511e-05,
+      "loss": 0.7531,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999641235667728e-05,
+      "loss": 1.1118,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996353304205e-05,
+      "loss": 1.0174,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996293769750173e-05,
+      "loss": 1.0389,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996233753314718e-05,
+      "loss": 0.8861,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996173254900564e-05,
+      "loss": 1.225,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999611227450965e-05,
+      "loss": 1.1055,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996050812143933e-05,
+      "loss": 0.5756,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995988867805403e-05,
+      "loss": 1.406,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995926441496032e-05,
+      "loss": 1.1934,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999586353321784e-05,
+      "loss": 0.7967,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995800142972844e-05,
+      "loss": 1.3859,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995736270763084e-05,
+      "loss": 0.7007,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995671916590607e-05,
+      "loss": 1.7758,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995607080457485e-05,
+      "loss": 1.5991,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995541762365804e-05,
+      "loss": 0.9362,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999547596231766e-05,
+      "loss": 0.7088,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999540968031517e-05,
+      "loss": 0.9171,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999534291636046e-05,
+      "loss": 0.6647,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995275670455684e-05,
+      "loss": 0.8844,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995207942602997e-05,
+      "loss": 1.3135,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995139732804578e-05,
+      "loss": 0.6841,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995071041062617e-05,
+      "loss": 1.4143,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995001867379325e-05,
+      "loss": 1.1116,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994932211756923e-05,
+      "loss": 1.2127,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999486207419765e-05,
+      "loss": 1.5853,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994791454703757e-05,
+      "loss": 1.2197,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999472035327752e-05,
+      "loss": 1.3884,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999464876992122e-05,
+      "loss": 1.5134,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999457670463716e-05,
+      "loss": 1.1629,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994504157427653e-05,
+      "loss": 0.2428,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999443112829504e-05,
+      "loss": 0.613,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994357617241654e-05,
+      "loss": 0.3914,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994283624269864e-05,
+      "loss": 1.7373,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994209149382056e-05,
+      "loss": 0.6966,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999413419258061e-05,
+      "loss": 0.8186,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994058753867942e-05,
+      "loss": 0.5366,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993982833246475e-05,
+      "loss": 0.659,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993906430718657e-05,
+      "loss": 1.4879,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999382954628693e-05,
+      "loss": 0.7123,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993752179953775e-05,
+      "loss": 0.7832,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993674331721676e-05,
+      "loss": 0.7825,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993596001593133e-05,
+      "loss": 0.8445,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993517189570666e-05,
+      "loss": 0.3429,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993437895656804e-05,
+      "loss": 0.5481,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99933581198541e-05,
+      "loss": 0.665,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999327786216512e-05,
+      "loss": 1.1202,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999319712259243e-05,
+      "loss": 0.5447,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993115901138644e-05,
+      "loss": 0.5664,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993034197806363e-05,
+      "loss": 0.5139,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999295201259821e-05,
+      "loss": 0.6218,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999286934551683e-05,
+      "loss": 1.1019,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992786196564882e-05,
+      "loss": 0.798,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992702565745036e-05,
+      "loss": 0.9223,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999261845305998e-05,
+      "loss": 0.8497,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992533858512414e-05,
+      "loss": 0.7207,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992448782105062e-05,
+      "loss": 1.1288,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992363223840657e-05,
+      "loss": 1.161,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999227718372195e-05,
+      "loss": 0.87,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992190661751703e-05,
+      "loss": 0.8624,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992103657932698e-05,
+      "loss": 1.4901,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992016172267732e-05,
+      "loss": 1.6074,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991928204759617e-05,
+      "loss": 1.423,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999183975541118e-05,
+      "loss": 0.9646,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991750824225263e-05,
+      "loss": 0.6319,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991661411204727e-05,
+      "loss": 0.4981,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991571516352443e-05,
+      "loss": 0.7395,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99914811396713e-05,
+      "loss": 0.4252,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991390281164203e-05,
+      "loss": 1.379,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991298940834077e-05,
+      "loss": 0.4683,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991207118683845e-05,
+      "loss": 1.1226,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991114814716472e-05,
+      "loss": 1.0527,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991022028934918e-05,
+      "loss": 0.5555,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999092876134217e-05,
+      "loss": 1.0691,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999083501194122e-05,
+      "loss": 0.9959,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990740780735082e-05,
+      "loss": 1.1658,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990646067726786e-05,
+      "loss": 1.3981,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990550872919376e-05,
+      "loss": 0.7379,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999045519631591e-05,
+      "loss": 0.7221,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990359037919464e-05,
+      "loss": 1.0191,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999026239773313e-05,
+      "loss": 0.7425,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990165275760013e-05,
+      "loss": 0.2001,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990067672003237e-05,
+      "loss": 1.0392,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9989969586465935e-05,
+      "loss": 0.4916,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.998987101915126e-05,
+      "loss": 0.5072,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.998977197006238e-05,
+      "loss": 0.6986,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9989672439202482e-05,
+      "loss": 1.0116,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.998957242657476e-05,
+      "loss": 0.6874,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9989471932182427e-05,
+      "loss": 0.6757,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9989370956028723e-05,
+      "loss": 0.7693,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9989269498116884e-05,
+      "loss": 1.8492,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9989167558450178e-05,
+      "loss": 1.2112,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998906513703187e-05,
+      "loss": 0.6117,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998896223386526e-05,
+      "loss": 0.5276,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988858848953657e-05,
+      "loss": 1.8417,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998875498230038e-05,
+      "loss": 1.0569,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998865063390877e-05,
+      "loss": 1.9577,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998854580378218e-05,
+      "loss": 0.7601,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998844049192397e-05,
+      "loss": 0.6423,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988334698337542e-05,
+      "loss": 0.8546,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988228423026284e-05,
+      "loss": 2.089,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988121665993616e-05,
+      "loss": 0.9292,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998801442724296e-05,
+      "loss": 0.4901,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998790670677778e-05,
+      "loss": 0.3248,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987798504601527e-05,
+      "loss": 0.7664,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987689820717677e-05,
+      "loss": 0.588,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998758065512973e-05,
+      "loss": 0.7708,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987471007841192e-05,
+      "loss": 0.5872,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998736087885558e-05,
+      "loss": 1.4527,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987250268176445e-05,
+      "loss": 0.5513,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987139175807336e-05,
+      "loss": 1.3337,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987027601751826e-05,
+      "loss": 0.6061,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986915546013494e-05,
+      "loss": 1.7578,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998680300859595e-05,
+      "loss": 0.1747,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998668998950281e-05,
+      "loss": 0.8515,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.99865764887377e-05,
+      "loss": 1.1035,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986462506304272e-05,
+      "loss": 1.3064,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998634804220619e-05,
+      "loss": 0.1359,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986233096447133e-05,
+      "loss": 0.7047,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986117669030793e-05,
+      "loss": 1.3782,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986001759960888e-05,
+      "loss": 1.4152,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998588536924113e-05,
+      "loss": 0.7859,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985768496875266e-05,
+      "loss": 1.0571,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985651142867054e-05,
+      "loss": 1.2417,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985533307220266e-05,
+      "loss": 0.905,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998541498993869e-05,
+      "loss": 1.7134,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998529619102612e-05,
+      "loss": 0.7629,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998517691048639e-05,
+      "loss": 0.4634,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985057148323317e-05,
+      "loss": 0.5983,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998493690454076e-05,
+      "loss": 0.4711,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998481617914258e-05,
+      "loss": 0.5549,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9984694972132657e-05,
+      "loss": 0.2744,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998457328351489e-05,
+      "loss": 0.5607,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998445111329319e-05,
+      "loss": 1.0092,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998432846147148e-05,
+      "loss": 1.6041,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9984205328053703e-05,
+      "loss": 1.1939,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998408171304382e-05,
+      "loss": 0.6857,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983957616445797e-05,
+      "loss": 1.985,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983833038263625e-05,
+      "loss": 0.5186,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983707978501312e-05,
+      "loss": 0.6115,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983582437162875e-05,
+      "loss": 0.7975,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983456414252347e-05,
+      "loss": 1.173,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983329909773783e-05,
+      "loss": 0.6883,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998320292373125e-05,
+      "loss": 0.7631,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983075456128813e-05,
+      "loss": 0.6927,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982947506970594e-05,
+      "loss": 0.6911,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982819076260686e-05,
+      "loss": 0.8147,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982690164003223e-05,
+      "loss": 0.4218,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998256077020235e-05,
+      "loss": 0.7985,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998243089486222e-05,
+      "loss": 0.3147,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982300537987013e-05,
+      "loss": 0.4141,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998216969958092e-05,
+      "loss": 0.6379,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982038379648142e-05,
+      "loss": 0.5361,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.99819065781929e-05,
+      "loss": 1.0841,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998177429521943e-05,
+      "loss": 1.2379,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9981641530731986e-05,
+      "loss": 0.9302,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9981508284734834e-05,
+      "loss": 0.8473,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9981374557232252e-05,
+      "loss": 1.2734,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998124034822855e-05,
+      "loss": 0.6388,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9981105657728026e-05,
+      "loss": 0.4888,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998097048573502e-05,
+      "loss": 0.8031,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980834832253868e-05,
+      "loss": 0.8993,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998069869728894e-05,
+      "loss": 0.4776,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980562080844604e-05,
+      "loss": 0.981,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980424982925253e-05,
+      "loss": 0.6238,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980287403535294e-05,
+      "loss": 1.0145,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998014934267915e-05,
+      "loss": 1.0059,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980010800361254e-05,
+      "loss": 0.2006,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997987177658606e-05,
+      "loss": 1.0541,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997973227135804e-05,
+      "loss": 0.9727,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997959228468167e-05,
+      "loss": 0.2922,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997945181656146e-05,
+      "loss": 0.7019,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9979310867001915e-05,
+      "loss": 1.1058,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997916943600757e-05,
+      "loss": 1.063,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997902752358297e-05,
+      "loss": 0.6015,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997888512973268e-05,
+      "loss": 0.7154,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978742254461264e-05,
+      "loss": 0.6491,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978598897773323e-05,
+      "loss": 0.757,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997845505967347e-05,
+      "loss": 0.9207,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997831074016632e-05,
+      "loss": 0.8565,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997816593925651e-05,
+      "loss": 0.3835,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978020656948696e-05,
+      "loss": 0.2406,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997787489324755e-05,
+      "loss": 0.691,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9977728648157756e-05,
+      "loss": 0.4387,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997758192168401e-05,
+      "loss": 0.6552,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9977434713831035e-05,
+      "loss": 0.8331,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997728702460356e-05,
+      "loss": 1.6591,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9977138854006328e-05,
+      "loss": 0.5272,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9976990202044104e-05,
+      "loss": 0.6651,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997684106872166e-05,
+      "loss": 0.2164,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9976691454043798e-05,
+      "loss": 0.7968,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9976541358015327e-05,
+      "loss": 1.8874,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997639078064106e-05,
+      "loss": 0.6524,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9976239721925845e-05,
+      "loss": 0.2942,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9976088181874538e-05,
+      "loss": 0.7711,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9975936160492005e-05,
+      "loss": 0.9607,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9975783657783136e-05,
+      "loss": 0.4099,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9975630673752832e-05,
+      "loss": 1.1947,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9975477208406e-05,
+      "loss": 1.1312,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997532326174759e-05,
+      "loss": 1.2229,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997516883378254e-05,
+      "loss": 1.3697,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9975013924515807e-05,
+      "loss": 1.1513,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997485853395238e-05,
+      "loss": 0.3405,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997470266209725e-05,
+      "loss": 0.7461,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997454630895543e-05,
+      "loss": 0.6557,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9974389474531935e-05,
+      "loss": 0.6777,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9974232158831816e-05,
+      "loss": 1.2022,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9974074361860126e-05,
+      "loss": 0.6656,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9973916083621933e-05,
+      "loss": 0.8741,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997375732412233e-05,
+      "loss": 0.7916,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9973598083366414e-05,
+      "loss": 1.3013,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997343836135931e-05,
+      "loss": 1.4093,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9973278158106146e-05,
+      "loss": 0.4716,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997311747361207e-05,
+      "loss": 0.8477,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972956307882252e-05,
+      "loss": 0.1948,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972794660921868e-05,
+      "loss": 0.7546,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997263253273611e-05,
+      "loss": 0.3533,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972469923330192e-05,
+      "loss": 1.4601,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972306832709343e-05,
+      "loss": 0.6571,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972143260878804e-05,
+      "loss": 1.435,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997197920784383e-05,
+      "loss": 1.3984,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9971814673609692e-05,
+      "loss": 1.2059,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997164965818168e-05,
+      "loss": 0.8426,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9971484161565103e-05,
+      "loss": 0.7984,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9971318183765273e-05,
+      "loss": 0.6549,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9971151724787524e-05,
+      "loss": 0.2801,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9970984784637215e-05,
+      "loss": 0.8526,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.99708173633197e-05,
+      "loss": 0.5741,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9970649460840365e-05,
+      "loss": 0.7949,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9970481077204602e-05,
+      "loss": 0.2115,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9970312212417838e-05,
+      "loss": 0.6785,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.997014286648548e-05,
+      "loss": 1.1062,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9969973039412987e-05,
+      "loss": 1.4707,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9969802731205804e-05,
+      "loss": 0.7811,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996963194186942e-05,
+      "loss": 0.5495,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996946067140931e-05,
+      "loss": 0.4635,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9969288919830982e-05,
+      "loss": 1.1076,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9969116687139962e-05,
+      "loss": 1.3592,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9968943973341777e-05,
+      "loss": 1.1646,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996877077844199e-05,
+      "loss": 0.7203,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9968597102446156e-05,
+      "loss": 0.585,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996842294535986e-05,
+      "loss": 0.4449,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9968248307188702e-05,
+      "loss": 0.8173,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9968073187938293e-05,
+      "loss": 1.4929,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9967897587614264e-05,
+      "loss": 1.8908,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9967721506222253e-05,
+      "loss": 0.685,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9967544943767926e-05,
+      "loss": 0.6069,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996736790025695e-05,
+      "loss": 0.5538,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9967190375695026e-05,
+      "loss": 0.1748,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9967012370087847e-05,
+      "loss": 1.8905,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9966833883441143e-05,
+      "loss": 0.6067,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996665491576065e-05,
+      "loss": 1.7368,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9966475467052114e-05,
+      "loss": 1.3131,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9966295537321314e-05,
+      "loss": 2.1539,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996611512657402e-05,
+      "loss": 1.3738,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9965934234816032e-05,
+      "loss": 0.2598,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9965752862053175e-05,
+      "loss": 0.4273,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9965571008291267e-05,
+      "loss": 0.5249,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9965388673536158e-05,
+      "loss": 0.5676,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9965205857793707e-05,
+      "loss": 1.3819,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996502256106979e-05,
+      "loss": 0.6625,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9964838783370295e-05,
+      "loss": 1.1695,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9964654524701138e-05,
+      "loss": 1.2459,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9964469785068224e-05,
+      "loss": 1.2785,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996428456447751e-05,
+      "loss": 1.1017,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9964098862934935e-05,
+      "loss": 1.6079,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996391268044648e-05,
+      "loss": 0.8432,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9963726017018112e-05,
+      "loss": 1.1223,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9963538872655842e-05,
+      "loss": 0.6977,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9963351247365687e-05,
+      "loss": 0.4189,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996316314115367e-05,
+      "loss": 0.4508,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996297455402584e-05,
+      "loss": 0.3168,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9962785485988256e-05,
+      "loss": 1.0139,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9962595937047003e-05,
+      "loss": 0.3052,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9962405907208157e-05,
+      "loss": 1.2114,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9962215396477844e-05,
+      "loss": 0.874,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996202440486217e-05,
+      "loss": 1.2867,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9961832932367287e-05,
+      "loss": 0.8998,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9961640978999343e-05,
+      "loss": 0.7862,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9961448544764506e-05,
+      "loss": 0.8893,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9961255629668966e-05,
+      "loss": 0.2114,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9961062233718914e-05,
+      "loss": 1.1447,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9960868356920574e-05,
+      "loss": 0.6623,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996067399928018e-05,
+      "loss": 0.4047,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9960479160803965e-05,
+      "loss": 0.0994,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9960283841498206e-05,
+      "loss": 0.595,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9960088041369172e-05,
+      "loss": 0.6222,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9959891760423157e-05,
+      "loss": 0.2451,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9959694998666476e-05,
+      "loss": 0.4523,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9959497756105443e-05,
+      "loss": 0.7146,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99593000327464e-05,
+      "loss": 0.6075,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995910182859571e-05,
+      "loss": 0.6242,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995890314365973e-05,
+      "loss": 0.3816,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9958703977944857e-05,
+      "loss": 0.3516,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995850433145749e-05,
+      "loss": 0.2103,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995830420420404e-05,
+      "loss": 0.4599,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9958103596190945e-05,
+      "loss": 0.5005,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995790250742465e-05,
+      "loss": 0.5987,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995770093791162e-05,
+      "loss": 0.6961,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9957498887658325e-05,
+      "loss": 0.7288,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995729635667127e-05,
+      "loss": 0.3013,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995709334495696e-05,
+      "loss": 0.6421,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995688985252192e-05,
+      "loss": 2.0191,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995668587937269e-05,
+      "loss": 0.8917,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9956481425515825e-05,
+      "loss": 0.7563,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9956276490957895e-05,
+      "loss": 1.2259,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9956071075705494e-05,
+      "loss": 1.0131,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9955865179765213e-05,
+      "loss": 0.4833,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9955658803143678e-05,
+      "loss": 1.1743,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9955451945847513e-05,
+      "loss": 0.6852,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995524460788338e-05,
+      "loss": 1.2677,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9955036789257933e-05,
+      "loss": 1.1692,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995482848997785e-05,
+      "loss": 1.5601,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9954619710049832e-05,
+      "loss": 0.8616,
+      "step": 434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9954410449480584e-05,
+      "loss": 0.7812,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9954200708276837e-05,
+      "loss": 0.9193,
+      "step": 436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9953990486445327e-05,
+      "loss": 0.9947,
+      "step": 437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9953779783992815e-05,
+      "loss": 0.9416,
+      "step": 438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9953568600926067e-05,
+      "loss": 0.8197,
+      "step": 439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9953356937251876e-05,
+      "loss": 0.2731,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9953144792977043e-05,
+      "loss": 0.9043,
+      "step": 441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995293216810839e-05,
+      "loss": 0.7306,
+      "step": 442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995271906265274e-05,
+      "loss": 0.6951,
+      "step": 443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995250547661695e-05,
+      "loss": 0.5762,
+      "step": 444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9952291410007887e-05,
+      "loss": 0.5139,
+      "step": 445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9952076862832424e-05,
+      "loss": 1.0533,
+      "step": 446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9951861835097464e-05,
+      "loss": 0.7621,
+      "step": 447
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995164632680991e-05,
+      "loss": 0.5433,
+      "step": 448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9951430337976698e-05,
+      "loss": 1.0738,
+      "step": 449
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9951213868604767e-05,
+      "loss": 0.7553,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9950996918701063e-05,
+      "loss": 1.4164,
+      "step": 451
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9950779488272577e-05,
+      "loss": 1.6573,
+      "step": 452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995056157732628e-05,
+      "loss": 0.4969,
+      "step": 453
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9950343185869188e-05,
+      "loss": 0.5695,
+      "step": 454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9950124313908315e-05,
+      "loss": 0.6193,
+      "step": 455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9949904961450697e-05,
+      "loss": 1.2925,
+      "step": 456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9949685128503376e-05,
+      "loss": 1.5384,
+      "step": 457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9949464815073435e-05,
+      "loss": 0.3516,
+      "step": 458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9949244021167938e-05,
+      "loss": 0.3286,
+      "step": 459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9949022746793992e-05,
+      "loss": 0.8387,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99488009919587e-05,
+      "loss": 0.8673,
+      "step": 461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99485787566692e-05,
+      "loss": 0.3987,
+      "step": 462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9948356040932622e-05,
+      "loss": 0.6789,
+      "step": 463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9948132844756136e-05,
+      "loss": 0.4687,
+      "step": 464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994790916814691e-05,
+      "loss": 0.488,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9947685011112127e-05,
+      "loss": 0.8064,
+      "step": 466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9947460373659e-05,
+      "loss": 0.7538,
+      "step": 467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994723525579475e-05,
+      "loss": 0.6173,
+      "step": 468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9947009657526606e-05,
+      "loss": 0.5615,
+      "step": 469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9946783578861824e-05,
+      "loss": 0.3935,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9946557019807663e-05,
+      "loss": 1.3617,
+      "step": 471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9946329980371408e-05,
+      "loss": 0.3915,
+      "step": 472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9946102460560362e-05,
+      "loss": 0.2531,
+      "step": 473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994587446038183e-05,
+      "loss": 0.7226,
+      "step": 474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9945645979843143e-05,
+      "loss": 0.5897,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9945417018951643e-05,
+      "loss": 0.725,
+      "step": 476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994518757771469e-05,
+      "loss": 0.7072,
+      "step": 477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9944957656139657e-05,
+      "loss": 1.7122,
+      "step": 478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9944727254233936e-05,
+      "loss": 1.2856,
+      "step": 479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9944496372004934e-05,
+      "loss": 1.3938,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9944265009460062e-05,
+      "loss": 2.5845,
+      "step": 481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9944033166606763e-05,
+      "loss": 0.8775,
+      "step": 482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9943800843452494e-05,
+      "loss": 0.909,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9943568040004707e-05,
+      "loss": 0.8332,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99433347562709e-05,
+      "loss": 0.8881,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9943100992258557e-05,
+      "loss": 0.6004,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9942866747975203e-05,
+      "loss": 1.2214,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994263202342836e-05,
+      "loss": 1.4629,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9942396818625572e-05,
+      "loss": 0.7656,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99421611335744e-05,
+      "loss": 0.3609,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.994192496828242e-05,
+      "loss": 1.0734,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9941688322757218e-05,
+      "loss": 0.3235,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9941451197006406e-05,
+      "loss": 1.1538,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9941213591037604e-05,
+      "loss": 0.6937,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9940975504858445e-05,
+      "loss": 1.0034,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9940736938476583e-05,
+      "loss": 0.6381,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9940497891899684e-05,
+      "loss": 1.0199,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.994025836513543e-05,
+      "loss": 0.8213,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9940018358191525e-05,
+      "loss": 0.6842,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993977787107568e-05,
+      "loss": 0.7075,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9939536903795622e-05,
+      "loss": 0.1483,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9939295456359098e-05,
+      "loss": 0.3279,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9939053528773867e-05,
+      "loss": 0.1561,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9938811121047705e-05,
+      "loss": 0.3968,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9938568233188405e-05,
+      "loss": 0.8352,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9938324865203772e-05,
+      "loss": 1.1887,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9938081017101624e-05,
+      "loss": 0.4795,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9937836688889807e-05,
+      "loss": 0.4676,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993759188057616e-05,
+      "loss": 0.6786,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9937346592168565e-05,
+      "loss": 1.2996,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99371008236749e-05,
+      "loss": 0.4906,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9936854575103064e-05,
+      "loss": 1.0579,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993660784646097e-05,
+      "loss": 0.6014,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993636063775655e-05,
+      "loss": 0.5159,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993611294899775e-05,
+      "loss": 0.7773,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9935864780192528e-05,
+      "loss": 0.5817,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9935616131348863e-05,
+      "loss": 0.3059,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993536700247474e-05,
+      "loss": 1.0407,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9935117393578173e-05,
+      "loss": 2.2004,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9934867304667187e-05,
+      "loss": 0.9538,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9934616735749813e-05,
+      "loss": 0.6122,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99343656868341e-05,
+      "loss": 0.9644,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993411415792813e-05,
+      "loss": 0.5231,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993386214903998e-05,
+      "loss": 0.7125,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993360966017775e-05,
+      "loss": 0.6234,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9933356691349555e-05,
+      "loss": 1.1371,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9933103242563525e-05,
+      "loss": 0.7484,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9932849313827807e-05,
+      "loss": 1.286,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993259490515056e-05,
+      "loss": 1.5883,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9932340016539962e-05,
+      "loss": 0.6572,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9932084648004207e-05,
+      "loss": 0.576,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9931828799551496e-05,
+      "loss": 0.887,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9931572471190063e-05,
+      "loss": 1.1768,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9931315662928138e-05,
+      "loss": 1.3332,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9931058374773977e-05,
+      "loss": 0.9296,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9930800606735848e-05,
+      "loss": 0.9275,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9930542358822038e-05,
+      "loss": 0.3005,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9930283631040847e-05,
+      "loss": 0.388,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9930024423400588e-05,
+      "loss": 0.9647,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992976473590959e-05,
+      "loss": 0.9033,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992950456857621e-05,
+      "loss": 0.519,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99292439214088e-05,
+      "loss": 1.0241,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9928982794415737e-05,
+      "loss": 0.3346,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9928721187605415e-05,
+      "loss": 0.8096,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992845910098625e-05,
+      "loss": 0.6017,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9928196534566655e-05,
+      "loss": 0.9009,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9927933488355073e-05,
+      "loss": 0.3454,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992766996235996e-05,
+      "loss": 0.7731,
+      "step": 548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992740595658978e-05,
+      "loss": 0.6193,
+      "step": 549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9927141471053025e-05,
+      "loss": 0.968,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992687650575819e-05,
+      "loss": 1.3816,
+      "step": 551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9926611060713793e-05,
+      "loss": 1.7112,
+      "step": 552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992634513592837e-05,
+      "loss": 0.9549,
+      "step": 553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992607873141046e-05,
+      "loss": 0.5637,
+      "step": 554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992581184716863e-05,
+      "loss": 1.5673,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9925544483211456e-05,
+      "loss": 0.5425,
+      "step": 556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992527663954753e-05,
+      "loss": 0.6209,
+      "step": 557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992500831618547e-05,
+      "loss": 0.7288,
+      "step": 558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9924739513133886e-05,
+      "loss": 0.3908,
+      "step": 559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992447023040142e-05,
+      "loss": 0.3972,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9924200467996736e-05,
+      "loss": 1.3837,
+      "step": 561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9923930225928497e-05,
+      "loss": 1.6296,
+      "step": 562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9923659504205385e-05,
+      "loss": 1.5193,
+      "step": 563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992338830283611e-05,
+      "loss": 0.5386,
+      "step": 564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9923116621829386e-05,
+      "loss": 1.2319,
+      "step": 565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992284446119394e-05,
+      "loss": 0.7148,
+      "step": 566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9922571820938523e-05,
+      "loss": 0.218,
+      "step": 567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99222987010719e-05,
+      "loss": 1.3695,
+      "step": 568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992202510160284e-05,
+      "loss": 1.1522,
+      "step": 569
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9921751022540148e-05,
+      "loss": 0.6351,
+      "step": 570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9921476463892622e-05,
+      "loss": 0.7183,
+      "step": 571
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9921201425669093e-05,
+      "loss": 0.7003,
+      "step": 572
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9920925907878406e-05,
+      "loss": 0.5684,
+      "step": 573
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9920649910529404e-05,
+      "loss": 0.7982,
+      "step": 574
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992037343363096e-05,
+      "loss": 0.9444,
+      "step": 575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992009647719197e-05,
+      "loss": 0.7973,
+      "step": 576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9919819041221322e-05,
+      "loss": 0.6029,
+      "step": 577
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9919541125727942e-05,
+      "loss": 1.0407,
+      "step": 578
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991926273072076e-05,
+      "loss": 0.7409,
+      "step": 579
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9918983856208726e-05,
+      "loss": 1.4256,
+      "step": 580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9918704502200796e-05,
+      "loss": 1.6862,
+      "step": 581
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9918424668705953e-05,
+      "loss": 1.4191,
+      "step": 582
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991814435573319e-05,
+      "loss": 1.4715,
+      "step": 583
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991786356329152e-05,
+      "loss": 0.7654,
+      "step": 584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9917582291389963e-05,
+      "loss": 0.673,
+      "step": 585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9917300540037557e-05,
+      "loss": 1.3082,
+      "step": 586
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991701830924336e-05,
+      "loss": 0.5866,
+      "step": 587
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991673559901645e-05,
+      "loss": 1.3762,
+      "step": 588
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99164524093659e-05,
+      "loss": 1.2141,
+      "step": 589
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9916168740300822e-05,
+      "loss": 0.1723,
+      "step": 590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9915884591830332e-05,
+      "loss": 0.141,
+      "step": 591
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991559996396356e-05,
+      "loss": 1.2093,
+      "step": 592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991531485670965e-05,
+      "loss": 1.3145,
+      "step": 593
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9915029270077773e-05,
+      "loss": 0.311,
+      "step": 594
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9914743204077103e-05,
+      "loss": 0.793,
+      "step": 595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9914456658716838e-05,
+      "loss": 0.156,
+      "step": 596
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991416963400618e-05,
+      "loss": 0.764,
+      "step": 597
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9913882129954365e-05,
+      "loss": 0.449,
+      "step": 598
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9913594146570627e-05,
+      "loss": 0.4745,
+      "step": 599
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9913305683864217e-05,
+      "loss": 0.7808,
+      "step": 600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9913016741844415e-05,
+      "loss": 0.143,
+      "step": 601
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9912727320520506e-05,
+      "loss": 0.1744,
+      "step": 602
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9912437419901784e-05,
+      "loss": 1.1472,
+      "step": 603
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991214703999758e-05,
+      "loss": 0.4761,
+      "step": 604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9911856180817215e-05,
+      "loss": 1.7887,
+      "step": 605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991156484237004e-05,
+      "loss": 0.4888,
+      "step": 606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991127302466542e-05,
+      "loss": 1.8356,
+      "step": 607
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9910980727712738e-05,
+      "loss": 1.8214,
+      "step": 608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991068795152138e-05,
+      "loss": 0.7004,
+      "step": 609
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9910394696100764e-05,
+      "loss": 0.3392,
+      "step": 610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991010096146031e-05,
+      "loss": 0.4477,
+      "step": 611
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990980674760946e-05,
+      "loss": 1.4819,
+      "step": 612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990951205455767e-05,
+      "loss": 1.0139,
+      "step": 613
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9909216882314413e-05,
+      "loss": 0.8825,
+      "step": 614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9908921230889172e-05,
+      "loss": 1.4321,
+      "step": 615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9908625100291454e-05,
+      "loss": 0.6488,
+      "step": 616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9908328490530776e-05,
+      "loss": 0.69,
+      "step": 617
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9908031401616664e-05,
+      "loss": 1.4098,
+      "step": 618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9907733833558678e-05,
+      "loss": 0.5706,
+      "step": 619
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990743578636637e-05,
+      "loss": 0.2073,
+      "step": 620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9907137260049325e-05,
+      "loss": 0.6266,
+      "step": 621
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9906838254617142e-05,
+      "loss": 0.6691,
+      "step": 622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990653877007942e-05,
+      "loss": 0.5711,
+      "step": 623
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9906238806445797e-05,
+      "loss": 0.8139,
+      "step": 624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9905938363725907e-05,
+      "loss": 1.1896,
+      "step": 625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9905637441929406e-05,
+      "loss": 1.0121,
+      "step": 626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9905336041065965e-05,
+      "loss": 1.2083,
+      "step": 627
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990503416114527e-05,
+      "loss": 1.4228,
+      "step": 628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9904731802177033e-05,
+      "loss": 0.7859,
+      "step": 629
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990442896417096e-05,
+      "loss": 0.7621,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9904125647136793e-05,
+      "loss": 1.3591,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9903821851084273e-05,
+      "loss": 1.0802,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.990351757602317e-05,
+      "loss": 0.3501,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9903212821963257e-05,
+      "loss": 0.9227,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9902907588914338e-05,
+      "loss": 0.5626,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9902601876886217e-05,
+      "loss": 0.9318,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.990229568588872e-05,
+      "loss": 1.2879,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9901989015931686e-05,
+      "loss": 0.4128,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9901681867024975e-05,
+      "loss": 1.2454,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.990137423917846e-05,
+      "loss": 0.1628,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9901066132402025e-05,
+      "loss": 1.0323,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.990075754670557e-05,
+      "loss": 0.6997,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9900448482099024e-05,
+      "loss": 0.5695,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9900138938592302e-05,
+      "loss": 0.7115,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989982891619537e-05,
+      "loss": 0.7272,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9899518414918184e-05,
+      "loss": 0.3342,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989920743477073e-05,
+      "loss": 0.6703,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9898895975762994e-05,
+      "loss": 0.4781,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989858403790499e-05,
+      "loss": 0.5635,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9898271621206742e-05,
+      "loss": 0.7926,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9897958725678295e-05,
+      "loss": 0.5461,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9897645351329703e-05,
+      "loss": 1.1882,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9897331498171038e-05,
+      "loss": 0.6995,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989701716621239e-05,
+      "loss": 1.5697,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9896702355463854e-05,
+      "loss": 1.2401,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989638706593556e-05,
+      "loss": 1.7431,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989607129763763e-05,
+      "loss": 0.5568,
+      "step": 657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989575505058022e-05,
+      "loss": 1.3333,
+      "step": 658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9895438324773485e-05,
+      "loss": 1.7163,
+      "step": 659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9895121120227618e-05,
+      "loss": 0.1996,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9894803436952803e-05,
+      "loss": 0.3989,
+      "step": 661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9894485274959257e-05,
+      "loss": 0.2774,
+      "step": 662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9894166634257203e-05,
+      "loss": 0.4981,
+      "step": 663
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989384751485688e-05,
+      "loss": 0.5703,
+      "step": 664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9893527916768545e-05,
+      "loss": 1.4094,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9893207840002475e-05,
+      "loss": 1.2284,
+      "step": 666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9892887284568953e-05,
+      "loss": 0.599,
+      "step": 667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9892566250478287e-05,
+      "loss": 0.6528,
+      "step": 668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9892244737740783e-05,
+      "loss": 1.4036,
+      "step": 669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9891922746366782e-05,
+      "loss": 1.109,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989160027636664e-05,
+      "loss": 0.5524,
+      "step": 671
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989127732775071e-05,
+      "loss": 1.1252,
+      "step": 672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9890953900529373e-05,
+      "loss": 0.7611,
+      "step": 673
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9890629994713027e-05,
+      "loss": 0.5337,
+      "step": 674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989030561031208e-05,
+      "loss": 0.4992,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9889980747336965e-05,
+      "loss": 0.4987,
+      "step": 676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9889655405798113e-05,
+      "loss": 1.0264,
+      "step": 677
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9889329585705988e-05,
+      "loss": 1.1805,
+      "step": 678
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988900328707106e-05,
+      "loss": 0.8459,
+      "step": 679
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988867650990381e-05,
+      "loss": 0.5386,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9888349254214747e-05,
+      "loss": 0.8899,
+      "step": 681
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988802152001439e-05,
+      "loss": 0.9034,
+      "step": 682
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988769330731327e-05,
+      "loss": 1.0591,
+      "step": 683
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9887364616121936e-05,
+      "loss": 1.1058,
+      "step": 684
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9887035446450947e-05,
+      "loss": 0.8154,
+      "step": 685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9886705798310894e-05,
+      "loss": 0.2141,
+      "step": 686
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988637567171236e-05,
+      "loss": 1.2417,
+      "step": 687
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9886045066665967e-05,
+      "loss": 0.8259,
+      "step": 688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9885713983182326e-05,
+      "loss": 0.6783,
+      "step": 689
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9885382421272088e-05,
+      "loss": 0.782,
+      "step": 690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9885050380945915e-05,
+      "loss": 0.9542,
+      "step": 691
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9884717862214466e-05,
+      "loss": 0.4134,
+      "step": 692
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988438486508843e-05,
+      "loss": 0.6152,
+      "step": 693
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988405138957852e-05,
+      "loss": 0.6151,
+      "step": 694
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9883717435695445e-05,
+      "loss": 0.4285,
+      "step": 695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9883383003449934e-05,
+      "loss": 0.4072,
+      "step": 696
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988304809285275e-05,
+      "loss": 0.6039,
+      "step": 697
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9882712703914646e-05,
+      "loss": 0.5972,
+      "step": 698
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9882376836646404e-05,
+      "loss": 0.8737,
+      "step": 699
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988204049105882e-05,
+      "loss": 0.6462,
+      "step": 700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9881703667162703e-05,
+      "loss": 1.8465,
+      "step": 701
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9881366364968877e-05,
+      "loss": 0.1637,
+      "step": 702
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9881028584488184e-05,
+      "loss": 0.6858,
+      "step": 703
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9880690325731486e-05,
+      "loss": 1.2579,
+      "step": 704
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9880351588709642e-05,
+      "loss": 0.8804,
+      "step": 705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988001237343355e-05,
+      "loss": 0.6756,
+      "step": 706
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987967267991411e-05,
+      "loss": 0.5856,
+      "step": 707
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987933250816224e-05,
+      "loss": 1.9668,
+      "step": 708
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9878991858188868e-05,
+      "loss": 0.7719,
+      "step": 709
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987865073000495e-05,
+      "loss": 0.3519,
+      "step": 710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9878309123621446e-05,
+      "loss": 0.6618,
+      "step": 711
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987796703904933e-05,
+      "loss": 0.6118,
+      "step": 712
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9877624476299605e-05,
+      "loss": 0.3085,
+      "step": 713
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987728143538328e-05,
+      "loss": 0.9492,
+      "step": 714
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987693791631138e-05,
+      "loss": 0.6606,
+      "step": 715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9876593919094935e-05,
+      "loss": 0.7021,
+      "step": 716
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9876249443745014e-05,
+      "loss": 0.523,
+      "step": 717
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9875904490272686e-05,
+      "loss": 0.2954,
+      "step": 718
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9875559058689036e-05,
+      "loss": 0.5993,
+      "step": 719
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9875213149005168e-05,
+      "loss": 1.2121,
+      "step": 720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9874866761232197e-05,
+      "loss": 1.2995,
+      "step": 721
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9874519895381254e-05,
+      "loss": 0.4406,
+      "step": 722
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987417255146349e-05,
+      "loss": 0.7343,
+      "step": 723
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9873824729490067e-05,
+      "loss": 1.9843,
+      "step": 724
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987347642947217e-05,
+      "loss": 0.4574,
+      "step": 725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987312765142099e-05,
+      "loss": 0.7322,
+      "step": 726
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9872778395347732e-05,
+      "loss": 1.4835,
+      "step": 727
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9872428661263628e-05,
+      "loss": 0.8479,
+      "step": 728
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9872078449179913e-05,
+      "loss": 0.3854,
+      "step": 729
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9871727759107845e-05,
+      "loss": 1.0441,
+      "step": 730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9871376591058697e-05,
+      "loss": 0.323,
+      "step": 731
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987102494504375e-05,
+      "loss": 0.5046,
+      "step": 732
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9870672821074313e-05,
+      "loss": 0.867,
+      "step": 733
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.98703202191617e-05,
+      "loss": 0.8821,
+      "step": 734
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986996713931724e-05,
+      "loss": 0.665,
+      "step": 735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9869613581552288e-05,
+      "loss": 1.012,
+      "step": 736
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9869259545878206e-05,
+      "loss": 0.9045,
+      "step": 737
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9868905032306367e-05,
+      "loss": 1.1273,
+      "step": 738
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986855004084817e-05,
+      "loss": 1.1516,
+      "step": 739
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986819457151502e-05,
+      "loss": 0.4875,
+      "step": 740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986783862431835e-05,
+      "loss": 1.0883,
+      "step": 741
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986748219926959e-05,
+      "loss": 0.1626,
+      "step": 742
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.98671252963802e-05,
+      "loss": 0.8024,
+      "step": 743
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9866767915661654e-05,
+      "loss": 0.3829,
+      "step": 744
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9866410057125437e-05,
+      "loss": 0.4716,
+      "step": 745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9866051720783047e-05,
+      "loss": 0.3886,
+      "step": 746
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9865692906646004e-05,
+      "loss": 0.4895,
+      "step": 747
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986533361472584e-05,
+      "loss": 0.4047,
+      "step": 748
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9864973845034098e-05,
+      "loss": 0.647,
+      "step": 749
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9864613597582345e-05,
+      "loss": 0.4907,
+      "step": 750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9864252872382167e-05,
+      "loss": 1.1996,
+      "step": 751
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9863891669445144e-05,
+      "loss": 1.3857,
+      "step": 752
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986352998878289e-05,
+      "loss": 0.2635,
+      "step": 753
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9863167830407034e-05,
+      "loss": 0.5669,
+      "step": 754
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9862805194329208e-05,
+      "loss": 0.866,
+      "step": 755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9862442080561075e-05,
+      "loss": 0.6947,
+      "step": 756
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.98620784891143e-05,
+      "loss": 1.4251,
+      "step": 757
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9861714420000573e-05,
+      "loss": 0.4982,
+      "step": 758
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986134987323159e-05,
+      "loss": 0.234,
+      "step": 759
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9860984848819073e-05,
+      "loss": 0.7659,
+      "step": 760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986061934677475e-05,
+      "loss": 0.5232,
+      "step": 761
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986025336711037e-05,
+      "loss": 0.519,
+      "step": 762
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.985988690983769e-05,
+      "loss": 0.5848,
+      "step": 763
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9859519974968498e-05,
+      "loss": 0.3917,
+      "step": 764
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9859152562514583e-05,
+      "loss": 0.4327,
+      "step": 765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.985878467248775e-05,
+      "loss": 0.4392,
+      "step": 766
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9858416304899822e-05,
+      "loss": 0.3361,
+      "step": 767
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.985804745976265e-05,
+      "loss": 1.3461,
+      "step": 768
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9857678137088078e-05,
+      "loss": 0.7643,
+      "step": 769
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.985730833688798e-05,
+      "loss": 2.0162,
+      "step": 770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9856938059174236e-05,
+      "loss": 0.6745,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9856567303958753e-05,
+      "loss": 0.9636,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9856196071253448e-05,
+      "loss": 0.745,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9855824361070242e-05,
+      "loss": 0.2257,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9855452173421096e-05,
+      "loss": 0.7546,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9855079508317967e-05,
+      "loss": 0.8715,
+      "step": 776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9854706365772825e-05,
+      "loss": 2.1071,
+      "step": 777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.985433274579767e-05,
+      "loss": 0.7766,
+      "step": 778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9853958648404508e-05,
+      "loss": 0.4357,
+      "step": 779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.985358407360537e-05,
+      "loss": 0.4849,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9853209021412283e-05,
+      "loss": 0.258,
+      "step": 781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.98528334918373e-05,
+      "loss": 0.3133,
+      "step": 782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9852457484892506e-05,
+      "loss": 1.1372,
+      "step": 783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9852081000589972e-05,
+      "loss": 0.6104,
+      "step": 784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9851704038941803e-05,
+      "loss": 0.5715,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9851326599960112e-05,
+      "loss": 1.2441,
+      "step": 786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.985094868365704e-05,
+      "loss": 0.8722,
+      "step": 787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9850570290044718e-05,
+      "loss": 1.1815,
+      "step": 788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.985019141913532e-05,
+      "loss": 1.403,
+      "step": 789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984981207094101e-05,
+      "loss": 0.1757,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9849432245473997e-05,
+      "loss": 1.0214,
+      "step": 791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9849051942746475e-05,
+      "loss": 0.4757,
+      "step": 792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984867116277067e-05,
+      "loss": 1.5894,
+      "step": 793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9848289905558828e-05,
+      "loss": 0.4365,
+      "step": 794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984790817112319e-05,
+      "loss": 0.9287,
+      "step": 795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9847525959476036e-05,
+      "loss": 0.6139,
+      "step": 796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9847143270629644e-05,
+      "loss": 0.3367,
+      "step": 797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9846760104596315e-05,
+      "loss": 0.3483,
+      "step": 798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9846376461388364e-05,
+      "loss": 0.6767,
+      "step": 799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984599234101812e-05,
+      "loss": 0.2854,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9845607743497935e-05,
+      "loss": 0.3731,
+      "step": 801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984522266884016e-05,
+      "loss": 0.657,
+      "step": 802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984483711705718e-05,
+      "loss": 0.5412,
+      "step": 803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9844451088161383e-05,
+      "loss": 1.6755,
+      "step": 804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9844064582165175e-05,
+      "loss": 0.6557,
+      "step": 805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9843677599080976e-05,
+      "loss": 0.5379,
+      "step": 806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984329013892123e-05,
+      "loss": 0.706,
+      "step": 807
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9842902201698386e-05,
+      "loss": 1.534,
+      "step": 808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9842513787424918e-05,
+      "loss": 0.9928,
+      "step": 809
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9842124896113298e-05,
+      "loss": 1.8131,
+      "step": 810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9841735527776033e-05,
+      "loss": 0.6418,
+      "step": 811
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9841345682425634e-05,
+      "loss": 0.5124,
+      "step": 812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9840955360074642e-05,
+      "loss": 0.4893,
+      "step": 813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9840564560735586e-05,
+      "loss": 0.5215,
+      "step": 814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984017328442103e-05,
+      "loss": 1.5597,
+      "step": 815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983978153114356e-05,
+      "loss": 1.0864,
+      "step": 816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9839389300915753e-05,
+      "loss": 1.725,
+      "step": 817
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9838996593750224e-05,
+      "loss": 0.9251,
+      "step": 818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9838603409659595e-05,
+      "loss": 0.436,
+      "step": 819
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9838209748656494e-05,
+      "loss": 0.3741,
+      "step": 820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9837815610753586e-05,
+      "loss": 0.5423,
+      "step": 821
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983742099596353e-05,
+      "loss": 0.8987,
+      "step": 822
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9837025904299013e-05,
+      "loss": 1.3972,
+      "step": 823
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9836630335772724e-05,
+      "loss": 0.5693,
+      "step": 824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983623429039739e-05,
+      "loss": 1.6204,
+      "step": 825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983583776818573e-05,
+      "loss": 0.6601,
+      "step": 826
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9835440769150496e-05,
+      "loss": 0.5965,
+      "step": 827
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983504329330444e-05,
+      "loss": 0.5237,
+      "step": 828
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983464534066034e-05,
+      "loss": 1.2563,
+      "step": 829
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9834246911230993e-05,
+      "loss": 0.5967,
+      "step": 830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9833848005029194e-05,
+      "loss": 1.3614,
+      "step": 831
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983344862206777e-05,
+      "loss": 0.9826,
+      "step": 832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9833048762359548e-05,
+      "loss": 1.0273,
+      "step": 833
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9832648425917393e-05,
+      "loss": 0.8302,
+      "step": 834
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983224761275416e-05,
+      "loss": 0.5429,
+      "step": 835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983184632288274e-05,
+      "loss": 0.6032,
+      "step": 836
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983144455631603e-05,
+      "loss": 1.2967,
+      "step": 837
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9831042313066936e-05,
+      "loss": 0.663,
+      "step": 838
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983063959314839e-05,
+      "loss": 1.0531,
+      "step": 839
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9830236396573334e-05,
+      "loss": 0.8302,
+      "step": 840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982983272335473e-05,
+      "loss": 0.7276,
+      "step": 841
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982942857350555e-05,
+      "loss": 0.7889,
+      "step": 842
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9829023947038785e-05,
+      "loss": 0.7824,
+      "step": 843
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9828618843967437e-05,
+      "loss": 0.9999,
+      "step": 844
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982821326430452e-05,
+      "loss": 0.9503,
+      "step": 845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9827807208063084e-05,
+      "loss": 0.5431,
+      "step": 846
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9827400675256175e-05,
+      "loss": 0.6307,
+      "step": 847
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982699366589685e-05,
+      "loss": 0.8052,
+      "step": 848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9826586179998196e-05,
+      "loss": 0.6728,
+      "step": 849
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9826178217573313e-05,
+      "loss": 0.7914,
+      "step": 850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9825769778635308e-05,
+      "loss": 1.2991,
+      "step": 851
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9825360863197312e-05,
+      "loss": 0.2803,
+      "step": 852
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9824951471272465e-05,
+      "loss": 0.3089,
+      "step": 853
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982454160287392e-05,
+      "loss": 1.3199,
+      "step": 854
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9824131258014858e-05,
+      "loss": 1.1866,
+      "step": 855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9823720436708468e-05,
+      "loss": 0.4476,
+      "step": 856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982330913896795e-05,
+      "loss": 0.2733,
+      "step": 857
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9822897364806523e-05,
+      "loss": 0.1851,
+      "step": 858
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982248511423742e-05,
+      "loss": 1.2564,
+      "step": 859
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9822072387273896e-05,
+      "loss": 1.7125,
+      "step": 860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9821659183929214e-05,
+      "loss": 0.7121,
+      "step": 861
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982124550421665e-05,
+      "loss": 0.6267,
+      "step": 862
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9820831348149507e-05,
+      "loss": 0.4892,
+      "step": 863
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982041671574109e-05,
+      "loss": 1.1857,
+      "step": 864
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9820001607004723e-05,
+      "loss": 0.6631,
+      "step": 865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981958602195376e-05,
+      "loss": 0.1848,
+      "step": 866
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9819169960601545e-05,
+      "loss": 1.1689,
+      "step": 867
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9818753422961455e-05,
+      "loss": 1.4489,
+      "step": 868
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981833640904688e-05,
+      "loss": 0.456,
+      "step": 869
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9817918918871218e-05,
+      "loss": 1.6581,
+      "step": 870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9817500952447892e-05,
+      "loss": 1.3701,
+      "step": 871
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9817082509790327e-05,
+      "loss": 0.1849,
+      "step": 872
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981666359091198e-05,
+      "loss": 2.4518,
+      "step": 873
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9816244195826317e-05,
+      "loss": 1.1614,
+      "step": 874
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981582432454681e-05,
+      "loss": 0.9318,
+      "step": 875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9815403977086956e-05,
+      "loss": 0.7512,
+      "step": 876
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9814983153460268e-05,
+      "loss": 0.6449,
+      "step": 877
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981456185368027e-05,
+      "loss": 1.1307,
+      "step": 878
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.98141400777605e-05,
+      "loss": 0.6562,
+      "step": 879
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981371782571451e-05,
+      "loss": 0.9998,
+      "step": 880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9813295097555882e-05,
+      "loss": 0.5524,
+      "step": 881
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9812871893298196e-05,
+      "loss": 1.3063,
+      "step": 882
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9812448212955054e-05,
+      "loss": 0.6604,
+      "step": 883
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9812024056540073e-05,
+      "loss": 0.8133,
+      "step": 884
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981159942406689e-05,
+      "loss": 1.2649,
+      "step": 885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9811174315549147e-05,
+      "loss": 0.4101,
+      "step": 886
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9810748731000506e-05,
+      "loss": 0.7147,
+      "step": 887
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9810322670434653e-05,
+      "loss": 0.2153,
+      "step": 888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9809896133865276e-05,
+      "loss": 1.102,
+      "step": 889
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9809469121306084e-05,
+      "loss": 0.3925,
+      "step": 890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.98090416327708e-05,
+      "loss": 0.9051,
+      "step": 891
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9808613668273166e-05,
+      "loss": 0.2233,
+      "step": 892
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980818522782694e-05,
+      "loss": 0.2151,
+      "step": 893
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980775631144588e-05,
+      "loss": 0.2598,
+      "step": 894
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9807326919143786e-05,
+      "loss": 0.8539,
+      "step": 895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980689705093445e-05,
+      "loss": 0.9786,
+      "step": 896
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9806466706831688e-05,
+      "loss": 0.6382,
+      "step": 897
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9806035886849335e-05,
+      "loss": 0.7848,
+      "step": 898
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980560459100124e-05,
+      "loss": 0.91,
+      "step": 899
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9805172819301255e-05,
+      "loss": 0.856,
+      "step": 900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9804740571763264e-05,
+      "loss": 0.4116,
+      "step": 901
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980430784840116e-05,
+      "loss": 1.0912,
+      "step": 902
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980387464922885e-05,
+      "loss": 0.4106,
+      "step": 903
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9803440974260253e-05,
+      "loss": 1.4465,
+      "step": 904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9803006823509313e-05,
+      "loss": 1.7108,
+      "step": 905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980257219698998e-05,
+      "loss": 0.471,
+      "step": 906
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9802137094716222e-05,
+      "loss": 0.695,
+      "step": 907
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9801701516702026e-05,
+      "loss": 0.4002,
+      "step": 908
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980126546296139e-05,
+      "loss": 0.2765,
+      "step": 909
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980082893350833e-05,
+      "loss": 1.5977,
+      "step": 910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9800391928356878e-05,
+      "loss": 1.0767,
+      "step": 911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9799954447521072e-05,
+      "loss": 0.418,
+      "step": 912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9799516491014983e-05,
+      "loss": 1.8383,
+      "step": 913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9799078058852676e-05,
+      "loss": 1.4465,
+      "step": 914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.979863915104825e-05,
+      "loss": 0.9612,
+      "step": 915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9798199767615808e-05,
+      "loss": 1.9441,
+      "step": 916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9797759908569473e-05,
+      "loss": 0.7633,
+      "step": 917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.979731957392338e-05,
+      "loss": 1.119,
+      "step": 918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9796878763691687e-05,
+      "loss": 0.5621,
+      "step": 919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9796437477888555e-05,
+      "loss": 0.4596,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.979599571652817e-05,
+      "loss": 0.4298,
+      "step": 921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.979555347962473e-05,
+      "loss": 0.9593,
+      "step": 922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9795110767192448e-05,
+      "loss": 0.4346,
+      "step": 923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9794667579245552e-05,
+      "loss": 0.604,
+      "step": 924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9794223915798287e-05,
+      "loss": 0.7285,
+      "step": 925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9793779776864916e-05,
+      "loss": 0.4707,
+      "step": 926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9793335162459707e-05,
+      "loss": 0.9277,
+      "step": 927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9792890072596954e-05,
+      "loss": 2.2679,
+      "step": 928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.979244450729096e-05,
+      "loss": 0.4684,
+      "step": 929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9791998466556044e-05,
+      "loss": 0.7535,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.979155195040655e-05,
+      "loss": 0.6208,
+      "step": 931
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9791104958856822e-05,
+      "loss": 0.6476,
+      "step": 932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9790657491921226e-05,
+      "loss": 1.4163,
+      "step": 933
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9790209549614146e-05,
+      "loss": 0.4522,
+      "step": 934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9789761131949977e-05,
+      "loss": 0.3377,
+      "step": 935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9789312238943135e-05,
+      "loss": 0.6109,
+      "step": 936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9788862870608044e-05,
+      "loss": 0.4592,
+      "step": 937
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9788413026959144e-05,
+      "loss": 0.9014,
+      "step": 938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97879627080109e-05,
+      "loss": 0.6926,
+      "step": 939
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978751191377778e-05,
+      "loss": 1.1108,
+      "step": 940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9787060644274275e-05,
+      "loss": 0.6459,
+      "step": 941
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9786608899514887e-05,
+      "loss": 0.6632,
+      "step": 942
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9786156679514136e-05,
+      "loss": 1.1219,
+      "step": 943
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9785703984286556e-05,
+      "loss": 0.5653,
+      "step": 944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97852508138467e-05,
+      "loss": 0.7207,
+      "step": 945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9784797168209127e-05,
+      "loss": 0.6952,
+      "step": 946
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978434304738842e-05,
+      "loss": 0.4408,
+      "step": 947
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9783888451399174e-05,
+      "loss": 0.8809,
+      "step": 948
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9783433380256e-05,
+      "loss": 0.5699,
+      "step": 949
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978297783397353e-05,
+      "loss": 0.7623,
+      "step": 950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978252181256639e-05,
+      "loss": 1.1275,
+      "step": 951
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9782065316049255e-05,
+      "loss": 0.3508,
+      "step": 952
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9781608344436778e-05,
+      "loss": 0.6992,
+      "step": 953
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9781150897743663e-05,
+      "loss": 1.4284,
+      "step": 954
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9780692975984602e-05,
+      "loss": 0.5363,
+      "step": 955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9780234579174315e-05,
+      "loss": 1.4317,
+      "step": 956
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9779775707327538e-05,
+      "loss": 0.2845,
+      "step": 957
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9779316360459016e-05,
+      "loss": 0.4542,
+      "step": 958
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977885653858351e-05,
+      "loss": 0.6382,
+      "step": 959
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9778396241715805e-05,
+      "loss": 0.3568,
+      "step": 960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977793546987069e-05,
+      "loss": 0.4677,
+      "step": 961
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977747422306298e-05,
+      "loss": 0.5054,
+      "step": 962
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9777012501307487e-05,
+      "loss": 1.5919,
+      "step": 963
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9776550304619067e-05,
+      "loss": 0.763,
+      "step": 964
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9776087633012565e-05,
+      "loss": 0.8916,
+      "step": 965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977562448650285e-05,
+      "loss": 0.5452,
+      "step": 966
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9775160865104815e-05,
+      "loss": 1.0649,
+      "step": 967
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977469676883335e-05,
+      "loss": 0.3331,
+      "step": 968
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9774232197703383e-05,
+      "loss": 1.1918,
+      "step": 969
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977376715172984e-05,
+      "loss": 0.7338,
+      "step": 970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9773301630927667e-05,
+      "loss": 1.1845,
+      "step": 971
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9772835635311827e-05,
+      "loss": 0.9874,
+      "step": 972
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9772369164897296e-05,
+      "loss": 1.0567,
+      "step": 973
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9771902219699066e-05,
+      "loss": 1.5323,
+      "step": 974
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9771434799732145e-05,
+      "loss": 0.6199,
+      "step": 975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9770966905011557e-05,
+      "loss": 0.9619,
+      "step": 976
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977049853555234e-05,
+      "loss": 1.1139,
+      "step": 977
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977002969136955e-05,
+      "loss": 0.9078,
+      "step": 978
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9769560372478247e-05,
+      "loss": 0.435,
+      "step": 979
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976909057889352e-05,
+      "loss": 0.2125,
+      "step": 980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976862031063047e-05,
+      "loss": 0.6946,
+      "step": 981
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976814956770421e-05,
+      "loss": 0.8615,
+      "step": 982
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976767835012987e-05,
+      "loss": 1.1745,
+      "step": 983
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9767206657922594e-05,
+      "loss": 1.5939,
+      "step": 984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9766734491097543e-05,
+      "loss": 0.2304,
+      "step": 985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9766261849669893e-05,
+      "loss": 0.2159,
+      "step": 986
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9765788733654835e-05,
+      "loss": 0.9941,
+      "step": 987
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976531514306757e-05,
+      "loss": 0.2684,
+      "step": 988
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9764841077923323e-05,
+      "loss": 1.064,
+      "step": 989
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9764366538237335e-05,
+      "loss": 0.7203,
+      "step": 990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976389152402485e-05,
+      "loss": 0.9756,
+      "step": 991
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9763416035301136e-05,
+      "loss": 1.5435,
+      "step": 992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976294007208148e-05,
+      "loss": 0.5478,
+      "step": 993
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9762463634381173e-05,
+      "loss": 0.9686,
+      "step": 994
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976198672221554e-05,
+      "loss": 1.0614,
+      "step": 995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976150933559989e-05,
+      "loss": 0.9037,
+      "step": 996
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9761031474549585e-05,
+      "loss": 0.7484,
+      "step": 997
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976055313907997e-05,
+      "loss": 0.879,
+      "step": 998
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9760074329206425e-05,
+      "loss": 0.5735,
+      "step": 999
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975959504494434e-05,
+      "loss": 0.5406,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9759115286309112e-05,
+      "loss": 0.3301,
+      "step": 1001
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975863505331617e-05,
+      "loss": 1.6343,
+      "step": 1002
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9758154345980938e-05,
+      "loss": 0.6,
+      "step": 1003
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9757673164318873e-05,
+      "loss": 1.7375,
+      "step": 1004
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9757191508345443e-05,
+      "loss": 0.4464,
+      "step": 1005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975670937807612e-05,
+      "loss": 1.4036,
+      "step": 1006
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975622677352641e-05,
+      "loss": 0.551,
+      "step": 1007
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9755743694711813e-05,
+      "loss": 0.3018,
+      "step": 1008
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9755260141647863e-05,
+      "loss": 0.3948,
+      "step": 1009
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9754776114350098e-05,
+      "loss": 0.6865,
+      "step": 1010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9754291612834076e-05,
+      "loss": 0.3001,
+      "step": 1011
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9753806637115363e-05,
+      "loss": 1.0699,
+      "step": 1012
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9753321187209558e-05,
+      "loss": 0.3944,
+      "step": 1013
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9752835263132252e-05,
+      "loss": 0.9181,
+      "step": 1014
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975234886489907e-05,
+      "loss": 0.1641,
+      "step": 1015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975186199252564e-05,
+      "loss": 0.9112,
+      "step": 1016
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975137464602761e-05,
+      "loss": 0.7322,
+      "step": 1017
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9750886825420646e-05,
+      "loss": 0.5251,
+      "step": 1018
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9750398530720426e-05,
+      "loss": 0.1295,
+      "step": 1019
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974990976194265e-05,
+      "loss": 0.5846,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9749420519103015e-05,
+      "loss": 0.4502,
+      "step": 1021
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974893080221725e-05,
+      "loss": 0.9123,
+      "step": 1022
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9748440611301094e-05,
+      "loss": 0.7735,
+      "step": 1023
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974794994637031e-05,
+      "loss": 0.823,
+      "step": 1024
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9747458807440657e-05,
+      "loss": 1.1445,
+      "step": 1025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974696719452792e-05,
+      "loss": 1.0638,
+      "step": 1026
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9746475107647905e-05,
+      "loss": 0.7744,
+      "step": 1027
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974598254681643e-05,
+      "loss": 0.6045,
+      "step": 1028
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9745489512049325e-05,
+      "loss": 1.0046,
+      "step": 1029
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974499600336243e-05,
+      "loss": 1.5903,
+      "step": 1030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9744502020771608e-05,
+      "loss": 0.7622,
+      "step": 1031
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9744007564292737e-05,
+      "loss": 0.9851,
+      "step": 1032
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974351263394171e-05,
+      "loss": 1.1635,
+      "step": 1033
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9743017229734434e-05,
+      "loss": 1.126,
+      "step": 1034
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9742521351686832e-05,
+      "loss": 0.8878,
+      "step": 1035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974202499981484e-05,
+      "loss": 0.5218,
+      "step": 1036
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9741528174134412e-05,
+      "loss": 0.4948,
+      "step": 1037
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974103087466151e-05,
+      "loss": 1.0953,
+      "step": 1038
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9740533101412125e-05,
+      "loss": 0.7753,
+      "step": 1039
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9740034854402252e-05,
+      "loss": 1.9002,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97395361336479e-05,
+      "loss": 0.3582,
+      "step": 1041
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.973903693916511e-05,
+      "loss": 0.3172,
+      "step": 1042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9738537270969917e-05,
+      "loss": 1.5297,
+      "step": 1043
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9738037129078382e-05,
+      "loss": 0.3031,
+      "step": 1044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.973753651350658e-05,
+      "loss": 0.9481,
+      "step": 1045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9737035424270596e-05,
+      "loss": 0.673,
+      "step": 1046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9736533861386543e-05,
+      "loss": 0.5143,
+      "step": 1047
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9736031824870543e-05,
+      "loss": 0.7905,
+      "step": 1048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.973552931473872e-05,
+      "loss": 0.5522,
+      "step": 1049
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9735026331007232e-05,
+      "loss": 0.6414,
+      "step": 1050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9734522873692244e-05,
+      "loss": 0.3981,
+      "step": 1051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9734018942809937e-05,
+      "loss": 1.8226,
+      "step": 1052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.973351453837651e-05,
+      "loss": 0.3626,
+      "step": 1053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.973300966040817e-05,
+      "loss": 1.5782,
+      "step": 1054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9732504308921144e-05,
+      "loss": 1.2202,
+      "step": 1055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9731998483931675e-05,
+      "loss": 0.787,
+      "step": 1056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.973149218545602e-05,
+      "loss": 0.1914,
+      "step": 1057
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.973098541351046e-05,
+      "loss": 0.5692,
+      "step": 1058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9730478168111272e-05,
+      "loss": 0.184,
+      "step": 1059
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972997044927476e-05,
+      "loss": 0.5102,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9729462257017237e-05,
+      "loss": 0.4516,
+      "step": 1061
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972895359135505e-05,
+      "loss": 1.0164,
+      "step": 1062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972844445230454e-05,
+      "loss": 0.3895,
+      "step": 1063
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972793483988207e-05,
+      "loss": 0.9863,
+      "step": 1064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9727424754104022e-05,
+      "loss": 0.4317,
+      "step": 1065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9726914194986788e-05,
+      "loss": 0.6995,
+      "step": 1066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9726403162546777e-05,
+      "loss": 0.3391,
+      "step": 1067
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9725891656800418e-05,
+      "loss": 1.1505,
+      "step": 1068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9725379677764144e-05,
+      "loss": 0.5945,
+      "step": 1069
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9724867225454414e-05,
+      "loss": 0.7453,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9724354299887692e-05,
+      "loss": 0.4098,
+      "step": 1071
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9723840901080474e-05,
+      "loss": 0.2985,
+      "step": 1072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972332702904926e-05,
+      "loss": 0.5463,
+      "step": 1073
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972281268381055e-05,
+      "loss": 0.7508,
+      "step": 1074
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9722297865380894e-05,
+      "loss": 0.8901,
+      "step": 1075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9721782573776833e-05,
+      "loss": 0.8638,
+      "step": 1076
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972126680901492e-05,
+      "loss": 0.5067,
+      "step": 1077
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972075057111174e-05,
+      "loss": 0.654,
+      "step": 1078
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972023386008388e-05,
+      "loss": 0.7626,
+      "step": 1079
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971971667594795e-05,
+      "loss": 0.7189,
+      "step": 1080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9719199018720576e-05,
+      "loss": 0.6496,
+      "step": 1081
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9718680888418387e-05,
+      "loss": 0.8386,
+      "step": 1082
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9718162285058044e-05,
+      "loss": 0.4641,
+      "step": 1083
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9717643208656208e-05,
+      "loss": 1.0721,
+      "step": 1084
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9717123659229565e-05,
+      "loss": 0.9141,
+      "step": 1085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9716603636794814e-05,
+      "loss": 0.3017,
+      "step": 1086
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971608314136867e-05,
+      "loss": 0.5588,
+      "step": 1087
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9715562172967855e-05,
+      "loss": 0.8546,
+      "step": 1088
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9715040731609125e-05,
+      "loss": 0.827,
+      "step": 1089
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9714518817309227e-05,
+      "loss": 0.6364,
+      "step": 1090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9713996430084936e-05,
+      "loss": 0.4492,
+      "step": 1091
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971347356995305e-05,
+      "loss": 1.0531,
+      "step": 1092
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971295023693037e-05,
+      "loss": 0.6398,
+      "step": 1093
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9712426431033713e-05,
+      "loss": 0.9127,
+      "step": 1094
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971190215227992e-05,
+      "loss": 1.3736,
+      "step": 1095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9711377400685834e-05,
+      "loss": 0.1838,
+      "step": 1096
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9710852176268325e-05,
+      "loss": 0.9494,
+      "step": 1097
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971032647904427e-05,
+      "loss": 0.723,
+      "step": 1098
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9709800309030574e-05,
+      "loss": 1.5363,
+      "step": 1099
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970927366624414e-05,
+      "loss": 0.5538,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970874655070189e-05,
+      "loss": 0.303,
+      "step": 1101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9708218962420773e-05,
+      "loss": 1.827,
+      "step": 1102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970769090141775e-05,
+      "loss": 1.5139,
+      "step": 1103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9707162367709784e-05,
+      "loss": 1.5729,
+      "step": 1104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970663336131387e-05,
+      "loss": 0.3694,
+      "step": 1105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9706103882247e-05,
+      "loss": 0.4518,
+      "step": 1106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9705573930526193e-05,
+      "loss": 1.1556,
+      "step": 1107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970504350616849e-05,
+      "loss": 1.2532,
+      "step": 1108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9704512609190933e-05,
+      "loss": 0.5303,
+      "step": 1109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970398123961059e-05,
+      "loss": 0.3668,
+      "step": 1110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9703449397444533e-05,
+      "loss": 0.1662,
+      "step": 1111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970291708270986e-05,
+      "loss": 0.3436,
+      "step": 1112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9702384295423677e-05,
+      "loss": 1.2115,
+      "step": 1113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9701851035603106e-05,
+      "loss": 1.0371,
+      "step": 1114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970131730326529e-05,
+      "loss": 0.1677,
+      "step": 1115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970078309842738e-05,
+      "loss": 0.7339,
+      "step": 1116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9700248421106545e-05,
+      "loss": 0.5085,
+      "step": 1117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9699713271319976e-05,
+      "loss": 0.4819,
+      "step": 1118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9699177649084867e-05,
+      "loss": 0.4136,
+      "step": 1119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9698641554418432e-05,
+      "loss": 1.8047,
+      "step": 1120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9698104987337904e-05,
+      "loss": 0.7772,
+      "step": 1121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.969756794786053e-05,
+      "loss": 0.5915,
+      "step": 1122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.969703043600356e-05,
+      "loss": 1.7493,
+      "step": 1123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9696492451784285e-05,
+      "loss": 0.8168,
+      "step": 1124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9695953995219982e-05,
+      "loss": 0.8137,
+      "step": 1125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.969541506632797e-05,
+      "loss": 1.2438,
+      "step": 1126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.969487566512556e-05,
+      "loss": 0.7803,
+      "step": 1127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9694335791630088e-05,
+      "loss": 0.9307,
+      "step": 1128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9693795445858912e-05,
+      "loss": 0.3013,
+      "step": 1129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9693254627829396e-05,
+      "loss": 1.2434,
+      "step": 1130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9692713337558923e-05,
+      "loss": 1.2772,
+      "step": 1131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9692171575064885e-05,
+      "loss": 1.2287,
+      "step": 1132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9691629340364704e-05,
+      "loss": 0.6465,
+      "step": 1133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9691086633475795e-05,
+      "loss": 0.6391,
+      "step": 1134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.969054345441561e-05,
+      "loss": 0.6802,
+      "step": 1135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9689999803201604e-05,
+      "loss": 0.8625,
+      "step": 1136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9689455679851247e-05,
+      "loss": 0.4291,
+      "step": 1137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9688911084382032e-05,
+      "loss": 0.4183,
+      "step": 1138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968836601681146e-05,
+      "loss": 0.9413,
+      "step": 1139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968782047715705e-05,
+      "loss": 1.17,
+      "step": 1140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9687274465436332e-05,
+      "loss": 0.9689,
+      "step": 1141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9686727981666864e-05,
+      "loss": 0.575,
+      "step": 1142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.96861810258662e-05,
+      "loss": 0.2597,
+      "step": 1143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968563359805193e-05,
+      "loss": 0.7654,
+      "step": 1144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9685085698241632e-05,
+      "loss": 0.5447,
+      "step": 1145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9684537326452927e-05,
+      "loss": 0.613,
+      "step": 1146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968398848270344e-05,
+      "loss": 0.6045,
+      "step": 1147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9683439167010805e-05,
+      "loss": 0.5878,
+      "step": 1148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968288937939268e-05,
+      "loss": 0.8565,
+      "step": 1149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9682339119866745e-05,
+      "loss": 0.6027,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9681788388450666e-05,
+      "loss": 0.2958,
+      "step": 1151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9681237185162155e-05,
+      "loss": 0.456,
+      "step": 1152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9680685510018928e-05,
+      "loss": 1.2949,
+      "step": 1153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9680133363038715e-05,
+      "loss": 1.6037,
+      "step": 1154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.967958074423926e-05,
+      "loss": 0.397,
+      "step": 1155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9679027653638323e-05,
+      "loss": 0.3956,
+      "step": 1156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9678474091253685e-05,
+      "loss": 0.4606,
+      "step": 1157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9677920057103133e-05,
+      "loss": 1.7485,
+      "step": 1158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9677365551204477e-05,
+      "loss": 1.3751,
+      "step": 1159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9676810573575537e-05,
+      "loss": 1.8788,
+      "step": 1160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9676255124234152e-05,
+      "loss": 0.5339,
+      "step": 1161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.967569920319817e-05,
+      "loss": 0.406,
+      "step": 1162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.967514281048546e-05,
+      "loss": 0.2996,
+      "step": 1163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.967458594611391e-05,
+      "loss": 0.3261,
+      "step": 1164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9674028610101407e-05,
+      "loss": 1.1719,
+      "step": 1165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9673470802465873e-05,
+      "loss": 0.9088,
+      "step": 1166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9672912523225233e-05,
+      "loss": 0.4854,
+      "step": 1167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9672353772397427e-05,
+      "loss": 1.3498,
+      "step": 1168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9671794550000417e-05,
+      "loss": 0.6161,
+      "step": 1169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9671234856052176e-05,
+      "loss": 1.2851,
+      "step": 1170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.967067469057069e-05,
+      "loss": 1.1063,
+      "step": 1171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9670114053573967e-05,
+      "loss": 1.3366,
+      "step": 1172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966955294508002e-05,
+      "loss": 1.28,
+      "step": 1173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966899136510689e-05,
+      "loss": 0.6946,
+      "step": 1174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966842931367262e-05,
+      "loss": 0.4216,
+      "step": 1175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9667866790795277e-05,
+      "loss": 1.4016,
+      "step": 1176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966730379649294e-05,
+      "loss": 1.7573,
+      "step": 1177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9666740330783708e-05,
+      "loss": 0.4975,
+      "step": 1178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9666176393685683e-05,
+      "loss": 1.5985,
+      "step": 1179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9665611985216994e-05,
+      "loss": 0.6463,
+      "step": 1180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966504710539578e-05,
+      "loss": 0.6026,
+      "step": 1181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9664481754240198e-05,
+      "loss": 0.8557,
+      "step": 1182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966391593176842e-05,
+      "loss": 0.7954,
+      "step": 1183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9663349637998625e-05,
+      "loss": 0.8484,
+      "step": 1184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966278287294902e-05,
+      "loss": 0.1332,
+      "step": 1185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9662215636637817e-05,
+      "loss": 0.511,
+      "step": 1186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9661647929083246e-05,
+      "loss": 0.8456,
+      "step": 1187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966107975030356e-05,
+      "loss": 0.8205,
+      "step": 1188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9660511100317014e-05,
+      "loss": 0.7778,
+      "step": 1189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9659941979141887e-05,
+      "loss": 1.1563,
+      "step": 1190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9659372386796467e-05,
+      "loss": 0.5743,
+      "step": 1191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9658802323299063e-05,
+      "loss": 0.6515,
+      "step": 1192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9658231788668e-05,
+      "loss": 1.3516,
+      "step": 1193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9657660782921613e-05,
+      "loss": 0.9265,
+      "step": 1194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.965708930607825e-05,
+      "loss": 1.3534,
+      "step": 1195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9656517358156284e-05,
+      "loss": 0.7063,
+      "step": 1196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9655944939174095e-05,
+      "loss": 1.0286,
+      "step": 1197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.965537204915008e-05,
+      "loss": 0.3473,
+      "step": 1198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.965479868810265e-05,
+      "loss": 0.3675,
+      "step": 1199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9654224856050237e-05,
+      "loss": 0.8686,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.965365055301128e-05,
+      "loss": 0.3013,
+      "step": 1201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9653075779004246e-05,
+      "loss": 0.7494,
+      "step": 1202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9652500534047593e-05,
+      "loss": 1.3259,
+      "step": 1203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.965192481815982e-05,
+      "loss": 0.6244,
+      "step": 1204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9651348631359426e-05,
+      "loss": 1.6282,
+      "step": 1205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9650771973664937e-05,
+      "loss": 0.5879,
+      "step": 1206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9650194845094878e-05,
+      "loss": 0.7783,
+      "step": 1207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96496172456678e-05,
+      "loss": 1.5696,
+      "step": 1208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9649039175402272e-05,
+      "loss": 1.2149,
+      "step": 1209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.964846063431687e-05,
+      "loss": 1.0164,
+      "step": 1210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9647881622430183e-05,
+      "loss": 0.622,
+      "step": 1211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.964730213976083e-05,
+      "loss": 0.5903,
+      "step": 1212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9646722186327427e-05,
+      "loss": 0.5364,
+      "step": 1213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.964614176214862e-05,
+      "loss": 0.4667,
+      "step": 1214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9645560867243062e-05,
+      "loss": 0.5733,
+      "step": 1215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.964497950162942e-05,
+      "loss": 0.2538,
+      "step": 1216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.964439766532638e-05,
+      "loss": 0.3849,
+      "step": 1217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9643815358352647e-05,
+      "loss": 0.8021,
+      "step": 1218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.964323258072694e-05,
+      "loss": 0.9764,
+      "step": 1219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9642649332467967e-05,
+      "loss": 1.9394,
+      "step": 1220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96420656135945e-05,
+      "loss": 0.5053,
+      "step": 1221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9641481424125286e-05,
+      "loss": 1.4392,
+      "step": 1222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9640896764079103e-05,
+      "loss": 0.869,
+      "step": 1223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9640311633474743e-05,
+      "loss": 0.5238,
+      "step": 1224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963972603233101e-05,
+      "loss": 0.5098,
+      "step": 1225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963913996066673e-05,
+      "loss": 0.5179,
+      "step": 1226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9638553418500733e-05,
+      "loss": 1.3423,
+      "step": 1227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9637966405851876e-05,
+      "loss": 0.2629,
+      "step": 1228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9637378922739022e-05,
+      "loss": 1.1825,
+      "step": 1229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9636790969181054e-05,
+      "loss": 0.8631,
+      "step": 1230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9636202545196868e-05,
+      "loss": 0.7252,
+      "step": 1231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963561365080538e-05,
+      "loss": 0.7714,
+      "step": 1232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9635024286025512e-05,
+      "loss": 1.3558,
+      "step": 1233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963443445087621e-05,
+      "loss": 0.8908,
+      "step": 1234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9633844145376425e-05,
+      "loss": 0.3189,
+      "step": 1235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963325336954513e-05,
+      "loss": 1.1034,
+      "step": 1236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9632662123401324e-05,
+      "loss": 0.6465,
+      "step": 1237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9632070406964e-05,
+      "loss": 0.3391,
+      "step": 1238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963147822025218e-05,
+      "loss": 0.4522,
+      "step": 1239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9630885563284886e-05,
+      "loss": 0.4094,
+      "step": 1240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9630292436081182e-05,
+      "loss": 0.3144,
+      "step": 1241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962969883866012e-05,
+      "loss": 0.7906,
+      "step": 1242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9629104771040785e-05,
+      "loss": 0.7,
+      "step": 1243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9628510233242262e-05,
+      "loss": 1.1935,
+      "step": 1244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9627915225283668e-05,
+      "loss": 0.8062,
+      "step": 1245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9627319747184123e-05,
+      "loss": 0.7525,
+      "step": 1246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9626723798962765e-05,
+      "loss": 0.4995,
+      "step": 1247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9626127380638746e-05,
+      "loss": 0.4688,
+      "step": 1248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9625530492231237e-05,
+      "loss": 0.6305,
+      "step": 1249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9624933133759427e-05,
+      "loss": 0.543,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9624335305242508e-05,
+      "loss": 0.7525,
+      "step": 1251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96237370066997e-05,
+      "loss": 0.4077,
+      "step": 1252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9623138238150223e-05,
+      "loss": 1.043,
+      "step": 1253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9622538999613333e-05,
+      "loss": 1.663,
+      "step": 1254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9621939291108275e-05,
+      "loss": 0.5246,
+      "step": 1255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962133911265434e-05,
+      "loss": 0.3023,
+      "step": 1256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962073846427081e-05,
+      "loss": 1.7629,
+      "step": 1257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9620137345976986e-05,
+      "loss": 0.6454,
+      "step": 1258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9619535757792194e-05,
+      "loss": 0.5011,
+      "step": 1259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9618933699735763e-05,
+      "loss": 0.6047,
+      "step": 1260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9618331171827056e-05,
+      "loss": 0.5557,
+      "step": 1261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9617728174085423e-05,
+      "loss": 0.4064,
+      "step": 1262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9617124706530253e-05,
+      "loss": 1.0158,
+      "step": 1263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9616520769180934e-05,
+      "loss": 0.3721,
+      "step": 1264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9615916362056887e-05,
+      "loss": 0.2858,
+      "step": 1265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9615311485177536e-05,
+      "loss": 1.2161,
+      "step": 1266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.961470613856231e-05,
+      "loss": 1.2208,
+      "step": 1267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9614100322230674e-05,
+      "loss": 0.4251,
+      "step": 1268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96134940362021e-05,
+      "loss": 1.0411,
+      "step": 1269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9612887280496072e-05,
+      "loss": 0.4242,
+      "step": 1270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9612280055132087e-05,
+      "loss": 0.9607,
+      "step": 1271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9611672360129667e-05,
+      "loss": 0.716,
+      "step": 1272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9611064195508347e-05,
+      "loss": 0.5752,
+      "step": 1273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9610455561287664e-05,
+      "loss": 0.7996,
+      "step": 1274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960984645748718e-05,
+      "loss": 0.4107,
+      "step": 1275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9609236884126478e-05,
+      "loss": 1.3455,
+      "step": 1276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960862684122515e-05,
+      "loss": 1.2466,
+      "step": 1277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9608016328802793e-05,
+      "loss": 0.5921,
+      "step": 1278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960740534687904e-05,
+      "loss": 0.8827,
+      "step": 1279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9606793895473518e-05,
+      "loss": 0.8977,
+      "step": 1280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9606181974605886e-05,
+      "loss": 0.5757,
+      "step": 1281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9605569584295814e-05,
+      "loss": 1.0363,
+      "step": 1282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9604956724562976e-05,
+      "loss": 0.9014,
+      "step": 1283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960434339542707e-05,
+      "loss": 0.3815,
+      "step": 1284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9603729596907813e-05,
+      "loss": 0.3223,
+      "step": 1285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9603115329024936e-05,
+      "loss": 0.3059,
+      "step": 1286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9602500591798173e-05,
+      "loss": 0.7249,
+      "step": 1287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9601885385247282e-05,
+      "loss": 0.9811,
+      "step": 1288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9601269709392037e-05,
+      "loss": 0.8626,
+      "step": 1289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9600653564252232e-05,
+      "loss": 0.5167,
+      "step": 1290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9600036949847666e-05,
+      "loss": 0.6028,
+      "step": 1291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9599419866198154e-05,
+      "loss": 0.684,
+      "step": 1292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.959880231332353e-05,
+      "loss": 0.1528,
+      "step": 1293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9598184291243646e-05,
+      "loss": 0.5415,
+      "step": 1294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.959756579997836e-05,
+      "loss": 0.7431,
+      "step": 1295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9596946839547558e-05,
+      "loss": 0.3149,
+      "step": 1296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9596327409971126e-05,
+      "loss": 0.523,
+      "step": 1297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9595707511268973e-05,
+      "loss": 0.6504,
+      "step": 1298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9595087143461027e-05,
+      "loss": 0.5339,
+      "step": 1299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.959446630656722e-05,
+      "loss": 0.3156,
+      "step": 1300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9593845000607513e-05,
+      "loss": 1.0786,
+      "step": 1301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.959322322560187e-05,
+      "loss": 1.5367,
+      "step": 1302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9592600981570282e-05,
+      "loss": 1.4002,
+      "step": 1303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9591978268532735e-05,
+      "loss": 0.5503,
+      "step": 1304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9591355086509253e-05,
+      "loss": 0.4239,
+      "step": 1305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.959073143551986e-05,
+      "loss": 0.614,
+      "step": 1306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9590107315584604e-05,
+      "loss": 1.6556,
+      "step": 1307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9589482726723544e-05,
+      "loss": 1.5075,
+      "step": 1308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958885766895675e-05,
+      "loss": 0.4412,
+      "step": 1309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958823214230432e-05,
+      "loss": 0.3583,
+      "step": 1310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9587606146786345e-05,
+      "loss": 0.3845,
+      "step": 1311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9586979682422955e-05,
+      "loss": 0.2505,
+      "step": 1312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9586352749234283e-05,
+      "loss": 0.5025,
+      "step": 1313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9585725347240478e-05,
+      "loss": 1.8943,
+      "step": 1314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9585097476461702e-05,
+      "loss": 0.4297,
+      "step": 1315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9584469136918136e-05,
+      "loss": 1.3125,
+      "step": 1316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958384032862998e-05,
+      "loss": 0.5469,
+      "step": 1317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9583211051617434e-05,
+      "loss": 0.4549,
+      "step": 1318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958258130590073e-05,
+      "loss": 0.6301,
+      "step": 1319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9581951091500106e-05,
+      "loss": 0.3949,
+      "step": 1320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9581320408435817e-05,
+      "loss": 0.2625,
+      "step": 1321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958068925672814e-05,
+      "loss": 0.6502,
+      "step": 1322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9580057636397345e-05,
+      "loss": 0.3581,
+      "step": 1323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.957942554746375e-05,
+      "loss": 0.8989,
+      "step": 1324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9578792989947653e-05,
+      "loss": 1.0273,
+      "step": 1325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9578159963869398e-05,
+      "loss": 1.0671,
+      "step": 1326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9577526469249324e-05,
+      "loss": 1.1308,
+      "step": 1327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.957689250610779e-05,
+      "loss": 0.8631,
+      "step": 1328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9576258074465175e-05,
+      "loss": 0.8221,
+      "step": 1329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.957562317434187e-05,
+      "loss": 0.8391,
+      "step": 1330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.957498780575828e-05,
+      "loss": 1.1928,
+      "step": 1331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9574351968734826e-05,
+      "loss": 0.8067,
+      "step": 1332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9573715663291943e-05,
+      "loss": 0.777,
+      "step": 1333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.957307888945008e-05,
+      "loss": 0.4489,
+      "step": 1334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.957244164722971e-05,
+      "loss": 0.3869,
+      "step": 1335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9571803936651305e-05,
+      "loss": 1.347,
+      "step": 1336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9571165757735366e-05,
+      "loss": 2.0703,
+      "step": 1337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9570527110502402e-05,
+      "loss": 0.5353,
+      "step": 1338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956988799497294e-05,
+      "loss": 0.4771,
+      "step": 1339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9569248411167522e-05,
+      "loss": 0.3535,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.95686083591067e-05,
+      "loss": 1.0112,
+      "step": 1341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9567967838811058e-05,
+      "loss": 0.2552,
+      "step": 1342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956732685030116e-05,
+      "loss": 1.0101,
+      "step": 1343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9566685393597632e-05,
+      "loss": 0.2933,
+      "step": 1344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9566043468721072e-05,
+      "loss": 1.026,
+      "step": 1345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956540107569212e-05,
+      "loss": 1.1288,
+      "step": 1346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9564758214531423e-05,
+      "loss": 0.4214,
+      "step": 1347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9564114885259634e-05,
+      "loss": 0.7784,
+      "step": 1348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956347108789744e-05,
+      "loss": 0.7225,
+      "step": 1349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956282682246553e-05,
+      "loss": 0.6609,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9562182088984606e-05,
+      "loss": 0.4548,
+      "step": 1351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9561536887475396e-05,
+      "loss": 0.1848,
+      "step": 1352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9560891217958635e-05,
+      "loss": 0.2226,
+      "step": 1353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9560245080455067e-05,
+      "loss": 0.2861,
+      "step": 1354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955959847498547e-05,
+      "loss": 0.508,
+      "step": 1355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955895140157062e-05,
+      "loss": 0.3553,
+      "step": 1356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9558303860231312e-05,
+      "loss": 0.6302,
+      "step": 1357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9557655850988368e-05,
+      "loss": 0.4018,
+      "step": 1358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9557007373862604e-05,
+      "loss": 1.7575,
+      "step": 1359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9556358428874867e-05,
+      "loss": 0.9643,
+      "step": 1360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955570901604601e-05,
+      "loss": 0.4927,
+      "step": 1361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9555059135396914e-05,
+      "loss": 0.5135,
+      "step": 1362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9554408786948453e-05,
+      "loss": 0.4946,
+      "step": 1363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9553757970721544e-05,
+      "loss": 0.934,
+      "step": 1364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9553106686737095e-05,
+      "loss": 0.7065,
+      "step": 1365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955245493501604e-05,
+      "loss": 0.6651,
+      "step": 1366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955180271557933e-05,
+      "loss": 0.4643,
+      "step": 1367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9551150028447916e-05,
+      "loss": 1.2664,
+      "step": 1368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9550496873642786e-05,
+      "loss": 0.6602,
+      "step": 1369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.954984325118493e-05,
+      "loss": 0.5881,
+      "step": 1370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9549189161095358e-05,
+      "loss": 0.4429,
+      "step": 1371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9548534603395084e-05,
+      "loss": 0.1429,
+      "step": 1372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9547879578105155e-05,
+      "loss": 1.1334,
+      "step": 1373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9547224085246615e-05,
+      "loss": 0.7026,
+      "step": 1374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9546568124840542e-05,
+      "loss": 0.7424,
+      "step": 1375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.954591169690801e-05,
+      "loss": 0.6846,
+      "step": 1376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9545254801470117e-05,
+      "loss": 1.6402,
+      "step": 1377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9544597438547978e-05,
+      "loss": 0.8709,
+      "step": 1378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9543939608162724e-05,
+      "loss": 0.4656,
+      "step": 1379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.954328131033549e-05,
+      "loss": 0.9656,
+      "step": 1380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9542622545087437e-05,
+      "loss": 1.1144,
+      "step": 1381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9541963312439742e-05,
+      "loss": 0.5506,
+      "step": 1382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.954130361241359e-05,
+      "loss": 0.4427,
+      "step": 1383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9540643445030176e-05,
+      "loss": 1.6415,
+      "step": 1384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.953998281031073e-05,
+      "loss": 1.499,
+      "step": 1385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9539321708276476e-05,
+      "loss": 1.0541,
+      "step": 1386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.953866013894867e-05,
+      "loss": 0.5822,
+      "step": 1387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9537998102348562e-05,
+      "loss": 0.4507,
+      "step": 1388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9537335598497443e-05,
+      "loss": 1.6394,
+      "step": 1389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9536672627416604e-05,
+      "loss": 1.2942,
+      "step": 1390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9536009189127344e-05,
+      "loss": 1.0843,
+      "step": 1391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9535345283650996e-05,
+      "loss": 1.3417,
+      "step": 1392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9534680911008887e-05,
+      "loss": 0.891,
+      "step": 1393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9534016071222385e-05,
+      "loss": 0.4058,
+      "step": 1394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9533350764312842e-05,
+      "loss": 0.4579,
+      "step": 1395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9532684990301654e-05,
+      "loss": 0.4145,
+      "step": 1396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9532018749210207e-05,
+      "loss": 0.3303,
+      "step": 1397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9531352041059924e-05,
+      "loss": 0.236,
+      "step": 1398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9530684865872225e-05,
+      "loss": 0.6704,
+      "step": 1399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.953001722366856e-05,
+      "loss": 0.7428,
+      "step": 1400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9529349114470387e-05,
+      "loss": 0.7566,
+      "step": 1401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9528680538299175e-05,
+      "loss": 0.3382,
+      "step": 1402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952801149517641e-05,
+      "loss": 0.6085,
+      "step": 1403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.95273419851236e-05,
+      "loss": 0.0858,
+      "step": 1404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952667200816226e-05,
+      "loss": 0.742,
+      "step": 1405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9526001564313923e-05,
+      "loss": 0.7703,
+      "step": 1406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9525330653600138e-05,
+      "loss": 0.3939,
+      "step": 1407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9524659276042474e-05,
+      "loss": 1.0466,
+      "step": 1408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9523987431662503e-05,
+      "loss": 0.2865,
+      "step": 1409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9523315120481814e-05,
+      "loss": 0.4623,
+      "step": 1410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9522642342522024e-05,
+      "loss": 0.318,
+      "step": 1411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952196909780475e-05,
+      "loss": 0.4212,
+      "step": 1412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952129538635163e-05,
+      "loss": 0.2426,
+      "step": 1413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952062120818432e-05,
+      "loss": 1.9959,
+      "step": 1414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9519946563324487e-05,
+      "loss": 0.3863,
+      "step": 1415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9519271451793812e-05,
+      "loss": 0.664,
+      "step": 1416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9518595873614e-05,
+      "loss": 0.71,
+      "step": 1417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9517919828806756e-05,
+      "loss": 0.0852,
+      "step": 1418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.951724331739381e-05,
+      "loss": 0.2233,
+      "step": 1419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9516566339396908e-05,
+      "loss": 0.4701,
+      "step": 1420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9515888894837803e-05,
+      "loss": 0.3869,
+      "step": 1421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9515210983738274e-05,
+      "loss": 0.6546,
+      "step": 1422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.951453260612011e-05,
+      "loss": 0.7228,
+      "step": 1423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9513853762005107e-05,
+      "loss": 0.8109,
+      "step": 1424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.951317445141509e-05,
+      "loss": 0.586,
+      "step": 1425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9512494674371884e-05,
+      "loss": 0.7766,
+      "step": 1426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9511814430897342e-05,
+      "loss": 0.8284,
+      "step": 1427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9511133721013323e-05,
+      "loss": 0.8798,
+      "step": 1428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9510452544741715e-05,
+      "loss": 1.4785,
+      "step": 1429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9509770902104405e-05,
+      "loss": 1.0848,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9509088793123297e-05,
+      "loss": 1.9017,
+      "step": 1431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950840621782032e-05,
+      "loss": 0.8331,
+      "step": 1432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9507723176217403e-05,
+      "loss": 1.4211,
+      "step": 1433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9507039668336518e-05,
+      "loss": 0.3238,
+      "step": 1434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950635569419961e-05,
+      "loss": 0.9774,
+      "step": 1435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9505671253828678e-05,
+      "loss": 1.1056,
+      "step": 1436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9504986347245708e-05,
+      "loss": 0.5001,
+      "step": 1437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950430097447272e-05,
+      "loss": 0.4883,
+      "step": 1438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9503615135531747e-05,
+      "loss": 1.1171,
+      "step": 1439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9502928830444822e-05,
+      "loss": 0.4091,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9502242059234005e-05,
+      "loss": 0.1814,
+      "step": 1441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9501554821921375e-05,
+      "loss": 0.2346,
+      "step": 1442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9500867118529014e-05,
+      "loss": 0.276,
+      "step": 1443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9500178949079024e-05,
+      "loss": 0.8222,
+      "step": 1444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9499490313593525e-05,
+      "loss": 0.2874,
+      "step": 1445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949880121209465e-05,
+      "loss": 0.5904,
+      "step": 1446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9498111644604546e-05,
+      "loss": 0.8968,
+      "step": 1447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949742161114538e-05,
+      "loss": 0.5841,
+      "step": 1448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949673111173932e-05,
+      "loss": 1.2787,
+      "step": 1449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9496040146408564e-05,
+      "loss": 0.8141,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949534871517532e-05,
+      "loss": 1.2113,
+      "step": 1451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9494656818061815e-05,
+      "loss": 0.5809,
+      "step": 1452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949396445509028e-05,
+      "loss": 0.2149,
+      "step": 1453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949327162628297e-05,
+      "loss": 0.5376,
+      "step": 1454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949257833166215e-05,
+      "loss": 0.1731,
+      "step": 1455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9491884571250106e-05,
+      "loss": 0.577,
+      "step": 1456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949119034506913e-05,
+      "loss": 0.5048,
+      "step": 1457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949049565314154e-05,
+      "loss": 0.4685,
+      "step": 1458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9489800495489666e-05,
+      "loss": 0.6282,
+      "step": 1459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9489104872135842e-05,
+      "loss": 0.2985,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9488408783102433e-05,
+      "loss": 0.4744,
+      "step": 1461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9487712228411804e-05,
+      "loss": 1.8337,
+      "step": 1462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9487015208086347e-05,
+      "loss": 0.2885,
+      "step": 1463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9486317722148465e-05,
+      "loss": 0.517,
+      "step": 1464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.948561977062057e-05,
+      "loss": 0.5809,
+      "step": 1465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.94849213535251e-05,
+      "loss": 0.6287,
+      "step": 1466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.94842224708845e-05,
+      "loss": 0.3302,
+      "step": 1467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9483523122721233e-05,
+      "loss": 0.5134,
+      "step": 1468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.948282330905777e-05,
+      "loss": 1.0321,
+      "step": 1469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9482123029916614e-05,
+      "loss": 1.1338,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.948142228532026e-05,
+      "loss": 1.4897,
+      "step": 1471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.948072107529124e-05,
+      "loss": 0.6301,
+      "step": 1472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.948001939985208e-05,
+      "loss": 1.2513,
+      "step": 1473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9479317259025342e-05,
+      "loss": 1.3361,
+      "step": 1474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9478614652833593e-05,
+      "loss": 1.7311,
+      "step": 1475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9477911581299407e-05,
+      "loss": 2.0104,
+      "step": 1476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9477208044445386e-05,
+      "loss": 0.423,
+      "step": 1477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.947650404229414e-05,
+      "loss": 1.5771,
+      "step": 1478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9475799574868295e-05,
+      "loss": 0.5982,
+      "step": 1479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9475094642190493e-05,
+      "loss": 1.231,
+      "step": 1480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.947438924428339e-05,
+      "loss": 1.4904,
+      "step": 1481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9473683381169663e-05,
+      "loss": 1.5732,
+      "step": 1482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.947297705287199e-05,
+      "loss": 0.7774,
+      "step": 1483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.947227025941308e-05,
+      "loss": 0.783,
+      "step": 1484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.947156300081564e-05,
+      "loss": 1.39,
+      "step": 1485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9470855277102414e-05,
+      "loss": 0.3615,
+      "step": 1486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9470147088296137e-05,
+      "loss": 0.3154,
+      "step": 1487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9469438434419573e-05,
+      "loss": 0.6486,
+      "step": 1488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.94687293154955e-05,
+      "loss": 0.1408,
+      "step": 1489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.946801973154671e-05,
+      "loss": 1.0136,
+      "step": 1490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9467309682596008e-05,
+      "loss": 0.2071,
+      "step": 1491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9466599168666212e-05,
+      "loss": 0.7256,
+      "step": 1492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9465888189780164e-05,
+      "loss": 0.4106,
+      "step": 1493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.946517674596071e-05,
+      "loss": 1.2395,
+      "step": 1494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9464464837230713e-05,
+      "loss": 0.7408,
+      "step": 1495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9463752463613058e-05,
+      "loss": 0.649,
+      "step": 1496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9463039625130643e-05,
+      "loss": 1.2721,
+      "step": 1497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9462326321806375e-05,
+      "loss": 1.0046,
+      "step": 1498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9461612553663186e-05,
+      "loss": 0.5415,
+      "step": 1499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9460898320724007e-05,
+      "loss": 0.781,
+      "step": 1500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9460183623011797e-05,
+      "loss": 1.5794,
+      "step": 1501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9459468460549527e-05,
+      "loss": 1.0115,
+      "step": 1502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.945875283336018e-05,
+      "loss": 0.5245,
+      "step": 1503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9458036741466763e-05,
+      "loss": 0.5409,
+      "step": 1504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9457320184892284e-05,
+      "loss": 1.0374,
+      "step": 1505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9456603163659776e-05,
+      "loss": 0.6134,
+      "step": 1506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9455885677792284e-05,
+      "loss": 0.4745,
+      "step": 1507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9455167727312865e-05,
+      "loss": 1.0242,
+      "step": 1508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9454449312244603e-05,
+      "loss": 1.5,
+      "step": 1509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9453730432610576e-05,
+      "loss": 0.4962,
+      "step": 1510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.94530110884339e-05,
+      "loss": 0.7874,
+      "step": 1511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9452291279737684e-05,
+      "loss": 1.5916,
+      "step": 1512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.945157100654507e-05,
+      "loss": 0.2186,
+      "step": 1513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9450850268879204e-05,
+      "loss": 1.4358,
+      "step": 1514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9450129066763253e-05,
+      "loss": 0.448,
+      "step": 1515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944940740022039e-05,
+      "loss": 0.1325,
+      "step": 1516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944868526927382e-05,
+      "loss": 0.5456,
+      "step": 1517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9447962673946748e-05,
+      "loss": 0.285,
+      "step": 1518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9447239614262393e-05,
+      "loss": 0.4869,
+      "step": 1519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9446516090243995e-05,
+      "loss": 1.3819,
+      "step": 1520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944579210191482e-05,
+      "loss": 0.4802,
+      "step": 1521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944506764929812e-05,
+      "loss": 0.6801,
+      "step": 1522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9444342732417196e-05,
+      "loss": 0.8678,
+      "step": 1523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944361735129533e-05,
+      "loss": 0.7294,
+      "step": 1524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9442891505955847e-05,
+      "loss": 0.6941,
+      "step": 1525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944216519642207e-05,
+      "loss": 0.6383,
+      "step": 1526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9441438422717345e-05,
+      "loss": 0.3685,
+      "step": 1527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9440711184865033e-05,
+      "loss": 1.4503,
+      "step": 1528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9439983482888506e-05,
+      "loss": 1.487,
+      "step": 1529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.943925531681115e-05,
+      "loss": 1.2955,
+      "step": 1530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9438526686656363e-05,
+      "loss": 1.158,
+      "step": 1531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9437797592447576e-05,
+      "loss": 1.6921,
+      "step": 1532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9437068034208214e-05,
+      "loss": 1.2319,
+      "step": 1533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9436338011961727e-05,
+      "loss": 0.5137,
+      "step": 1534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9435607525731574e-05,
+      "loss": 0.7303,
+      "step": 1535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.943487657554124e-05,
+      "loss": 1.5605,
+      "step": 1536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9434145161414213e-05,
+      "loss": 0.8638,
+      "step": 1537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9433413283374e-05,
+      "loss": 1.0028,
+      "step": 1538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9432680941444126e-05,
+      "loss": 1.0629,
+      "step": 1539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.943194813564813e-05,
+      "loss": 0.8601,
+      "step": 1540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.943121486600956e-05,
+      "loss": 0.9301,
+      "step": 1541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.943048113255199e-05,
+      "loss": 0.8666,
+      "step": 1542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9429746935298988e-05,
+      "loss": 1.5554,
+      "step": 1543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9429012274274166e-05,
+      "loss": 0.8031,
+      "step": 1544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9428277149501133e-05,
+      "loss": 1.4556,
+      "step": 1545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.942754156100351e-05,
+      "loss": 0.7503,
+      "step": 1546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.942680550880494e-05,
+      "loss": 1.6515,
+      "step": 1547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.942606899292909e-05,
+      "loss": 0.5658,
+      "step": 1548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9425332013399616e-05,
+      "loss": 0.8181,
+      "step": 1549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.942459457024021e-05,
+      "loss": 1.0877,
+      "step": 1550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9423856663474583e-05,
+      "loss": 0.2842,
+      "step": 1551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.942311829312644e-05,
+      "loss": 0.6529,
+      "step": 1552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9422379459219516e-05,
+      "loss": 0.2489,
+      "step": 1553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.942164016177756e-05,
+      "loss": 0.3872,
+      "step": 1554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9420900400824324e-05,
+      "loss": 0.6067,
+      "step": 1555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9420160176383594e-05,
+      "loss": 0.1641,
+      "step": 1556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9419419488479158e-05,
+      "loss": 1.0505,
+      "step": 1557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9418678337134817e-05,
+      "loss": 0.7491,
+      "step": 1558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.94179367223744e-05,
+      "loss": 1.9272,
+      "step": 1559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9417194644221734e-05,
+      "loss": 0.4221,
+      "step": 1560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9416452102700672e-05,
+      "loss": 0.5821,
+      "step": 1561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9415709097835082e-05,
+      "loss": 0.4481,
+      "step": 1562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941496562964884e-05,
+      "loss": 0.4708,
+      "step": 1563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9414221698165847e-05,
+      "loss": 0.7223,
+      "step": 1564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9413477303410006e-05,
+      "loss": 1.219,
+      "step": 1565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941273244540525e-05,
+      "loss": 1.8039,
+      "step": 1566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941198712417551e-05,
+      "loss": 1.6811,
+      "step": 1567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941124133974475e-05,
+      "loss": 0.5859,
+      "step": 1568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9410495092136926e-05,
+      "loss": 1.34,
+      "step": 1569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9409748381376038e-05,
+      "loss": 1.5484,
+      "step": 1570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.940900120748607e-05,
+      "loss": 0.2484,
+      "step": 1571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9408253570491048e-05,
+      "loss": 0.7271,
+      "step": 1572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9407505470414997e-05,
+      "loss": 0.673,
+      "step": 1573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9406756907281958e-05,
+      "loss": 0.9361,
+      "step": 1574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9406007881115996e-05,
+      "loss": 1.0384,
+      "step": 1575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9405258391941182e-05,
+      "loss": 1.0284,
+      "step": 1576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9404508439781605e-05,
+      "loss": 0.8318,
+      "step": 1577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.940375802466136e-05,
+      "loss": 0.5615,
+      "step": 1578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9403007146604576e-05,
+      "loss": 0.2984,
+      "step": 1579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9402255805635386e-05,
+      "loss": 0.4577,
+      "step": 1580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9401504001777933e-05,
+      "loss": 1.418,
+      "step": 1581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9400751735056385e-05,
+      "loss": 1.3564,
+      "step": 1582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9399999005494913e-05,
+      "loss": 0.7335,
+      "step": 1583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9399245813117715e-05,
+      "loss": 0.333,
+      "step": 1584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9398492157948996e-05,
+      "loss": 1.2528,
+      "step": 1585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9397738040012978e-05,
+      "loss": 0.7997,
+      "step": 1586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9396983459333902e-05,
+      "loss": 0.6056,
+      "step": 1587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9396228415936018e-05,
+      "loss": 0.6867,
+      "step": 1588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939547290984359e-05,
+      "loss": 0.983,
+      "step": 1589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939471694108091e-05,
+      "loss": 0.9966,
+      "step": 1590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9393960509672268e-05,
+      "loss": 1.294,
+      "step": 1591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9393203615641967e-05,
+      "loss": 0.6132,
+      "step": 1592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939244625901435e-05,
+      "loss": 0.4935,
+      "step": 1593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939168843981375e-05,
+      "loss": 1.1722,
+      "step": 1594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939093015806452e-05,
+      "loss": 0.3962,
+      "step": 1595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9390171413791037e-05,
+      "loss": 0.7934,
+      "step": 1596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9389412207017684e-05,
+      "loss": 0.4539,
+      "step": 1597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9388652537768866e-05,
+      "loss": 0.5406,
+      "step": 1598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938789240606899e-05,
+      "loss": 0.2573,
+      "step": 1599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9387131811942497e-05,
+      "loss": 0.5022,
+      "step": 1600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9386370755413823e-05,
+      "loss": 0.5483,
+      "step": 1601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9385609236507442e-05,
+      "loss": 1.1951,
+      "step": 1602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9384847255247814e-05,
+      "loss": 1.126,
+      "step": 1603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9384084811659437e-05,
+      "loss": 0.4615,
+      "step": 1604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938332190576682e-05,
+      "loss": 1.4432,
+      "step": 1605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938255853759447e-05,
+      "loss": 1.4126,
+      "step": 1606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938179470716693e-05,
+      "loss": 1.2814,
+      "step": 1607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9381030414508747e-05,
+      "loss": 0.0696,
+      "step": 1608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938026565964449e-05,
+      "loss": 0.3672,
+      "step": 1609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9379500442598732e-05,
+      "loss": 1.5553,
+      "step": 1610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9378734763396074e-05,
+      "loss": 0.6412,
+      "step": 1611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9377968622061116e-05,
+      "loss": 0.7421,
+      "step": 1612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9377202018618488e-05,
+      "loss": 0.2664,
+      "step": 1613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.937643495309283e-05,
+      "loss": 0.2337,
+      "step": 1614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.937566742550879e-05,
+      "loss": 0.7855,
+      "step": 1615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9374899435891042e-05,
+      "loss": 0.3322,
+      "step": 1616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9374130984264264e-05,
+      "loss": 1.065,
+      "step": 1617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9373362070653154e-05,
+      "loss": 0.5015,
+      "step": 1618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9372592695082428e-05,
+      "loss": 0.1694,
+      "step": 1619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.937182285757682e-05,
+      "loss": 0.168,
+      "step": 1620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.937105255816106e-05,
+      "loss": 0.6516,
+      "step": 1621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.937028179685991e-05,
+      "loss": 0.5592,
+      "step": 1622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9369510573698144e-05,
+      "loss": 0.7329,
+      "step": 1623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9368738888700547e-05,
+      "loss": 0.6786,
+      "step": 1624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9367966741891924e-05,
+      "loss": 0.8568,
+      "step": 1625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.936719413329709e-05,
+      "loss": 1.0818,
+      "step": 1626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9366421062940875e-05,
+      "loss": 0.748,
+      "step": 1627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9365647530848126e-05,
+      "loss": 0.9744,
+      "step": 1628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9364873537043707e-05,
+      "loss": 0.8625,
+      "step": 1629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9364099081552494e-05,
+      "loss": 0.7577,
+      "step": 1630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9363324164399374e-05,
+      "loss": 1.0914,
+      "step": 1631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9362548785609255e-05,
+      "loss": 0.9043,
+      "step": 1632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.936177294520706e-05,
+      "loss": 1.1885,
+      "step": 1633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9360996643217723e-05,
+      "loss": 0.5794,
+      "step": 1634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9360219879666192e-05,
+      "loss": 0.9408,
+      "step": 1635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9359442654577433e-05,
+      "loss": 0.5961,
+      "step": 1636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9358664967976427e-05,
+      "loss": 1.3801,
+      "step": 1637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.935788681988817e-05,
+      "loss": 0.5273,
+      "step": 1638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9357108210337674e-05,
+      "loss": 0.6875,
+      "step": 1639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9356329139349952e-05,
+      "loss": 0.2293,
+      "step": 1640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9355549606950055e-05,
+      "loss": 0.6394,
+      "step": 1641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9354769613163038e-05,
+      "loss": 0.5918,
+      "step": 1642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9353989158013963e-05,
+      "loss": 0.4535,
+      "step": 1643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9353208241527915e-05,
+      "loss": 0.8816,
+      "step": 1644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9352426863729987e-05,
+      "loss": 0.5822,
+      "step": 1645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.935164502464531e-05,
+      "loss": 0.734,
+      "step": 1646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9350862724298998e-05,
+      "loss": 0.5295,
+      "step": 1647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9350079962716196e-05,
+      "loss": 1.0419,
+      "step": 1648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9349296739922066e-05,
+      "loss": 0.6482,
+      "step": 1649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9348513055941777e-05,
+      "loss": 0.2986,
+      "step": 1650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.934772891080052e-05,
+      "loss": 0.5085,
+      "step": 1651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9346944304523492e-05,
+      "loss": 0.1467,
+      "step": 1652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9346159237135918e-05,
+      "loss": 0.4431,
+      "step": 1653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9345373708663022e-05,
+      "loss": 0.3166,
+      "step": 1654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9344587719130055e-05,
+      "loss": 0.367,
+      "step": 1655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.934380126856228e-05,
+      "loss": 1.2543,
+      "step": 1656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9343014356984968e-05,
+      "loss": 1.597,
+      "step": 1657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9342226984423415e-05,
+      "loss": 0.3692,
+      "step": 1658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.934143915090293e-05,
+      "loss": 0.6855,
+      "step": 1659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9340650856448828e-05,
+      "loss": 0.555,
+      "step": 1660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9339862101086453e-05,
+      "loss": 1.8328,
+      "step": 1661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9339072884841142e-05,
+      "loss": 0.3572,
+      "step": 1662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9338283207738273e-05,
+      "loss": 0.4574,
+      "step": 1663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9337493069803215e-05,
+      "loss": 0.6623,
+      "step": 1664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9336702471061373e-05,
+      "loss": 0.2116,
+      "step": 1665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9335911411538155e-05,
+      "loss": 0.2837,
+      "step": 1666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9335119891258982e-05,
+      "loss": 0.449,
+      "step": 1667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9334327910249295e-05,
+      "loss": 1.5392,
+      "step": 1668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9333535468534545e-05,
+      "loss": 1.0754,
+      "step": 1669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.933274256614021e-05,
+      "loss": 0.2716,
+      "step": 1670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9331949203091765e-05,
+      "loss": 0.236,
+      "step": 1671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9331155379414712e-05,
+      "loss": 0.8426,
+      "step": 1672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9330361095134566e-05,
+      "loss": 0.5846,
+      "step": 1673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932956635027685e-05,
+      "loss": 1.7235,
+      "step": 1674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9328771144867117e-05,
+      "loss": 1.7019,
+      "step": 1675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9327975478930913e-05,
+      "loss": 0.5121,
+      "step": 1676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932717935249382e-05,
+      "loss": 0.2904,
+      "step": 1677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932638276558142e-05,
+      "loss": 1.1772,
+      "step": 1678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9325585718219322e-05,
+      "loss": 0.7032,
+      "step": 1679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932478821043313e-05,
+      "loss": 0.5892,
+      "step": 1680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932399024224849e-05,
+      "loss": 0.737,
+      "step": 1681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9323191813691043e-05,
+      "loss": 0.7938,
+      "step": 1682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932239292478645e-05,
+      "loss": 0.8363,
+      "step": 1683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9321593575560387e-05,
+      "loss": 0.6832,
+      "step": 1684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9320793766038545e-05,
+      "loss": 0.5446,
+      "step": 1685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.931999349624663e-05,
+      "loss": 0.7412,
+      "step": 1686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9319192766210365e-05,
+      "loss": 0.777,
+      "step": 1687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9318391575955484e-05,
+      "loss": 0.2339,
+      "step": 1688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9317589925507733e-05,
+      "loss": 1.5953,
+      "step": 1689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9316787814892883e-05,
+      "loss": 0.9448,
+      "step": 1690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9315985244136712e-05,
+      "loss": 0.6662,
+      "step": 1691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9315182213265014e-05,
+      "loss": 1.2278,
+      "step": 1692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.93143787223036e-05,
+      "loss": 0.4288,
+      "step": 1693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9313574771278297e-05,
+      "loss": 0.1773,
+      "step": 1694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9312770360214933e-05,
+      "loss": 1.1328,
+      "step": 1695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.931196548913937e-05,
+      "loss": 0.8742,
+      "step": 1696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.931116015807748e-05,
+      "loss": 0.6861,
+      "step": 1697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9310354367055135e-05,
+      "loss": 0.4555,
+      "step": 1698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930954811609824e-05,
+      "loss": 0.5972,
+      "step": 1699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930874140523271e-05,
+      "loss": 0.4699,
+      "step": 1700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9307934234484474e-05,
+      "loss": 0.3754,
+      "step": 1701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930712660387947e-05,
+      "loss": 0.2352,
+      "step": 1702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930631851344365e-05,
+      "loss": 0.1822,
+      "step": 1703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9305509963202998e-05,
+      "loss": 0.1772,
+      "step": 1704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9304700953183486e-05,
+      "loss": 0.2503,
+      "step": 1705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9303891483411133e-05,
+      "loss": 1.8151,
+      "step": 1706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930308155391194e-05,
+      "loss": 0.6405,
+      "step": 1707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930227116471195e-05,
+      "loss": 0.3314,
+      "step": 1708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9301460315837205e-05,
+      "loss": 0.4817,
+      "step": 1709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9300649007313763e-05,
+      "loss": 0.2518,
+      "step": 1710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9299837239167696e-05,
+      "loss": 0.4692,
+      "step": 1711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9299025011425103e-05,
+      "loss": 0.9254,
+      "step": 1712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9298212324112085e-05,
+      "loss": 1.3958,
+      "step": 1713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9297399177254762e-05,
+      "loss": 0.1034,
+      "step": 1714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9296585570879263e-05,
+      "loss": 0.5698,
+      "step": 1715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.929577150501175e-05,
+      "loss": 1.3478,
+      "step": 1716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9294956979678374e-05,
+      "loss": 0.3764,
+      "step": 1717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9294141994905323e-05,
+      "loss": 0.3124,
+      "step": 1718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9293326550718783e-05,
+      "loss": 0.4468,
+      "step": 1719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.929251064714497e-05,
+      "loss": 0.5363,
+      "step": 1720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.92916942842101e-05,
+      "loss": 1.6153,
+      "step": 1721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.929087746194042e-05,
+      "loss": 0.4987,
+      "step": 1722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9290060180362174e-05,
+      "loss": 1.251,
+      "step": 1723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9289242439501634e-05,
+      "loss": 1.497,
+      "step": 1724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.928842423938508e-05,
+      "loss": 1.5454,
+      "step": 1725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.928760558003881e-05,
+      "loss": 0.5282,
+      "step": 1726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.928678646148913e-05,
+      "loss": 1.9497,
+      "step": 1727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.928596688376238e-05,
+      "loss": 0.3284,
+      "step": 1728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9285146846884887e-05,
+      "loss": 0.5948,
+      "step": 1729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9284326350883017e-05,
+      "loss": 1.0452,
+      "step": 1730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9283505395783136e-05,
+      "loss": 1.1671,
+      "step": 1731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.928268398161163e-05,
+      "loss": 0.5122,
+      "step": 1732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9281862108394895e-05,
+      "loss": 0.6153,
+      "step": 1733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9281039776159352e-05,
+      "loss": 0.5686,
+      "step": 1734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9280216984931434e-05,
+      "loss": 0.9025,
+      "step": 1735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.927939373473758e-05,
+      "loss": 1.0239,
+      "step": 1736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9278570025604247e-05,
+      "loss": 1.0256,
+      "step": 1737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.927774585755791e-05,
+      "loss": 1.0957,
+      "step": 1738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.927692123062506e-05,
+      "loss": 0.7681,
+      "step": 1739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9276096144832203e-05,
+      "loss": 1.0124,
+      "step": 1740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9275270600205854e-05,
+      "loss": 0.9517,
+      "step": 1741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9274444596772546e-05,
+      "loss": 0.646,
+      "step": 1742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9273618134558827e-05,
+      "loss": 0.2712,
+      "step": 1743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9272791213591257e-05,
+      "loss": 0.3199,
+      "step": 1744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.927196383389642e-05,
+      "loss": 0.2058,
+      "step": 1745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9271135995500903e-05,
+      "loss": 0.6011,
+      "step": 1746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.927030769843131e-05,
+      "loss": 0.9844,
+      "step": 1747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9269478942714268e-05,
+      "loss": 0.1643,
+      "step": 1748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9268649728376415e-05,
+      "loss": 0.6818,
+      "step": 1749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.926782005544439e-05,
+      "loss": 0.7136,
+      "step": 1750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.926698992394487e-05,
+      "loss": 0.2863,
+      "step": 1751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9266159333904537e-05,
+      "loss": 0.6687,
+      "step": 1752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9265328285350078e-05,
+      "loss": 0.5143,
+      "step": 1753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9264496778308204e-05,
+      "loss": 1.4403,
+      "step": 1754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9263664812805644e-05,
+      "loss": 1.426,
+      "step": 1755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9262832388869134e-05,
+      "loss": 0.684,
+      "step": 1756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9261999506525434e-05,
+      "loss": 0.3343,
+      "step": 1757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.92611661658013e-05,
+      "loss": 0.5481,
+      "step": 1758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9260332366723534e-05,
+      "loss": 0.3854,
+      "step": 1759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9259498109318914e-05,
+      "loss": 1.5474,
+      "step": 1760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9258663393614266e-05,
+      "loss": 1.2573,
+      "step": 1761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9257828219636415e-05,
+      "loss": 1.3706,
+      "step": 1762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9256992587412205e-05,
+      "loss": 0.4246,
+      "step": 1763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.925615649696849e-05,
+      "loss": 0.8458,
+      "step": 1764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9255319948332145e-05,
+      "loss": 1.6312,
+      "step": 1765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.925448294153005e-05,
+      "loss": 0.779,
+      "step": 1766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9253645476589116e-05,
+      "loss": 0.5867,
+      "step": 1767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9252807553536253e-05,
+      "loss": 0.9929,
+      "step": 1768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9251969172398394e-05,
+      "loss": 0.491,
+      "step": 1769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9251130333202484e-05,
+      "loss": 0.5326,
+      "step": 1770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9250291035975484e-05,
+      "loss": 1.2859,
+      "step": 1771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9249451280744365e-05,
+      "loss": 0.5138,
+      "step": 1772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9248611067536122e-05,
+      "loss": 1.9088,
+      "step": 1773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9247770396377754e-05,
+      "loss": 0.5627,
+      "step": 1774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.924692926729629e-05,
+      "loss": 0.7429,
+      "step": 1775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.924608768031875e-05,
+      "loss": 0.5132,
+      "step": 1776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9245245635472193e-05,
+      "loss": 0.5463,
+      "step": 1777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.924440313278368e-05,
+      "loss": 0.3796,
+      "step": 1778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.924356017228029e-05,
+      "loss": 0.7557,
+      "step": 1779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9242716753989113e-05,
+      "loss": 1.3866,
+      "step": 1780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9241872877937255e-05,
+      "loss": 0.5535,
+      "step": 1781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.924102854415184e-05,
+      "loss": 0.3443,
+      "step": 1782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9240183752660008e-05,
+      "loss": 0.2074,
+      "step": 1783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9239338503488907e-05,
+      "loss": 0.4089,
+      "step": 1784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9238492796665705e-05,
+      "loss": 1.0684,
+      "step": 1785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9237646632217584e-05,
+      "loss": 0.3414,
+      "step": 1786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9236800010171735e-05,
+      "loss": 0.7967,
+      "step": 1787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.923595293055537e-05,
+      "loss": 0.1706,
+      "step": 1788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9235105393395722e-05,
+      "loss": 0.9964,
+      "step": 1789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9234257398720016e-05,
+      "loss": 0.531,
+      "step": 1790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9233408946555523e-05,
+      "loss": 0.6233,
+      "step": 1791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.92325600369295e-05,
+      "loss": 0.9806,
+      "step": 1792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9231710669869238e-05,
+      "loss": 0.4903,
+      "step": 1793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9230860845402028e-05,
+      "loss": 0.8192,
+      "step": 1794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.923001056355519e-05,
+      "loss": 0.6661,
+      "step": 1795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.922915982435605e-05,
+      "loss": 1.1931,
+      "step": 1796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9228308627831955e-05,
+      "loss": 0.3631,
+      "step": 1797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9227456974010254e-05,
+      "loss": 0.6072,
+      "step": 1798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9226604862918325e-05,
+      "loss": 0.6284,
+      "step": 1799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9225752294583554e-05,
+      "loss": 0.3548,
+      "step": 1800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9224899269033337e-05,
+      "loss": 1.3703,
+      "step": 1801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.92240457862951e-05,
+      "loss": 0.4626,
+      "step": 1802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.922319184639627e-05,
+      "loss": 1.3504,
+      "step": 1803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.922233744936429e-05,
+      "loss": 1.5652,
+      "step": 1804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9221482595226622e-05,
+      "loss": 0.9095,
+      "step": 1805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.922062728401074e-05,
+      "loss": 0.1869,
+      "step": 1806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9219771515744137e-05,
+      "loss": 0.6021,
+      "step": 1807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9218915290454313e-05,
+      "loss": 0.7944,
+      "step": 1808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9218058608168788e-05,
+      "loss": 0.3744,
+      "step": 1809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9217201468915096e-05,
+      "loss": 1.2053,
+      "step": 1810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.921634387272079e-05,
+      "loss": 0.7063,
+      "step": 1811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9215485819613428e-05,
+      "loss": 2.2214,
+      "step": 1812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.921462730962059e-05,
+      "loss": 0.7703,
+      "step": 1813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.921376834276987e-05,
+      "loss": 0.4719,
+      "step": 1814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.921290891908887e-05,
+      "loss": 0.6161,
+      "step": 1815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9212049038605212e-05,
+      "loss": 0.4624,
+      "step": 1816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9211188701346536e-05,
+      "loss": 1.1212,
+      "step": 1817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9210327907340495e-05,
+      "loss": 1.5718,
+      "step": 1818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9209466656614753e-05,
+      "loss": 1.1579,
+      "step": 1819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9208604949196987e-05,
+      "loss": 1.9026,
+      "step": 1820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9207742785114896e-05,
+      "loss": 0.8902,
+      "step": 1821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.920688016439619e-05,
+      "loss": 0.4502,
+      "step": 1822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9206017087068595e-05,
+      "loss": 0.6382,
+      "step": 1823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.920515355315984e-05,
+      "loss": 0.5737,
+      "step": 1824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.92042895626977e-05,
+      "loss": 0.9004,
+      "step": 1825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9203425115709922e-05,
+      "loss": 1.0132,
+      "step": 1826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.92025602122243e-05,
+      "loss": 1.5044,
+      "step": 1827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.920169485226863e-05,
+      "loss": 0.411,
+      "step": 1828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9200829035870723e-05,
+      "loss": 0.6173,
+      "step": 1829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9199962763058414e-05,
+      "loss": 0.5676,
+      "step": 1830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919909603385953e-05,
+      "loss": 1.1609,
+      "step": 1831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9198228848301943e-05,
+      "loss": 2.1159,
+      "step": 1832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9197361206413516e-05,
+      "loss": 0.4519,
+      "step": 1833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919649310822214e-05,
+      "loss": 0.3423,
+      "step": 1834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9195624553755705e-05,
+      "loss": 0.5671,
+      "step": 1835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919475554304214e-05,
+      "loss": 0.6237,
+      "step": 1836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919388607610937e-05,
+      "loss": 1.0109,
+      "step": 1837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919301615298533e-05,
+      "loss": 0.8931,
+      "step": 1838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9192145773697994e-05,
+      "loss": 0.8515,
+      "step": 1839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919127493827533e-05,
+      "loss": 0.6641,
+      "step": 1840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9190403646745328e-05,
+      "loss": 0.5684,
+      "step": 1841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.918953189913599e-05,
+      "loss": 0.2118,
+      "step": 1842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.918865969547533e-05,
+      "loss": 0.8208,
+      "step": 1843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.918778703579139e-05,
+      "loss": 0.5778,
+      "step": 1844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9186913920112204e-05,
+      "loss": 0.544,
+      "step": 1845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9186040348465845e-05,
+      "loss": 0.5991,
+      "step": 1846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9185166320880394e-05,
+      "loss": 0.7007,
+      "step": 1847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9184291837383924e-05,
+      "loss": 0.6145,
+      "step": 1848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9183416898004557e-05,
+      "loss": 0.9464,
+      "step": 1849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.91825415027704e-05,
+      "loss": 0.5545,
+      "step": 1850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9181665651709602e-05,
+      "loss": 0.3987,
+      "step": 1851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9180789344850308e-05,
+      "loss": 1.1552,
+      "step": 1852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9179912582220678e-05,
+      "loss": 0.3892,
+      "step": 1853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9179035363848897e-05,
+      "loss": 0.6697,
+      "step": 1854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9178157689763154e-05,
+      "loss": 1.0365,
+      "step": 1855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.917727955999166e-05,
+      "loss": 0.4081,
+      "step": 1856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.917640097456264e-05,
+      "loss": 0.3601,
+      "step": 1857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9175521933504328e-05,
+      "loss": 1.534,
+      "step": 1858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.917464243684498e-05,
+      "loss": 0.3313,
+      "step": 1859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9173762484612856e-05,
+      "loss": 0.2117,
+      "step": 1860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9172882076836246e-05,
+      "loss": 1.6052,
+      "step": 1861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9172001213543444e-05,
+      "loss": 0.8216,
+      "step": 1862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9171119894762757e-05,
+      "loss": 0.1535,
+      "step": 1863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9170238120522513e-05,
+      "loss": 1.4351,
+      "step": 1864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9169355890851055e-05,
+      "loss": 1.5612,
+      "step": 1865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9168473205776734e-05,
+      "loss": 0.3468,
+      "step": 1866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.916759006532792e-05,
+      "loss": 0.1975,
+      "step": 1867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9166706469532998e-05,
+      "loss": 0.3901,
+      "step": 1868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.916582241842037e-05,
+      "loss": 0.4705,
+      "step": 1869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9164937912018444e-05,
+      "loss": 1.5786,
+      "step": 1870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9164052950355645e-05,
+      "loss": 0.6407,
+      "step": 1871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9163167533460427e-05,
+      "loss": 1.7578,
+      "step": 1872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.916228166136124e-05,
+      "loss": 1.0263,
+      "step": 1873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9161395334086553e-05,
+      "loss": 0.3395,
+      "step": 1874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9160508551664862e-05,
+      "loss": 1.5726,
+      "step": 1875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9159621314124664e-05,
+      "loss": 0.6785,
+      "step": 1876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9158733621494466e-05,
+      "loss": 1.3808,
+      "step": 1877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9157845473802808e-05,
+      "loss": 0.4406,
+      "step": 1878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.915695687107824e-05,
+      "loss": 1.8838,
+      "step": 1879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.915606781334931e-05,
+      "loss": 0.3773,
+      "step": 1880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9155178300644595e-05,
+      "loss": 0.4027,
+      "step": 1881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9154288332992687e-05,
+      "loss": 0.4383,
+      "step": 1882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9153397910422185e-05,
+      "loss": 0.9208,
+      "step": 1883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9152507032961716e-05,
+      "loss": 0.4548,
+      "step": 1884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9151615700639906e-05,
+      "loss": 1.1467,
+      "step": 1885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9150723913485404e-05,
+      "loss": 1.2333,
+      "step": 1886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.914983167152687e-05,
+      "loss": 0.4396,
+      "step": 1887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9148938974792985e-05,
+      "loss": 0.3448,
+      "step": 1888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9148045823312436e-05,
+      "loss": 0.415,
+      "step": 1889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.914715221711393e-05,
+      "loss": 0.8041,
+      "step": 1890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9146258156226192e-05,
+      "loss": 1.188,
+      "step": 1891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9145363640677946e-05,
+      "loss": 1.2171,
+      "step": 1892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9144468670497957e-05,
+      "loss": 0.0796,
+      "step": 1893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9143573245714974e-05,
+      "loss": 0.3432,
+      "step": 1894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9142677366357788e-05,
+      "loss": 0.2228,
+      "step": 1895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9141781032455186e-05,
+      "loss": 0.4979,
+      "step": 1896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.914088424403598e-05,
+      "loss": 0.2647,
+      "step": 1897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9139987001128986e-05,
+      "loss": 0.565,
+      "step": 1898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9139089303763047e-05,
+      "loss": 0.6907,
+      "step": 1899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9138191151967018e-05,
+      "loss": 0.4492,
+      "step": 1900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9137292545769757e-05,
+      "loss": 0.5768,
+      "step": 1901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9136393485200152e-05,
+      "loss": 0.3037,
+      "step": 1902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9135493970287097e-05,
+      "loss": 1.3091,
+      "step": 1903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.91345940010595e-05,
+      "loss": 0.4736,
+      "step": 1904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9133693577546287e-05,
+      "loss": 0.7031,
+      "step": 1905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9132792699776403e-05,
+      "loss": 0.1295,
+      "step": 1906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9131891367778792e-05,
+      "loss": 0.4646,
+      "step": 1907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9130989581582433e-05,
+      "loss": 1.9656,
+      "step": 1908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.91300873412163e-05,
+      "loss": 1.5133,
+      "step": 1909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9129184646709403e-05,
+      "loss": 1.0556,
+      "step": 1910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9128281498090743e-05,
+      "loss": 0.2497,
+      "step": 1911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9127377895389352e-05,
+      "loss": 1.1375,
+      "step": 1912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9126473838634273e-05,
+      "loss": 0.3241,
+      "step": 1913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9125569327854558e-05,
+      "loss": 0.6035,
+      "step": 1914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9124664363079284e-05,
+      "loss": 0.866,
+      "step": 1915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9123758944337533e-05,
+      "loss": 0.7824,
+      "step": 1916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9122853071658405e-05,
+      "loss": 0.456,
+      "step": 1917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.912194674507101e-05,
+      "loss": 0.7059,
+      "step": 1918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.912103996460449e-05,
+      "loss": 0.9105,
+      "step": 1919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.912013273028798e-05,
+      "loss": 0.6146,
+      "step": 1920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9119225042150632e-05,
+      "loss": 0.3688,
+      "step": 1921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9118316900221635e-05,
+      "loss": 0.5622,
+      "step": 1922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9117408304530166e-05,
+      "loss": 0.557,
+      "step": 1923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.911649925510543e-05,
+      "loss": 1.1236,
+      "step": 1924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9115589751976646e-05,
+      "loss": 0.6198,
+      "step": 1925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.911467979517304e-05,
+      "loss": 1.3883,
+      "step": 1926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9113769384723863e-05,
+      "loss": 0.3726,
+      "step": 1927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9112858520658375e-05,
+      "loss": 1.5205,
+      "step": 1928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.911194720300585e-05,
+      "loss": 0.686,
+      "step": 1929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9111035431795572e-05,
+      "loss": 0.2378,
+      "step": 1930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9110123207056856e-05,
+      "loss": 0.6982,
+      "step": 1931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9109210528819016e-05,
+      "loss": 1.1787,
+      "step": 1932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9108297397111383e-05,
+      "loss": 0.8145,
+      "step": 1933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.910738381196331e-05,
+      "loss": 1.4939,
+      "step": 1934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9106469773404153e-05,
+      "loss": 0.5969,
+      "step": 1935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9105555281463297e-05,
+      "loss": 0.418,
+      "step": 1936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9104640336170126e-05,
+      "loss": 0.6245,
+      "step": 1937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9103724937554053e-05,
+      "loss": 0.7713,
+      "step": 1938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9102809085644494e-05,
+      "loss": 1.5436,
+      "step": 1939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9101892780470887e-05,
+      "loss": 1.271,
+      "step": 1940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9100976022062683e-05,
+      "loss": 0.9886,
+      "step": 1941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9100058810449345e-05,
+      "loss": 0.6296,
+      "step": 1942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9099141145660352e-05,
+      "loss": 0.4864,
+      "step": 1943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.90982230277252e-05,
+      "loss": 0.5941,
+      "step": 1944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9097304456673393e-05,
+      "loss": 1.8866,
+      "step": 1945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9096385432534454e-05,
+      "loss": 0.7623,
+      "step": 1946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9095465955337928e-05,
+      "loss": 0.6018,
+      "step": 1947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9094546025113358e-05,
+      "loss": 1.3619,
+      "step": 1948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9093625641890316e-05,
+      "loss": 0.5357,
+      "step": 1949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9092704805698378e-05,
+      "loss": 0.3563,
+      "step": 1950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9091783516567148e-05,
+      "loss": 0.4549,
+      "step": 1951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.909086177452623e-05,
+      "loss": 0.1239,
+      "step": 1952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9089939579605246e-05,
+      "loss": 0.442,
+      "step": 1953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9089016931833843e-05,
+      "loss": 0.4379,
+      "step": 1954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9088093831241672e-05,
+      "loss": 0.3341,
+      "step": 1955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9087170277858394e-05,
+      "loss": 1.3353,
+      "step": 1956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9086246271713704e-05,
+      "loss": 0.3374,
+      "step": 1957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9085321812837294e-05,
+      "loss": 0.3451,
+      "step": 1958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9084396901258873e-05,
+      "loss": 0.9781,
+      "step": 1959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9083471537008176e-05,
+      "loss": 0.7234,
+      "step": 1960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9082545720114934e-05,
+      "loss": 0.2631,
+      "step": 1961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.908161945060891e-05,
+      "loss": 0.4156,
+      "step": 1962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9080692728519874e-05,
+      "loss": 0.5625,
+      "step": 1963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9079765553877605e-05,
+      "loss": 1.4645,
+      "step": 1964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.907883792671191e-05,
+      "loss": 1.4629,
+      "step": 1965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9077909847052598e-05,
+      "loss": 0.5691,
+      "step": 1966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9076981314929497e-05,
+      "loss": 1.1147,
+      "step": 1967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.907605233037245e-05,
+      "loss": 0.1594,
+      "step": 1968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9075122893411324e-05,
+      "loss": 0.2297,
+      "step": 1969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9074193004075978e-05,
+      "loss": 0.2034,
+      "step": 1970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9073262662396303e-05,
+      "loss": 0.5511,
+      "step": 1971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9072331868402204e-05,
+      "loss": 1.0268,
+      "step": 1972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9071400622123593e-05,
+      "loss": 0.1455,
+      "step": 1973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.90704689235904e-05,
+      "loss": 0.8885,
+      "step": 1974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9069536772832568e-05,
+      "loss": 0.7088,
+      "step": 1975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9068604169880063e-05,
+      "loss": 1.3444,
+      "step": 1976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.906767111476285e-05,
+      "loss": 0.3551,
+      "step": 1977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9066737607510934e-05,
+      "loss": 1.587,
+      "step": 1978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9065803648154294e-05,
+      "loss": 0.8735,
+      "step": 1979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9064869236722966e-05,
+      "loss": 1.2105,
+      "step": 1980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9063934373246977e-05,
+      "loss": 0.3904,
+      "step": 1981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9062999057756368e-05,
+      "loss": 0.9611,
+      "step": 1982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9062063290281206e-05,
+      "loss": 1.2595,
+      "step": 1983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9061127070851564e-05,
+      "loss": 0.8274,
+      "step": 1984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9060190399497532e-05,
+      "loss": 1.4447,
+      "step": 1985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9059253276249218e-05,
+      "loss": 0.8996,
+      "step": 1986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9058315701136737e-05,
+      "loss": 0.7922,
+      "step": 1987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9057377674190224e-05,
+      "loss": 1.156,
+      "step": 1988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9056439195439827e-05,
+      "loss": 1.1203,
+      "step": 1989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.905550026491571e-05,
+      "loss": 1.2253,
+      "step": 1990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.905456088264805e-05,
+      "loss": 1.2263,
+      "step": 1991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9053621048667033e-05,
+      "loss": 0.2415,
+      "step": 1992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9052680763002873e-05,
+      "loss": 0.3483,
+      "step": 1993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9051740025685787e-05,
+      "loss": 0.2077,
+      "step": 1994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9050798836746018e-05,
+      "loss": 0.9111,
+      "step": 1995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9049857196213803e-05,
+      "loss": 0.7413,
+      "step": 1996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9048915104119414e-05,
+      "loss": 1.0647,
+      "step": 1997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.904797256049313e-05,
+      "loss": 0.2548,
+      "step": 1998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9047029565365243e-05,
+      "loss": 0.5014,
+      "step": 1999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9046086118766055e-05,
+      "loss": 0.4747,
+      "step": 2000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9045142220725896e-05,
+      "loss": 0.1585,
+      "step": 2001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9044197871275105e-05,
+      "loss": 1.5022,
+      "step": 2002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9043253070444022e-05,
+      "loss": 0.6597,
+      "step": 2003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9042307818263027e-05,
+      "loss": 0.3555,
+      "step": 2004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9041362114762487e-05,
+      "loss": 0.583,
+      "step": 2005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9040415959972805e-05,
+      "loss": 0.3742,
+      "step": 2006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.903946935392439e-05,
+      "loss": 0.2418,
+      "step": 2007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9038522296647665e-05,
+      "loss": 0.4749,
+      "step": 2008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9037574788173065e-05,
+      "loss": 0.352,
+      "step": 2009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9036626828531048e-05,
+      "loss": 0.4164,
+      "step": 2010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9035678417752076e-05,
+      "loss": 0.1052,
+      "step": 2011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.903472955586664e-05,
+      "loss": 1.0579,
+      "step": 2012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9033780242905226e-05,
+      "loss": 0.368,
+      "step": 2013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.903283047889835e-05,
+      "loss": 0.6441,
+      "step": 2014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9031880263876538e-05,
+      "loss": 1.7694,
+      "step": 2015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9030929597870328e-05,
+      "loss": 0.7942,
+      "step": 2016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.902997848091028e-05,
+      "loss": 0.4777,
+      "step": 2017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9029026913026948e-05,
+      "loss": 0.2921,
+      "step": 2018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9028074894250927e-05,
+      "loss": 0.3058,
+      "step": 2019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.902712242461282e-05,
+      "loss": 0.8928,
+      "step": 2020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.902616950414323e-05,
+      "loss": 0.9296,
+      "step": 2021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9025216132872786e-05,
+      "loss": 0.6007,
+      "step": 2022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9024262310832132e-05,
+      "loss": 0.3736,
+      "step": 2023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9023308038051914e-05,
+      "loss": 1.3139,
+      "step": 2024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.902235331456282e-05,
+      "loss": 1.3935,
+      "step": 2025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.902139814039552e-05,
+      "loss": 0.8092,
+      "step": 2026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9020442515580714e-05,
+      "loss": 0.7716,
+      "step": 2027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9019486440149125e-05,
+      "loss": 1.3424,
+      "step": 2028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9018529914131475e-05,
+      "loss": 0.3329,
+      "step": 2029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9017572937558504e-05,
+      "loss": 0.6521,
+      "step": 2030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9016615510460974e-05,
+      "loss": 0.9821,
+      "step": 2031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9015657632869662e-05,
+      "loss": 1.2733,
+      "step": 2032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9014699304815346e-05,
+      "loss": 1.0415,
+      "step": 2033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9013740526328824e-05,
+      "loss": 1.0355,
+      "step": 2034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9012781297440918e-05,
+      "loss": 0.9341,
+      "step": 2035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9011821618182453e-05,
+      "loss": 0.3435,
+      "step": 2036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9010861488584274e-05,
+      "loss": 0.8049,
+      "step": 2037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9009900908677245e-05,
+      "loss": 0.8076,
+      "step": 2038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9008939878492235e-05,
+      "loss": 0.749,
+      "step": 2039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.900797839806013e-05,
+      "loss": 0.7836,
+      "step": 2040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.900701646741183e-05,
+      "loss": 0.4659,
+      "step": 2041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.900605408657826e-05,
+      "loss": 1.3611,
+      "step": 2042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9005091255590342e-05,
+      "loss": 1.2872,
+      "step": 2043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.900412797447903e-05,
+      "loss": 0.6276,
+      "step": 2044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.900316424327527e-05,
+      "loss": 1.2368,
+      "step": 2045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9002200062010054e-05,
+      "loss": 1.6945,
+      "step": 2046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9001235430714355e-05,
+      "loss": 0.7055,
+      "step": 2047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.900027034941919e-05,
+      "loss": 0.9503,
+      "step": 2048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8999304818155564e-05,
+      "loss": 0.9521,
+      "step": 2049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.899833883695452e-05,
+      "loss": 0.4244,
+      "step": 2050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8997372405847093e-05,
+      "loss": 0.2034,
+      "step": 2051
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8996405524864355e-05,
+      "loss": 1.3616,
+      "step": 2052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8995438194037377e-05,
+      "loss": 0.4024,
+      "step": 2053
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.899447041339725e-05,
+      "loss": 0.501,
+      "step": 2054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8993502182975076e-05,
+      "loss": 0.2922,
+      "step": 2055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8992533502801974e-05,
+      "loss": 0.7667,
+      "step": 2056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.899156437290908e-05,
+      "loss": 0.2262,
+      "step": 2057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.899059479332754e-05,
+      "loss": 0.2871,
+      "step": 2058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8989624764088523e-05,
+      "loss": 0.2907,
+      "step": 2059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8988654285223193e-05,
+      "loss": 1.2716,
+      "step": 2060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8987683356762753e-05,
+      "loss": 0.5989,
+      "step": 2061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.89867119787384e-05,
+      "loss": 0.397,
+      "step": 2062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8985740151181356e-05,
+      "loss": 1.1608,
+      "step": 2063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8984767874122864e-05,
+      "loss": 1.6679,
+      "step": 2064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.898379514759416e-05,
+      "loss": 0.5231,
+      "step": 2065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8982821971626517e-05,
+      "loss": 0.8377,
+      "step": 2066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8981848346251208e-05,
+      "loss": 0.0895,
+      "step": 2067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8980874271499526e-05,
+      "loss": 0.8065,
+      "step": 2068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8979899747402775e-05,
+      "loss": 0.9887,
+      "step": 2069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8978924773992287e-05,
+      "loss": 0.6387,
+      "step": 2070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8977949351299386e-05,
+      "loss": 0.4755,
+      "step": 2071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8976973479355427e-05,
+      "loss": 0.4981,
+      "step": 2072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8975997158191775e-05,
+      "loss": 0.5411,
+      "step": 2073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8975020387839806e-05,
+      "loss": 0.8386,
+      "step": 2074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8974043168330917e-05,
+      "loss": 0.5819,
+      "step": 2075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8973065499696513e-05,
+      "loss": 0.9937,
+      "step": 2076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8972087381968015e-05,
+      "loss": 0.3693,
+      "step": 2077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8971108815176868e-05,
+      "loss": 0.8434,
+      "step": 2078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.897012979935451e-05,
+      "loss": 0.5726,
+      "step": 2079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8969150334532417e-05,
+      "loss": 0.8991,
+      "step": 2080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8968170420742068e-05,
+      "loss": 0.1425,
+      "step": 2081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8967190058014952e-05,
+      "loss": 1.0651,
+      "step": 2082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8966209246382577e-05,
+      "loss": 0.844,
+      "step": 2083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.896522798587648e-05,
+      "loss": 0.8033,
+      "step": 2084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.896424627652818e-05,
+      "loss": 0.9609,
+      "step": 2085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8963264118369242e-05,
+      "loss": 1.4342,
+      "step": 2086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8962281511431226e-05,
+      "loss": 0.2674,
+      "step": 2087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.896129845574572e-05,
+      "loss": 0.6854,
+      "step": 2088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8960314951344305e-05,
+      "loss": 0.7232,
+      "step": 2089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.895933099825861e-05,
+      "loss": 0.4223,
+      "step": 2090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8958346596520245e-05,
+      "loss": 0.9039,
+      "step": 2091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.895736174616086e-05,
+      "loss": 0.3457,
+      "step": 2092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.89563764472121e-05,
+      "loss": 0.4076,
+      "step": 2093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.895539069970563e-05,
+      "loss": 0.7996,
+      "step": 2094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8954404503673135e-05,
+      "loss": 0.2967,
+      "step": 2095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8953417859146314e-05,
+      "loss": 0.7693,
+      "step": 2096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.895243076615688e-05,
+      "loss": 1.2088,
+      "step": 2097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.895144322473655e-05,
+      "loss": 0.3849,
+      "step": 2098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.895045523491707e-05,
+      "loss": 0.8703,
+      "step": 2099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.894946679673019e-05,
+      "loss": 0.4392,
+      "step": 2100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.894847791020768e-05,
+      "loss": 0.3574,
+      "step": 2101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8947488575381323e-05,
+      "loss": 1.5164,
+      "step": 2102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8946498792282922e-05,
+      "loss": 1.5001,
+      "step": 2103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8945508560944274e-05,
+      "loss": 0.3985,
+      "step": 2104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8944517881397217e-05,
+      "loss": 0.5957,
+      "step": 2105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8943526753673592e-05,
+      "loss": 0.1305,
+      "step": 2106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.894253517780524e-05,
+      "loss": 1.4062,
+      "step": 2107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.894154315382405e-05,
+      "loss": 0.2527,
+      "step": 2108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8940550681761893e-05,
+      "loss": 1.5712,
+      "step": 2109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.893955776165067e-05,
+      "loss": 0.7047,
+      "step": 2110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8938564393522294e-05,
+      "loss": 0.8762,
+      "step": 2111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8937570577408693e-05,
+      "loss": 0.4227,
+      "step": 2112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.89365763133418e-05,
+      "loss": 1.3143,
+      "step": 2113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.893558160135358e-05,
+      "loss": 0.7392,
+      "step": 2114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8934586441476006e-05,
+      "loss": 1.403,
+      "step": 2115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8933590833741052e-05,
+      "loss": 0.6943,
+      "step": 2116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8932594778180717e-05,
+      "loss": 0.8371,
+      "step": 2117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8931598274827023e-05,
+      "loss": 0.599,
+      "step": 2118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8930601323711998e-05,
+      "loss": 0.424,
+      "step": 2119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.892960392486767e-05,
+      "loss": 0.6293,
+      "step": 2120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8928606078326108e-05,
+      "loss": 0.5766,
+      "step": 2121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.892760778411938e-05,
+      "loss": 0.6334,
+      "step": 2122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8926609042279568e-05,
+      "loss": 0.6667,
+      "step": 2123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.892560985283878e-05,
+      "loss": 0.8136,
+      "step": 2124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.892461021582912e-05,
+      "loss": 0.6767,
+      "step": 2125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.892361013128272e-05,
+      "loss": 0.3322,
+      "step": 2126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.892260959923172e-05,
+      "loss": 0.8807,
+      "step": 2127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8921608619708284e-05,
+      "loss": 0.6012,
+      "step": 2128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.892060719274458e-05,
+      "loss": 0.5631,
+      "step": 2129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8919605318372794e-05,
+      "loss": 0.4087,
+      "step": 2130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8918602996625122e-05,
+      "loss": 0.8641,
+      "step": 2131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8917600227533784e-05,
+      "loss": 0.3451,
+      "step": 2132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8916597011131004e-05,
+      "loss": 0.3665,
+      "step": 2133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8915593347449034e-05,
+      "loss": 0.1666,
+      "step": 2134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8914589236520123e-05,
+      "loss": 1.0443,
+      "step": 2135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8913584678376548e-05,
+      "loss": 0.9668,
+      "step": 2136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8912579673050595e-05,
+      "loss": 1.3535,
+      "step": 2137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.891157422057456e-05,
+      "loss": 0.2364,
+      "step": 2138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8910568320980763e-05,
+      "loss": 1.0469,
+      "step": 2139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8909561974301537e-05,
+      "loss": 0.744,
+      "step": 2140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8908555180569216e-05,
+      "loss": 0.9625,
+      "step": 2141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8907547939816163e-05,
+      "loss": 1.0009,
+      "step": 2142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8906540252074757e-05,
+      "loss": 0.5691,
+      "step": 2143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8905532117377374e-05,
+      "loss": 0.6995,
+      "step": 2144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8904523535756424e-05,
+      "loss": 0.2039,
+      "step": 2145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8903514507244322e-05,
+      "loss": 0.6956,
+      "step": 2146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8902505031873494e-05,
+      "loss": 0.7644,
+      "step": 2147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8901495109676385e-05,
+      "loss": 0.7441,
+      "step": 2148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8900484740685452e-05,
+      "loss": 0.1459,
+      "step": 2149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.889947392493318e-05,
+      "loss": 0.2854,
+      "step": 2150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8898462662452044e-05,
+      "loss": 0.0814,
+      "step": 2151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.889745095327455e-05,
+      "loss": 1.3644,
+      "step": 2152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8896438797433216e-05,
+      "loss": 1.6556,
+      "step": 2153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.889542619496057e-05,
+      "loss": 1.4556,
+      "step": 2154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8894413145889155e-05,
+      "loss": 0.4819,
+      "step": 2155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8893399650251538e-05,
+      "loss": 1.6216,
+      "step": 2156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.889238570808029e-05,
+      "loss": 0.1747,
+      "step": 2157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.889137131940799e-05,
+      "loss": 1.1222,
+      "step": 2158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8890356484267256e-05,
+      "loss": 0.3339,
+      "step": 2159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8889341202690692e-05,
+      "loss": 0.6167,
+      "step": 2160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8888325474710932e-05,
+      "loss": 0.4446,
+      "step": 2161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8887309300360626e-05,
+      "loss": 0.8163,
+      "step": 2162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8886292679672433e-05,
+      "loss": 0.3701,
+      "step": 2163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8885275612679023e-05,
+      "loss": 1.0773,
+      "step": 2164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8884258099413085e-05,
+      "loss": 0.2305,
+      "step": 2165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8883240139907326e-05,
+      "loss": 0.214,
+      "step": 2166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8882221734194464e-05,
+      "loss": 0.1703,
+      "step": 2167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8881202882307217e-05,
+      "loss": 1.6798,
+      "step": 2168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8880183584278352e-05,
+      "loss": 0.3237,
+      "step": 2169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.887916384014061e-05,
+      "loss": 1.4028,
+      "step": 2170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8878143649926783e-05,
+      "loss": 1.7982,
+      "step": 2171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8877123013669646e-05,
+      "loss": 0.3232,
+      "step": 2172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.887610193140201e-05,
+      "loss": 0.8005,
+      "step": 2173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8875080403156684e-05,
+      "loss": 0.9173,
+      "step": 2174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.887405842896651e-05,
+      "loss": 0.8157,
+      "step": 2175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.887303600886433e-05,
+      "loss": 1.2332,
+      "step": 2176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8872013142883e-05,
+      "loss": 0.8514,
+      "step": 2177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.88709898310554e-05,
+      "loss": 0.6381,
+      "step": 2178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8869966073414426e-05,
+      "loss": 1.4182,
+      "step": 2179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8868941869992964e-05,
+      "loss": 0.5788,
+      "step": 2180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8867917220823947e-05,
+      "loss": 1.3335,
+      "step": 2181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8866892125940302e-05,
+      "loss": 0.7891,
+      "step": 2182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8865866585374978e-05,
+      "loss": 0.175,
+      "step": 2183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.886484059916093e-05,
+      "loss": 1.4613,
+      "step": 2184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8863814167331136e-05,
+      "loss": 0.7284,
+      "step": 2185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.886278728991859e-05,
+      "loss": 0.8397,
+      "step": 2186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8861759966956292e-05,
+      "loss": 0.5896,
+      "step": 2187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8860732198477255e-05,
+      "loss": 0.9617,
+      "step": 2188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8859703984514522e-05,
+      "loss": 0.1397,
+      "step": 2189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8858675325101132e-05,
+      "loss": 1.2931,
+      "step": 2190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8857646220270148e-05,
+      "loss": 0.3814,
+      "step": 2191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8856616670054647e-05,
+      "loss": 1.1054,
+      "step": 2192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8855586674487718e-05,
+      "loss": 1.4621,
+      "step": 2193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8854556233602464e-05,
+      "loss": 1.1998,
+      "step": 2194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8853525347432003e-05,
+      "loss": 0.2894,
+      "step": 2195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8852494016009474e-05,
+      "loss": 0.7577,
+      "step": 2196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8851462239368014e-05,
+      "loss": 1.1164,
+      "step": 2197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.885043001754079e-05,
+      "loss": 0.6726,
+      "step": 2198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8849397350560974e-05,
+      "loss": 0.817,
+      "step": 2199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8848364238461758e-05,
+      "loss": 0.6661,
+      "step": 2200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.884733068127635e-05,
+      "loss": 1.5277,
+      "step": 2201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8846296679037966e-05,
+      "loss": 0.9763,
+      "step": 2202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.884526223177984e-05,
+      "loss": 0.4344,
+      "step": 2203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.884422733953521e-05,
+      "loss": 1.5935,
+      "step": 2204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.884319200233735e-05,
+      "loss": 1.6554,
+      "step": 2205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.884215622021953e-05,
+      "loss": 0.3929,
+      "step": 2206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8841119993215038e-05,
+      "loss": 0.8225,
+      "step": 2207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8840083321357184e-05,
+      "loss": 1.2536,
+      "step": 2208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.883904620467928e-05,
+      "loss": 0.4607,
+      "step": 2209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8838008643214664e-05,
+      "loss": 1.283,
+      "step": 2210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.883697063699668e-05,
+      "loss": 0.1372,
+      "step": 2211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.883593218605869e-05,
+      "loss": 0.2972,
+      "step": 2212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8834893290434074e-05,
+      "loss": 0.4041,
+      "step": 2213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8833853950156217e-05,
+      "loss": 0.8553,
+      "step": 2214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.883281416525852e-05,
+      "loss": 1.1373,
+      "step": 2215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8831773935774414e-05,
+      "loss": 0.4686,
+      "step": 2216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8830733261737326e-05,
+      "loss": 1.3029,
+      "step": 2217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8829692143180695e-05,
+      "loss": 1.2433,
+      "step": 2218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8828650580137992e-05,
+      "loss": 1.1669,
+      "step": 2219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8827608572642692e-05,
+      "loss": 1.1672,
+      "step": 2220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8826566120728283e-05,
+      "loss": 1.979,
+      "step": 2221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8825523224428267e-05,
+      "loss": 0.1186,
+      "step": 2222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.882447988377617e-05,
+      "loss": 0.408,
+      "step": 2223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8823436098805517e-05,
+      "loss": 0.3214,
+      "step": 2224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.882239186954986e-05,
+      "loss": 0.5103,
+      "step": 2225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.882134719604276e-05,
+      "loss": 0.8972,
+      "step": 2226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.882030207831779e-05,
+      "loss": 0.6946,
+      "step": 2227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8819256516408544e-05,
+      "loss": 0.2102,
+      "step": 2228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8818210510348625e-05,
+      "loss": 0.9763,
+      "step": 2229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.881716406017165e-05,
+      "loss": 0.6055,
+      "step": 2230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.881611716591125e-05,
+      "loss": 0.2568,
+      "step": 2231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.881506982760108e-05,
+      "loss": 0.5982,
+      "step": 2232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8814022045274794e-05,
+      "loss": 0.7526,
+      "step": 2233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.881297381896607e-05,
+      "loss": 0.581,
+      "step": 2234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.88119251487086e-05,
+      "loss": 0.7847,
+      "step": 2235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8810876034536084e-05,
+      "loss": 0.7624,
+      "step": 2236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8809826476482244e-05,
+      "loss": 0.8493,
+      "step": 2237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8808776474580812e-05,
+      "loss": 0.6045,
+      "step": 2238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8807726028865534e-05,
+      "loss": 0.5933,
+      "step": 2239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8806675139370172e-05,
+      "loss": 0.6012,
+      "step": 2240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.88056238061285e-05,
+      "loss": 1.4217,
+      "step": 2241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8804572029174314e-05,
+      "loss": 0.741,
+      "step": 2242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.880351980854141e-05,
+      "loss": 0.5657,
+      "step": 2243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.880246714426361e-05,
+      "loss": 0.8062,
+      "step": 2244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8801414036374753e-05,
+      "loss": 0.7972,
+      "step": 2245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.880036048490867e-05,
+      "loss": 0.9422,
+      "step": 2246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8799306489899235e-05,
+      "loss": 0.8029,
+      "step": 2247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.879825205138032e-05,
+      "loss": 0.2798,
+      "step": 2248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8797197169385818e-05,
+      "loss": 0.9238,
+      "step": 2249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8796141843949624e-05,
+      "loss": 0.6019,
+      "step": 2250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8795086075105663e-05,
+      "loss": 1.3661,
+      "step": 2251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.879402986288787e-05,
+      "loss": 1.309,
+      "step": 2252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8792973207330185e-05,
+      "loss": 0.0779,
+      "step": 2253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.879191610846657e-05,
+      "loss": 0.4828,
+      "step": 2254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8790858566331003e-05,
+      "loss": 0.4012,
+      "step": 2255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8789800580957473e-05,
+      "loss": 0.9181,
+      "step": 2256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.878874215237998e-05,
+      "loss": 1.7421,
+      "step": 2257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8787683280632545e-05,
+      "loss": 1.517,
+      "step": 2258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8786623965749202e-05,
+      "loss": 0.7241,
+      "step": 2259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.878556420776399e-05,
+      "loss": 0.7448,
+      "step": 2260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.878450400671098e-05,
+      "loss": 0.8421,
+      "step": 2261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.878344336262424e-05,
+      "loss": 1.2401,
+      "step": 2262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.878238227553786e-05,
+      "loss": 0.622,
+      "step": 2263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8781320745485945e-05,
+      "loss": 0.9412,
+      "step": 2264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.878025877250261e-05,
+      "loss": 0.3177,
+      "step": 2265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8779196356621992e-05,
+      "loss": 0.552,
+      "step": 2266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8778133497878226e-05,
+      "loss": 0.5107,
+      "step": 2267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8777070196305483e-05,
+      "loss": 0.2925,
+      "step": 2268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8776006451937936e-05,
+      "loss": 0.2262,
+      "step": 2269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.877494226480977e-05,
+      "loss": 0.7009,
+      "step": 2270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8773877634955192e-05,
+      "loss": 1.0226,
+      "step": 2271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8772812562408416e-05,
+      "loss": 0.4441,
+      "step": 2272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8771747047203675e-05,
+      "loss": 1.2682,
+      "step": 2273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8770681089375207e-05,
+      "loss": 0.8474,
+      "step": 2274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8769614688957285e-05,
+      "loss": 0.8205,
+      "step": 2275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8768547845984174e-05,
+      "loss": 1.0749,
+      "step": 2276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8767480560490172e-05,
+      "loss": 0.8512,
+      "step": 2277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8766412832509567e-05,
+      "loss": 0.5393,
+      "step": 2278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8765344662076687e-05,
+      "loss": 0.6057,
+      "step": 2279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8764276049225858e-05,
+      "loss": 0.9679,
+      "step": 2280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8763206993991428e-05,
+      "loss": 0.8633,
+      "step": 2281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8762137496407757e-05,
+      "loss": 1.2597,
+      "step": 2282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.876106755650921e-05,
+      "loss": 0.5281,
+      "step": 2283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8759997174330186e-05,
+      "loss": 0.8095,
+      "step": 2284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8758926349905083e-05,
+      "loss": 0.4955,
+      "step": 2285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8757855083268317e-05,
+      "loss": 1.6798,
+      "step": 2286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.875678337445432e-05,
+      "loss": 0.7104,
+      "step": 2287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8755711223497534e-05,
+      "loss": 0.8286,
+      "step": 2288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8754638630432414e-05,
+      "loss": 0.1532,
+      "step": 2289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8753565595293444e-05,
+      "loss": 0.3631,
+      "step": 2290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8752492118115103e-05,
+      "loss": 1.6659,
+      "step": 2291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8751418198931894e-05,
+      "loss": 1.4002,
+      "step": 2292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.875034383777834e-05,
+      "loss": 0.3506,
+      "step": 2293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8749269034688958e-05,
+      "loss": 0.9087,
+      "step": 2294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.87481937896983e-05,
+      "loss": 0.4911,
+      "step": 2295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8747118102840927e-05,
+      "loss": 0.2696,
+      "step": 2296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8746041974151408e-05,
+      "loss": 0.7222,
+      "step": 2297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8744965403664323e-05,
+      "loss": 0.1506,
+      "step": 2298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8743888391414286e-05,
+      "loss": 0.4502,
+      "step": 2299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8742810937435902e-05,
+      "loss": 0.3111,
+      "step": 2300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8741733041763803e-05,
+      "loss": 0.2127,
+      "step": 2301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8740654704432638e-05,
+      "loss": 1.4996,
+      "step": 2302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8739575925477055e-05,
+      "loss": 0.6439,
+      "step": 2303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8738496704931737e-05,
+      "loss": 0.1488,
+      "step": 2304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8737417042831363e-05,
+      "loss": 0.0939,
+      "step": 2305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8736336939210633e-05,
+      "loss": 0.2226,
+      "step": 2306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.873525639410426e-05,
+      "loss": 0.8828,
+      "step": 2307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.873417540754698e-05,
+      "loss": 0.9065,
+      "step": 2308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.873309397957353e-05,
+      "loss": 1.1924,
+      "step": 2309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8732012110218673e-05,
+      "loss": 0.9426,
+      "step": 2310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8730929799517173e-05,
+      "loss": 0.5493,
+      "step": 2311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.872984704750382e-05,
+      "loss": 0.7185,
+      "step": 2312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.872876385421341e-05,
+      "loss": 0.2335,
+      "step": 2313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.872768021968076e-05,
+      "loss": 0.6047,
+      "step": 2314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8726596143940696e-05,
+      "loss": 0.3576,
+      "step": 2315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8725511627028063e-05,
+      "loss": 0.0998,
+      "step": 2316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8724426668977718e-05,
+      "loss": 1.6012,
+      "step": 2317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8723341269824523e-05,
+      "loss": 0.4894,
+      "step": 2318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8722255429603372e-05,
+      "loss": 1.0596,
+      "step": 2319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.872116914834916e-05,
+      "loss": 0.7986,
+      "step": 2320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8720082426096804e-05,
+      "loss": 0.3873,
+      "step": 2321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8718995262881226e-05,
+      "loss": 0.8012,
+      "step": 2322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8717907658737365e-05,
+      "loss": 1.6098,
+      "step": 2323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8716819613700186e-05,
+      "loss": 0.9713,
+      "step": 2324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8715731127804653e-05,
+      "loss": 0.6149,
+      "step": 2325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.871464220108575e-05,
+      "loss": 0.5183,
+      "step": 2326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8713552833578477e-05,
+      "loss": 1.4239,
+      "step": 2327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8712463025317844e-05,
+      "loss": 0.9211,
+      "step": 2328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8711372776338877e-05,
+      "loss": 0.2214,
+      "step": 2329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.871028208667662e-05,
+      "loss": 0.5765,
+      "step": 2330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8709190956366123e-05,
+      "loss": 0.3228,
+      "step": 2331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.870809938544246e-05,
+      "loss": 0.2145,
+      "step": 2332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.870700737394071e-05,
+      "loss": 0.5272,
+      "step": 2333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8705914921895976e-05,
+      "loss": 0.7061,
+      "step": 2334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8704822029343363e-05,
+      "loss": 1.2544,
+      "step": 2335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8703728696317997e-05,
+      "loss": 0.7301,
+      "step": 2336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8702634922855022e-05,
+      "loss": 0.2187,
+      "step": 2337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8701540708989588e-05,
+      "loss": 0.2157,
+      "step": 2338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8700446054756865e-05,
+      "loss": 1.1949,
+      "step": 2339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8699350960192034e-05,
+      "loss": 0.6174,
+      "step": 2340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8698255425330293e-05,
+      "loss": 0.6511,
+      "step": 2341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8697159450206853e-05,
+      "loss": 0.6235,
+      "step": 2342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.869606303485693e-05,
+      "loss": 0.8855,
+      "step": 2343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8694966179315776e-05,
+      "loss": 0.7084,
+      "step": 2344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.869386888361864e-05,
+      "loss": 0.918,
+      "step": 2345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8692771147800783e-05,
+      "loss": 0.7067,
+      "step": 2346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8691672971897494e-05,
+      "loss": 0.4805,
+      "step": 2347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8690574355944056e-05,
+      "loss": 0.9063,
+      "step": 2348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8689475299975795e-05,
+      "loss": 0.5892,
+      "step": 2349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8688375804028024e-05,
+      "loss": 0.4694,
+      "step": 2350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.868727586813608e-05,
+      "loss": 0.5105,
+      "step": 2351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8686175492335324e-05,
+      "loss": 1.4982,
+      "step": 2352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8685074676661113e-05,
+      "loss": 1.5014,
+      "step": 2353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8683973421148828e-05,
+      "loss": 1.1118,
+      "step": 2354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.868287172583387e-05,
+      "loss": 1.451,
+      "step": 2355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8681769590751642e-05,
+      "loss": 1.4357,
+      "step": 2356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.868066701593757e-05,
+      "loss": 1.7175,
+      "step": 2357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.867956400142708e-05,
+      "loss": 0.2883,
+      "step": 2358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.867846054725564e-05,
+      "loss": 1.7866,
+      "step": 2359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8677356653458706e-05,
+      "loss": 0.3266,
+      "step": 2360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8676252320071756e-05,
+      "loss": 0.2782,
+      "step": 2361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8675147547130285e-05,
+      "loss": 0.1164,
+      "step": 2362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8674042334669797e-05,
+      "loss": 0.2165,
+      "step": 2363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8672936682725824e-05,
+      "loss": 0.2016,
+      "step": 2364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8671830591333888e-05,
+      "loss": 0.4265,
+      "step": 2365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8670724060529546e-05,
+      "loss": 1.1497,
+      "step": 2366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8669617090348363e-05,
+      "loss": 0.7851,
+      "step": 2367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8668509680825916e-05,
+      "loss": 1.366,
+      "step": 2368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8667401831997793e-05,
+      "loss": 0.9659,
+      "step": 2369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8666293543899606e-05,
+      "loss": 0.5308,
+      "step": 2370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8665184816566972e-05,
+      "loss": 0.7283,
+      "step": 2371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8664075650035524e-05,
+      "loss": 0.6807,
+      "step": 2372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8662966044340918e-05,
+      "loss": 0.4803,
+      "step": 2373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8661855999518808e-05,
+      "loss": 1.1389,
+      "step": 2374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8660745515604874e-05,
+      "loss": 0.6919,
+      "step": 2375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.865963459263481e-05,
+      "loss": 0.8723,
+      "step": 2376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8658523230644313e-05,
+      "loss": 0.5791,
+      "step": 2377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8657411429669112e-05,
+      "loss": 0.8926,
+      "step": 2378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8656299189744934e-05,
+      "loss": 1.3137,
+      "step": 2379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.865518651090753e-05,
+      "loss": 0.7573,
+      "step": 2380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.865407339319266e-05,
+      "loss": 0.8198,
+      "step": 2381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8652959836636095e-05,
+      "loss": 0.9201,
+      "step": 2382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.865184584127363e-05,
+      "loss": 0.3599,
+      "step": 2383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8650731407141072e-05,
+      "loss": 0.7291,
+      "step": 2384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.864961653427423e-05,
+      "loss": 1.4278,
+      "step": 2385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8648501222708948e-05,
+      "loss": 0.1924,
+      "step": 2386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8647385472481058e-05,
+      "loss": 0.5161,
+      "step": 2387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8646269283626427e-05,
+      "loss": 0.1883,
+      "step": 2388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.864515265618093e-05,
+      "loss": 1.1803,
+      "step": 2389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8644035590180453e-05,
+      "loss": 1.1876,
+      "step": 2390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8642918085660905e-05,
+      "loss": 0.5549,
+      "step": 2391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8641800142658194e-05,
+      "loss": 0.8641,
+      "step": 2392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8640681761208252e-05,
+      "loss": 0.9793,
+      "step": 2393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.863956294134703e-05,
+      "loss": 0.421,
+      "step": 2394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8638443683110484e-05,
+      "loss": 1.1221,
+      "step": 2395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8637323986534582e-05,
+      "loss": 1.2688,
+      "step": 2396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8636203851655315e-05,
+      "loss": 1.1706,
+      "step": 2397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.863508327850869e-05,
+      "loss": 0.7834,
+      "step": 2398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.863396226713071e-05,
+      "loss": 1.2464,
+      "step": 2399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.863284081755741e-05,
+      "loss": 0.436,
+      "step": 2400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8631718929824836e-05,
+      "loss": 0.3324,
+      "step": 2401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8630596603969043e-05,
+      "loss": 1.3889,
+      "step": 2402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.86294738400261e-05,
+      "loss": 1.5457,
+      "step": 2403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8628350638032103e-05,
+      "loss": 0.7374,
+      "step": 2404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8627226998023137e-05,
+      "loss": 0.4655,
+      "step": 2405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8626102920035328e-05,
+      "loss": 1.6298,
+      "step": 2406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8624978404104794e-05,
+      "loss": 0.427,
+      "step": 2407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8623853450267682e-05,
+      "loss": 1.1844,
+      "step": 2408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.862272805856015e-05,
+      "loss": 0.5156,
+      "step": 2409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8621602229018365e-05,
+      "loss": 0.9863,
+      "step": 2410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8620475961678514e-05,
+      "loss": 0.1979,
+      "step": 2411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8619349256576793e-05,
+      "loss": 1.7939,
+      "step": 2412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.861822211374941e-05,
+      "loss": 0.2095,
+      "step": 2413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8617094533232598e-05,
+      "loss": 1.2687,
+      "step": 2414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8615966515062597e-05,
+      "loss": 1.3069,
+      "step": 2415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8614838059275657e-05,
+      "loss": 0.8212,
+      "step": 2416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.861370916590805e-05,
+      "loss": 0.8298,
+      "step": 2417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8612579834996057e-05,
+      "loss": 0.5012,
+      "step": 2418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.861145006657598e-05,
+      "loss": 0.4939,
+      "step": 2419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.861031986068412e-05,
+      "loss": 0.1099,
+      "step": 2420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.860918921735681e-05,
+      "loss": 0.6385,
+      "step": 2421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8608058136630385e-05,
+      "loss": 1.2342,
+      "step": 2422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.86069266185412e-05,
+      "loss": 1.5077,
+      "step": 2423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8605794663125618e-05,
+      "loss": 1.3419,
+      "step": 2424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8604662270420022e-05,
+      "loss": 0.4193,
+      "step": 2425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.860352944046081e-05,
+      "loss": 0.1544,
+      "step": 2426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.860239617328439e-05,
+      "loss": 1.3256,
+      "step": 2427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.860126246892718e-05,
+      "loss": 0.9231,
+      "step": 2428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8600128327425628e-05,
+      "loss": 1.2176,
+      "step": 2429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8598993748816173e-05,
+      "loss": 0.6884,
+      "step": 2430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8597858733135288e-05,
+      "loss": 1.4762,
+      "step": 2431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8596723280419452e-05,
+      "loss": 0.9452,
+      "step": 2432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8595587390705157e-05,
+      "loss": 1.5674,
+      "step": 2433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.859445106402891e-05,
+      "loss": 1.1882,
+      "step": 2434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.859331430042723e-05,
+      "loss": 2.6348,
+      "step": 2435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8592177099936658e-05,
+      "loss": 1.1411,
+      "step": 2436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8591039462593744e-05,
+      "loss": 0.8177,
+      "step": 2437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8589901388435044e-05,
+      "loss": 1.3295,
+      "step": 2438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8588762877497137e-05,
+      "loss": 0.7475,
+      "step": 2439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.858762392981662e-05,
+      "loss": 1.0583,
+      "step": 2440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8586484545430103e-05,
+      "loss": 0.9277,
+      "step": 2441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8585344724374197e-05,
+      "loss": 0.8022,
+      "step": 2442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8584204466685536e-05,
+      "loss": 0.6445,
+      "step": 2443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.858306377240077e-05,
+      "loss": 0.5153,
+      "step": 2444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8581922641556566e-05,
+      "loss": 0.989,
+      "step": 2445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.858078107418959e-05,
+      "loss": 0.9169,
+      "step": 2446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8579639070336542e-05,
+      "loss": 0.2389,
+      "step": 2447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8578496630034118e-05,
+      "loss": 1.0035,
+      "step": 2448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.857735375331904e-05,
+      "loss": 0.5904,
+      "step": 2449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.857621044022804e-05,
+      "loss": 0.4604,
+      "step": 2450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8575066690797863e-05,
+      "loss": 0.5141,
+      "step": 2451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8573922505065266e-05,
+      "loss": 0.6532,
+      "step": 2452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.857277788306703e-05,
+      "loss": 0.1229,
+      "step": 2453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8571632824839935e-05,
+      "loss": 0.1553,
+      "step": 2454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.857048733042079e-05,
+      "loss": 0.4609,
+      "step": 2455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8569341399846412e-05,
+      "loss": 0.2831,
+      "step": 2456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8568195033153627e-05,
+      "loss": 0.457,
+      "step": 2457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8567048230379278e-05,
+      "loss": 1.2696,
+      "step": 2458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8565900991560224e-05,
+      "loss": 0.6799,
+      "step": 2459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8564753316733347e-05,
+      "loss": 1.6036,
+      "step": 2460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8563605205935518e-05,
+      "loss": 0.595,
+      "step": 2461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8562456659203645e-05,
+      "loss": 0.5574,
+      "step": 2462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8561307676574644e-05,
+      "loss": 0.7621,
+      "step": 2463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8560158258085437e-05,
+      "loss": 0.6331,
+      "step": 2464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.855900840377297e-05,
+      "loss": 0.8554,
+      "step": 2465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8557858113674205e-05,
+      "loss": 0.0987,
+      "step": 2466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8556707387826103e-05,
+      "loss": 0.4011,
+      "step": 2467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.855555622626565e-05,
+      "loss": 0.5035,
+      "step": 2468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8554404629029848e-05,
+      "loss": 1.6396,
+      "step": 2469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.855325259615571e-05,
+      "loss": 0.2929,
+      "step": 2470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8552100127680256e-05,
+      "loss": 0.8578,
+      "step": 2471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8550947223640533e-05,
+      "loss": 1.2702,
+      "step": 2472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8549793884073593e-05,
+      "loss": 0.3413,
+      "step": 2473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8548640109016504e-05,
+      "loss": 0.8462,
+      "step": 2474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8547485898506346e-05,
+      "loss": 0.3116,
+      "step": 2475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8546331252580217e-05,
+      "loss": 0.587,
+      "step": 2476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.854517617127523e-05,
+      "loss": 0.7278,
+      "step": 2477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8544020654628506e-05,
+      "loss": 0.6471,
+      "step": 2478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8542864702677183e-05,
+      "loss": 0.8843,
+      "step": 2479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8541708315458417e-05,
+      "loss": 0.47,
+      "step": 2480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8540551493009376e-05,
+      "loss": 0.8458,
+      "step": 2481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8539394235367227e-05,
+      "loss": 0.5167,
+      "step": 2482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.853823654256918e-05,
+      "loss": 1.2695,
+      "step": 2483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8537078414652434e-05,
+      "loss": 0.3929,
+      "step": 2484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8535919851654214e-05,
+      "loss": 0.6529,
+      "step": 2485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8534760853611758e-05,
+      "loss": 0.7323,
+      "step": 2486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.853360142056231e-05,
+      "loss": 1.2683,
+      "step": 2487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8532441552543146e-05,
+      "loss": 1.1028,
+      "step": 2488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.853128124959153e-05,
+      "loss": 0.5946,
+      "step": 2489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8530120511744758e-05,
+      "loss": 1.1018,
+      "step": 2490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8528959339040145e-05,
+      "loss": 0.3972,
+      "step": 2491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8527797731515e-05,
+      "loss": 1.4108,
+      "step": 2492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8526635689206672e-05,
+      "loss": 0.4018,
+      "step": 2493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8525473212152486e-05,
+      "loss": 1.3297,
+      "step": 2494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8524310300389828e-05,
+      "loss": 1.4008,
+      "step": 2495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8523146953956055e-05,
+      "loss": 1.0788,
+      "step": 2496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8521983172888565e-05,
+      "loss": 1.0416,
+      "step": 2497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8520818957224767e-05,
+      "loss": 0.9185,
+      "step": 2498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.851965430700207e-05,
+      "loss": 0.9194,
+      "step": 2499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.851848922225791e-05,
+      "loss": 0.4778,
+      "step": 2500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.851732370302973e-05,
+      "loss": 1.412,
+      "step": 2501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8516157749354997e-05,
+      "loss": 1.0951,
+      "step": 2502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8514991361271176e-05,
+      "loss": 0.5136,
+      "step": 2503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.851382453881576e-05,
+      "loss": 1.3204,
+      "step": 2504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8512657282026245e-05,
+      "loss": 0.569,
+      "step": 2505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8511489590940155e-05,
+      "loss": 0.1728,
+      "step": 2506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8510321465595015e-05,
+      "loss": 0.861,
+      "step": 2507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.850915290602837e-05,
+      "loss": 0.291,
+      "step": 2508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.850798391227777e-05,
+      "loss": 0.6329,
+      "step": 2509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.85068144843808e-05,
+      "loss": 0.4937,
+      "step": 2510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.850564462237503e-05,
+      "loss": 0.3731,
+      "step": 2511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8504474326298078e-05,
+      "loss": 1.025,
+      "step": 2512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.850330359618754e-05,
+      "loss": 0.4395,
+      "step": 2513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8502132432081048e-05,
+      "loss": 0.4613,
+      "step": 2514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.850096083401625e-05,
+      "loss": 0.8914,
+      "step": 2515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8499788802030793e-05,
+      "loss": 1.0121,
+      "step": 2516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8498616336162354e-05,
+      "loss": 0.3874,
+      "step": 2517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8497443436448607e-05,
+      "loss": 0.4942,
+      "step": 2518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8496270102927254e-05,
+      "loss": 0.394,
+      "step": 2519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.849509633563601e-05,
+      "loss": 0.6425,
+      "step": 2520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.849392213461259e-05,
+      "loss": 0.964,
+      "step": 2521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8492747499894737e-05,
+      "loss": 0.3589,
+      "step": 2522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8491572431520207e-05,
+      "loss": 0.4012,
+      "step": 2523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8490396929526763e-05,
+      "loss": 0.5965,
+      "step": 2524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.848922099395219e-05,
+      "loss": 1.3616,
+      "step": 2525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8488044624834277e-05,
+      "loss": 1.6313,
+      "step": 2526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8486867822210836e-05,
+      "loss": 0.5735,
+      "step": 2527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.848569058611969e-05,
+      "loss": 1.1308,
+      "step": 2528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8484512916598673e-05,
+      "loss": 1.1976,
+      "step": 2529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.848333481368563e-05,
+      "loss": 0.8756,
+      "step": 2530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.848215627741844e-05,
+      "loss": 0.4622,
+      "step": 2531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.848097730783497e-05,
+      "loss": 0.4428,
+      "step": 2532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.847979790497311e-05,
+      "loss": 1.6956,
+      "step": 2533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.847861806887078e-05,
+      "loss": 0.7317,
+      "step": 2534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8477437799565882e-05,
+      "loss": 0.2973,
+      "step": 2535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8476257097096362e-05,
+      "loss": 0.6628,
+      "step": 2536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.847507596150016e-05,
+      "loss": 0.9529,
+      "step": 2537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8473894392815245e-05,
+      "loss": 1.0753,
+      "step": 2538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.847271239107959e-05,
+      "loss": 0.7955,
+      "step": 2539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.847152995633118e-05,
+      "loss": 0.8798,
+      "step": 2540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8470347088608024e-05,
+      "loss": 1.0175,
+      "step": 2541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.846916378794814e-05,
+      "loss": 0.8776,
+      "step": 2542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8467980054389555e-05,
+      "loss": 0.8042,
+      "step": 2543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8466795887970317e-05,
+      "loss": 0.5049,
+      "step": 2544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8465611288728484e-05,
+      "loss": 1.2079,
+      "step": 2545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.846442625670213e-05,
+      "loss": 1.6829,
+      "step": 2546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.846324079192934e-05,
+      "loss": 0.7494,
+      "step": 2547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.846205489444821e-05,
+      "loss": 0.4434,
+      "step": 2548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.846086856429687e-05,
+      "loss": 0.9962,
+      "step": 2549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.845968180151343e-05,
+      "loss": 1.4949,
+      "step": 2550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.845849460613605e-05,
+      "loss": 1.3227,
+      "step": 2551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8457306978202876e-05,
+      "loss": 1.3121,
+      "step": 2552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8456118917752076e-05,
+      "loss": 1.3103,
+      "step": 2553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8454930424821846e-05,
+      "loss": 0.3823,
+      "step": 2554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8453741499450372e-05,
+      "loss": 0.3763,
+      "step": 2555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.845255214167587e-05,
+      "loss": 1.1304,
+      "step": 2556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.845136235153657e-05,
+      "loss": 0.5448,
+      "step": 2557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.845017212907071e-05,
+      "loss": 0.1027,
+      "step": 2558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.844898147431654e-05,
+      "loss": 0.7691,
+      "step": 2559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8447790387312325e-05,
+      "loss": 1.7493,
+      "step": 2560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8446598868096357e-05,
+      "loss": 0.668,
+      "step": 2561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.844540691670693e-05,
+      "loss": 0.7252,
+      "step": 2562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.844421453318234e-05,
+      "loss": 0.1072,
+      "step": 2563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8443021717560925e-05,
+      "loss": 1.5936,
+      "step": 2564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8441828469881017e-05,
+      "loss": 0.1958,
+      "step": 2565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8440634790180966e-05,
+      "loss": 0.3713,
+      "step": 2566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8439440678499137e-05,
+      "loss": 0.5763,
+      "step": 2567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8438246134873907e-05,
+      "loss": 0.3588,
+      "step": 2568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8437051159343674e-05,
+      "loss": 0.6231,
+      "step": 2569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8435855751946838e-05,
+      "loss": 0.3123,
+      "step": 2570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8434659912721825e-05,
+      "loss": 1.3819,
+      "step": 2571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.843346364170707e-05,
+      "loss": 0.579,
+      "step": 2572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8432266938941012e-05,
+      "loss": 0.7093,
+      "step": 2573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8431069804462116e-05,
+      "loss": 1.3996,
+      "step": 2574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8429872238308868e-05,
+      "loss": 0.9614,
+      "step": 2575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.842867424051975e-05,
+      "loss": 1.1467,
+      "step": 2576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8427475811133262e-05,
+      "loss": 0.5366,
+      "step": 2577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.842627695018793e-05,
+      "loss": 0.8397,
+      "step": 2578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.842507765772228e-05,
+      "loss": 1.3538,
+      "step": 2579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.842387793377486e-05,
+      "loss": 1.0044,
+      "step": 2580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.842267777838422e-05,
+      "loss": 1.4048,
+      "step": 2581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8421477191588945e-05,
+      "loss": 0.5505,
+      "step": 2582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8420276173427618e-05,
+      "loss": 1.2902,
+      "step": 2583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.841907472393884e-05,
+      "loss": 0.9593,
+      "step": 2584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8417872843161222e-05,
+      "loss": 0.302,
+      "step": 2585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8416670531133395e-05,
+      "loss": 0.2543,
+      "step": 2586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8415467787894004e-05,
+      "loss": 1.2457,
+      "step": 2587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8414264613481703e-05,
+      "loss": 0.8372,
+      "step": 2588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8413061007935158e-05,
+      "loss": 0.5534,
+      "step": 2589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8411856971293058e-05,
+      "loss": 0.9844,
+      "step": 2590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.84106525035941e-05,
+      "loss": 0.9395,
+      "step": 2591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8409447604876993e-05,
+      "loss": 0.7069,
+      "step": 2592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.840824227518046e-05,
+      "loss": 0.9395,
+      "step": 2593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8407036514543248e-05,
+      "loss": 1.0873,
+      "step": 2594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.840583032300411e-05,
+      "loss": 0.9225,
+      "step": 2595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8404623700601803e-05,
+      "loss": 1.3251,
+      "step": 2596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.840341664737512e-05,
+      "loss": 0.7548,
+      "step": 2597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8402209163362842e-05,
+      "loss": 0.6075,
+      "step": 2598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.840100124860379e-05,
+      "loss": 0.1,
+      "step": 2599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8399792903136785e-05,
+      "loss": 0.239,
+      "step": 2600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8398584127000656e-05,
+      "loss": 0.379,
+      "step": 2601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8397374920234255e-05,
+      "loss": 0.3546,
+      "step": 2602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8396165282876453e-05,
+      "loss": 0.4022,
+      "step": 2603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.839495521496612e-05,
+      "loss": 0.1032,
+      "step": 2604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.839374471654215e-05,
+      "loss": 0.5529,
+      "step": 2605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8392533787643452e-05,
+      "loss": 1.149,
+      "step": 2606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8391322428308936e-05,
+      "loss": 0.3785,
+      "step": 2607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8390110638577544e-05,
+      "loss": 1.1914,
+      "step": 2608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8388898418488215e-05,
+      "loss": 0.5029,
+      "step": 2609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.838768576807992e-05,
+      "loss": 1.3251,
+      "step": 2610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8386472687391628e-05,
+      "loss": 0.9135,
+      "step": 2611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.838525917646233e-05,
+      "loss": 0.529,
+      "step": 2612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.838404523533102e-05,
+      "loss": 1.5151,
+      "step": 2613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8382830864036718e-05,
+      "loss": 1.3381,
+      "step": 2614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8381616062618462e-05,
+      "loss": 0.4338,
+      "step": 2615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8380400831115284e-05,
+      "loss": 0.3585,
+      "step": 2616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8379185169566248e-05,
+      "loss": 0.3536,
+      "step": 2617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8377969078010427e-05,
+      "loss": 1.21,
+      "step": 2618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.83767525564869e-05,
+      "loss": 0.317,
+      "step": 2619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.837553560503477e-05,
+      "loss": 1.3277,
+      "step": 2620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8374318223693154e-05,
+      "loss": 1.4938,
+      "step": 2621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8373100412501167e-05,
+      "loss": 0.9824,
+      "step": 2622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8371882171497957e-05,
+      "loss": 1.1234,
+      "step": 2623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.837066350072268e-05,
+      "loss": 0.5271,
+      "step": 2624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8369444400214503e-05,
+      "loss": 0.5436,
+      "step": 2625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8368224870012605e-05,
+      "loss": 1.2683,
+      "step": 2626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.836700491015618e-05,
+      "loss": 0.8307,
+      "step": 2627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8365784520684443e-05,
+      "loss": 2.1894,
+      "step": 2628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8364563701636613e-05,
+      "loss": 1.2428,
+      "step": 2629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.836334245305193e-05,
+      "loss": 1.4102,
+      "step": 2630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.836212077496964e-05,
+      "loss": 1.1258,
+      "step": 2631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8360898667429016e-05,
+      "loss": 0.7405,
+      "step": 2632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8359676130469335e-05,
+      "loss": 0.4268,
+      "step": 2633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.835845316412988e-05,
+      "loss": 1.4009,
+      "step": 2634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8357229768449966e-05,
+      "loss": 0.8387,
+      "step": 2635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8356005943468913e-05,
+      "loss": 0.4296,
+      "step": 2636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8354781689226043e-05,
+      "loss": 0.6996,
+      "step": 2637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.835355700576072e-05,
+      "loss": 0.115,
+      "step": 2638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.83523318931123e-05,
+      "loss": 0.4149,
+      "step": 2639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8351106351320152e-05,
+      "loss": 0.7441,
+      "step": 2640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.834988038042367e-05,
+      "loss": 1.0082,
+      "step": 2641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8348653980462253e-05,
+      "loss": 0.7981,
+      "step": 2642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.834742715147532e-05,
+      "loss": 1.2295,
+      "step": 2643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8346199893502307e-05,
+      "loss": 0.7152,
+      "step": 2644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.834497220658265e-05,
+      "loss": 0.9646,
+      "step": 2645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8343744090755807e-05,
+      "loss": 0.9443,
+      "step": 2646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8342515546061253e-05,
+      "loss": 0.2532,
+      "step": 2647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.834128657253847e-05,
+      "loss": 0.6817,
+      "step": 2648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8340057170226964e-05,
+      "loss": 0.4211,
+      "step": 2649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.833882733916624e-05,
+      "loss": 0.0951,
+      "step": 2650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.833759707939583e-05,
+      "loss": 1.7552,
+      "step": 2651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.833636639095527e-05,
+      "loss": 0.6195,
+      "step": 2652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8335135273884117e-05,
+      "loss": 0.2931,
+      "step": 2653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.833390372822194e-05,
+      "loss": 1.604,
+      "step": 2654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8332671754008324e-05,
+      "loss": 0.3364,
+      "step": 2655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8331439351282854e-05,
+      "loss": 0.8159,
+      "step": 2656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.833020652008515e-05,
+      "loss": 0.2523,
+      "step": 2657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8328973260454827e-05,
+      "loss": 0.346,
+      "step": 2658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.832773957243153e-05,
+      "loss": 0.0604,
+      "step": 2659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8326505456054906e-05,
+      "loss": 1.656,
+      "step": 2660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.832527091136462e-05,
+      "loss": 1.5345,
+      "step": 2661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8324035938400346e-05,
+      "loss": 0.5727,
+      "step": 2662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.832280053720178e-05,
+      "loss": 0.697,
+      "step": 2663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8321564707808627e-05,
+      "loss": 0.3593,
+      "step": 2664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8320328450260608e-05,
+      "loss": 0.1459,
+      "step": 2665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8319091764597458e-05,
+      "loss": 0.1134,
+      "step": 2666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8317854650858917e-05,
+      "loss": 1.4445,
+      "step": 2667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.831661710908475e-05,
+      "loss": 0.4159,
+      "step": 2668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8315379139314737e-05,
+      "loss": 0.1854,
+      "step": 2669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8314140741588658e-05,
+      "loss": 0.5231,
+      "step": 2670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8312901915946315e-05,
+      "loss": 0.2928,
+      "step": 2671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8311662662427525e-05,
+      "loss": 1.664,
+      "step": 2672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8310422981072126e-05,
+      "loss": 0.6653,
+      "step": 2673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8309182871919948e-05,
+      "loss": 0.6418,
+      "step": 2674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.830794233501086e-05,
+      "loss": 1.1259,
+      "step": 2675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8306701370384723e-05,
+      "loss": 1.7326,
+      "step": 2676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8305459978081433e-05,
+      "loss": 0.8961,
+      "step": 2677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.830421815814088e-05,
+      "loss": 0.369,
+      "step": 2678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8302975910602975e-05,
+      "loss": 0.3526,
+      "step": 2679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8301733235507645e-05,
+      "loss": 0.6175,
+      "step": 2680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.830049013289483e-05,
+      "loss": 1.0215,
+      "step": 2681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8299246602804493e-05,
+      "loss": 0.7032,
+      "step": 2682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8298002645276582e-05,
+      "loss": 0.8683,
+      "step": 2683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8296758260351092e-05,
+      "loss": 0.8134,
+      "step": 2684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8295513448068018e-05,
+      "loss": 1.3727,
+      "step": 2685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8294268208467358e-05,
+      "loss": 1.3364,
+      "step": 2686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8293022541589145e-05,
+      "loss": 0.4146,
+      "step": 2687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8291776447473405e-05,
+      "loss": 1.3901,
+      "step": 2688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8290529926160193e-05,
+      "loss": 0.5124,
+      "step": 2689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.828928297768957e-05,
+      "loss": 1.45,
+      "step": 2690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8288035602101617e-05,
+      "loss": 1.3447,
+      "step": 2691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.828678779943642e-05,
+      "loss": 0.747,
+      "step": 2692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8285539569734085e-05,
+      "loss": 1.0555,
+      "step": 2693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8284290913034733e-05,
+      "loss": 0.9445,
+      "step": 2694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.828304182937849e-05,
+      "loss": 0.438,
+      "step": 2695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.82817923188055e-05,
+      "loss": 0.2203,
+      "step": 2696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8280542381355933e-05,
+      "loss": 1.2567,
+      "step": 2697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8279292017069957e-05,
+      "loss": 0.8745,
+      "step": 2698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8278041225987755e-05,
+      "loss": 0.5853,
+      "step": 2699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.827679000814953e-05,
+      "loss": 0.5841,
+      "step": 2700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8275538363595494e-05,
+      "loss": 0.249,
+      "step": 2701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.827428629236588e-05,
+      "loss": 1.6121,
+      "step": 2702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.827303379450092e-05,
+      "loss": 0.3322,
+      "step": 2703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8271780870040884e-05,
+      "loss": 0.4625,
+      "step": 2704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8270527519026025e-05,
+      "loss": 1.2286,
+      "step": 2705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8269273741496643e-05,
+      "loss": 0.2808,
+      "step": 2706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8268019537493017e-05,
+      "loss": 0.627,
+      "step": 2707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8266764907055464e-05,
+      "loss": 0.1805,
+      "step": 2708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.826550985022431e-05,
+      "loss": 0.6349,
+      "step": 2709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.826425436703989e-05,
+      "loss": 1.6777,
+      "step": 2710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.826299845754256e-05,
+      "loss": 0.3319,
+      "step": 2711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.826174212177268e-05,
+      "loss": 1.3476,
+      "step": 2712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8260485359770628e-05,
+      "loss": 0.0649,
+      "step": 2713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.82592281715768e-05,
+      "loss": 1.7589,
+      "step": 2714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8257970557231596e-05,
+      "loss": 0.5533,
+      "step": 2715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8256712516775444e-05,
+      "loss": 1.3376,
+      "step": 2716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.825545405024877e-05,
+      "loss": 1.6099,
+      "step": 2717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8254195157692026e-05,
+      "loss": 0.3122,
+      "step": 2718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8252935839145672e-05,
+      "loss": 0.5557,
+      "step": 2719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.825167609465018e-05,
+      "loss": 0.6186,
+      "step": 2720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.825041592424604e-05,
+      "loss": 0.5267,
+      "step": 2721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8249155327973753e-05,
+      "loss": 1.2862,
+      "step": 2722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8247894305873837e-05,
+      "loss": 0.3423,
+      "step": 2723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8246632857986817e-05,
+      "loss": 0.3355,
+      "step": 2724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8245370984353238e-05,
+      "loss": 0.7559,
+      "step": 2725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8244108685013655e-05,
+      "loss": 1.1927,
+      "step": 2726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8242845960008646e-05,
+      "loss": 0.9725,
+      "step": 2727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8241582809378787e-05,
+      "loss": 0.8694,
+      "step": 2728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8240319233164677e-05,
+      "loss": 0.9101,
+      "step": 2729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.823905523140693e-05,
+      "loss": 0.625,
+      "step": 2730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8237790804146163e-05,
+      "loss": 0.7828,
+      "step": 2731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8236525951423027e-05,
+      "loss": 1.0513,
+      "step": 2732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8235260673278166e-05,
+      "loss": 0.7074,
+      "step": 2733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8233994969752244e-05,
+      "loss": 1.4394,
+      "step": 2734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.823272884088595e-05,
+      "loss": 0.6489,
+      "step": 2735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.823146228671997e-05,
+      "loss": 0.7142,
+      "step": 2736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.823019530729501e-05,
+      "loss": 0.9876,
+      "step": 2737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.82289279026518e-05,
+      "loss": 0.2055,
+      "step": 2738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8227660072831062e-05,
+      "loss": 0.6798,
+      "step": 2739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8226391817873557e-05,
+      "loss": 0.9862,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8225123137820038e-05,
+      "loss": 0.9451,
+      "step": 2741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.822385403271128e-05,
+      "loss": 1.5751,
+      "step": 2742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8222584502588073e-05,
+      "loss": 0.8883,
+      "step": 2743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8221314547491223e-05,
+      "loss": 0.9047,
+      "step": 2744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8220044167461546e-05,
+      "loss": 0.156,
+      "step": 2745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8218773362539867e-05,
+      "loss": 0.2202,
+      "step": 2746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.821750213276704e-05,
+      "loss": 0.8109,
+      "step": 2747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8216230478183907e-05,
+      "loss": 0.5488,
+      "step": 2748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8214958398831347e-05,
+      "loss": 0.5804,
+      "step": 2749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8213685894750246e-05,
+      "loss": 0.6368,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8212412965981507e-05,
+      "loss": 0.1669,
+      "step": 2751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.821113961256603e-05,
+      "loss": 0.2183,
+      "step": 2752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8209865834544744e-05,
+      "loss": 1.2869,
+      "step": 2753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.82085916319586e-05,
+      "loss": 0.8782,
+      "step": 2754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8207317004848535e-05,
+      "loss": 0.4946,
+      "step": 2755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.820604195325552e-05,
+      "loss": 1.907,
+      "step": 2756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8204766477220546e-05,
+      "loss": 0.3594,
+      "step": 2757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.820349057678459e-05,
+      "loss": 0.1781,
+      "step": 2758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.820221425198867e-05,
+      "loss": 0.5371,
+      "step": 2759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8200937502873805e-05,
+      "loss": 0.8056,
+      "step": 2760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.819966032948103e-05,
+      "loss": 0.5735,
+      "step": 2761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.819838273185139e-05,
+      "loss": 0.2458,
+      "step": 2762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8197104710025953e-05,
+      "loss": 0.6034,
+      "step": 2763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.819582626404579e-05,
+      "loss": 0.364,
+      "step": 2764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8194547393951986e-05,
+      "loss": 0.7166,
+      "step": 2765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8193268099785654e-05,
+      "loss": 0.0432,
+      "step": 2766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8191988381587907e-05,
+      "loss": 0.7169,
+      "step": 2767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.819070823939987e-05,
+      "loss": 0.4594,
+      "step": 2768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.818942767326269e-05,
+      "loss": 0.8765,
+      "step": 2769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.818814668321753e-05,
+      "loss": 0.3956,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8186865269305554e-05,
+      "loss": 0.4552,
+      "step": 2771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8185583431567945e-05,
+      "loss": 1.1844,
+      "step": 2772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.818430117004591e-05,
+      "loss": 0.5721,
+      "step": 2773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.818301848478065e-05,
+      "loss": 0.4765,
+      "step": 2774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8181735375813398e-05,
+      "loss": 0.579,
+      "step": 2775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8180451843185397e-05,
+      "loss": 1.4246,
+      "step": 2776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8179167886937885e-05,
+      "loss": 0.5301,
+      "step": 2777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.817788350711214e-05,
+      "loss": 0.8067,
+      "step": 2778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8176598703749442e-05,
+      "loss": 1.3952,
+      "step": 2779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.817531347689108e-05,
+      "loss": 1.4282,
+      "step": 2780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8174027826578364e-05,
+      "loss": 0.4639,
+      "step": 2781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8172741752852615e-05,
+      "loss": 0.498,
+      "step": 2782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8171455255755164e-05,
+      "loss": 0.444,
+      "step": 2783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.817016833532736e-05,
+      "loss": 1.1767,
+      "step": 2784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8168880991610568e-05,
+      "loss": 1.0059,
+      "step": 2785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8167593224646157e-05,
+      "loss": 0.8574,
+      "step": 2786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.816630503447552e-05,
+      "loss": 0.9142,
+      "step": 2787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8165016421140064e-05,
+      "loss": 1.2784,
+      "step": 2788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.81637273846812e-05,
+      "loss": 0.4947,
+      "step": 2789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8162437925140354e-05,
+      "loss": 0.409,
+      "step": 2790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8161148042558976e-05,
+      "loss": 0.2252,
+      "step": 2791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8159857736978512e-05,
+      "loss": 0.6709,
+      "step": 2792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.815856700844045e-05,
+      "loss": 0.4742,
+      "step": 2793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8157275856986263e-05,
+      "loss": 1.2234,
+      "step": 2794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8155984282657445e-05,
+      "loss": 0.8711,
+      "step": 2795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8154692285495512e-05,
+      "loss": 0.2263,
+      "step": 2796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8153399865541994e-05,
+      "loss": 0.8174,
+      "step": 2797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8152107022838417e-05,
+      "loss": 0.7065,
+      "step": 2798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8150813757426345e-05,
+      "loss": 0.736,
+      "step": 2799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.814952006934734e-05,
+      "loss": 0.9547,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8148225958642973e-05,
+      "loss": 0.1822,
+      "step": 2801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8146931425354848e-05,
+      "loss": 0.1404,
+      "step": 2802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.814563646952456e-05,
+      "loss": 0.5644,
+      "step": 2803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.814434109119375e-05,
+      "loss": 0.3192,
+      "step": 2804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8143045290404027e-05,
+      "loss": 0.6479,
+      "step": 2805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8141749067197044e-05,
+      "loss": 0.4005,
+      "step": 2806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.814045242161447e-05,
+      "loss": 0.7597,
+      "step": 2807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8139155353697977e-05,
+      "loss": 1.4114,
+      "step": 2808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.813785786348925e-05,
+      "loss": 0.7099,
+      "step": 2809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.813655995102999e-05,
+      "loss": 0.4817,
+      "step": 2810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.813526161636191e-05,
+      "loss": 1.3311,
+      "step": 2811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8133962859526742e-05,
+      "loss": 1.4751,
+      "step": 2812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8132663680566228e-05,
+      "loss": 0.5531,
+      "step": 2813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8131364079522124e-05,
+      "loss": 0.4391,
+      "step": 2814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8130064056436194e-05,
+      "loss": 0.5408,
+      "step": 2815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8128763611350227e-05,
+      "loss": 1.2218,
+      "step": 2816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.812746274430602e-05,
+      "loss": 1.9006,
+      "step": 2817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8126161455345375e-05,
+      "loss": 0.4376,
+      "step": 2818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.812485974451012e-05,
+      "loss": 0.3826,
+      "step": 2819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8123557611842095e-05,
+      "loss": 1.5419,
+      "step": 2820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8122255057383145e-05,
+      "loss": 0.4945,
+      "step": 2821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8120952081175138e-05,
+      "loss": 1.1852,
+      "step": 2822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.811964868325995e-05,
+      "loss": 0.6933,
+      "step": 2823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.811834486367947e-05,
+      "loss": 0.5782,
+      "step": 2824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.811704062247561e-05,
+      "loss": 1.4016,
+      "step": 2825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.811573595969028e-05,
+      "loss": 1.0842,
+      "step": 2826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8114430875365415e-05,
+      "loss": 1.9369,
+      "step": 2827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8113125369542964e-05,
+      "loss": 0.8848,
+      "step": 2828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.811181944226488e-05,
+      "loss": 1.9197,
+      "step": 2829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.811051309357314e-05,
+      "loss": 0.8891,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8109206323509728e-05,
+      "loss": 0.2508,
+      "step": 2831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8107899132116645e-05,
+      "loss": 1.3451,
+      "step": 2832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8106591519435896e-05,
+      "loss": 0.6463,
+      "step": 2833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8105283485509524e-05,
+      "loss": 0.8756,
+      "step": 2834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.810397503037955e-05,
+      "loss": 1.6592,
+      "step": 2835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8102666154088044e-05,
+      "loss": 1.5004,
+      "step": 2836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8101356856677066e-05,
+      "loss": 0.4056,
+      "step": 2837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8100047138188696e-05,
+      "loss": 1.2582,
+      "step": 2838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8098736998665026e-05,
+      "loss": 0.3468,
+      "step": 2839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8097426438148176e-05,
+      "loss": 1.3597,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8096115456680254e-05,
+      "loss": 0.7061,
+      "step": 2841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8094804054303403e-05,
+      "loss": 1.2007,
+      "step": 2842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8093492231059762e-05,
+      "loss": 1.1451,
+      "step": 2843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.809217998699151e-05,
+      "loss": 0.5657,
+      "step": 2844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.80908673221408e-05,
+      "loss": 0.7601,
+      "step": 2845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8089554236549838e-05,
+      "loss": 1.1128,
+      "step": 2846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8088240730260825e-05,
+      "loss": 0.1959,
+      "step": 2847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8086926803315972e-05,
+      "loss": 0.7657,
+      "step": 2848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.808561245575751e-05,
+      "loss": 0.8069,
+      "step": 2849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.808429768762768e-05,
+      "loss": 0.4231,
+      "step": 2850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.808298249896874e-05,
+      "loss": 0.1042,
+      "step": 2851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8081666889822964e-05,
+      "loss": 0.3365,
+      "step": 2852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.808035086023263e-05,
+      "loss": 0.0874,
+      "step": 2853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8079034410240035e-05,
+      "loss": 1.5469,
+      "step": 2854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8077717539887498e-05,
+      "loss": 1.6246,
+      "step": 2855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8076400249217338e-05,
+      "loss": 1.2563,
+      "step": 2856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.807508253827189e-05,
+      "loss": 0.881,
+      "step": 2857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8073764407093508e-05,
+      "loss": 0.6075,
+      "step": 2858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.807244585572456e-05,
+      "loss": 0.4558,
+      "step": 2859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8071126884207418e-05,
+      "loss": 0.3687,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8069807492584475e-05,
+      "loss": 0.7966,
+      "step": 2861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.806848768089814e-05,
+      "loss": 0.4594,
+      "step": 2862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8067167449190826e-05,
+      "loss": 0.8935,
+      "step": 2863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.806584679750497e-05,
+      "loss": 0.6654,
+      "step": 2864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8064525725883014e-05,
+      "loss": 0.8286,
+      "step": 2865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8063204234367423e-05,
+      "loss": 0.1315,
+      "step": 2866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8061882323000667e-05,
+      "loss": 1.3191,
+      "step": 2867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8060559991825225e-05,
+      "loss": 0.2735,
+      "step": 2868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.805923724088361e-05,
+      "loss": 1.2849,
+      "step": 2869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8057914070218326e-05,
+      "loss": 0.7137,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.80565904798719e-05,
+      "loss": 0.7802,
+      "step": 2871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8055266469886875e-05,
+      "loss": 0.9305,
+      "step": 2872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8053942040305806e-05,
+      "loss": 0.8986,
+      "step": 2873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8052617191171253e-05,
+      "loss": 0.9565,
+      "step": 2874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.80512919225258e-05,
+      "loss": 0.4159,
+      "step": 2875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.804996623441205e-05,
+      "loss": 0.5131,
+      "step": 2876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.80486401268726e-05,
+      "loss": 0.8571,
+      "step": 2877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8047313599950076e-05,
+      "loss": 0.8033,
+      "step": 2878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8045986653687107e-05,
+      "loss": 0.6612,
+      "step": 2879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.804465928812635e-05,
+      "loss": 0.9962,
+      "step": 2880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8043331503310458e-05,
+      "loss": 0.5933,
+      "step": 2881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.804200329928211e-05,
+      "loss": 0.5616,
+      "step": 2882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8040674676083992e-05,
+      "loss": 1.0056,
+      "step": 2883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.803934563375881e-05,
+      "loss": 1.4573,
+      "step": 2884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8038016172349275e-05,
+      "loss": 2.3446,
+      "step": 2885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.803668629189812e-05,
+      "loss": 0.7704,
+      "step": 2886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8035355992448084e-05,
+      "loss": 0.2751,
+      "step": 2887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8034025274041923e-05,
+      "loss": 1.1605,
+      "step": 2888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.803269413672241e-05,
+      "loss": 0.8698,
+      "step": 2889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8031362580532326e-05,
+      "loss": 1.1992,
+      "step": 2890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8030030605514465e-05,
+      "loss": 0.1606,
+      "step": 2891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.802869821171164e-05,
+      "loss": 1.3224,
+      "step": 2892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8027365399166668e-05,
+      "loss": 0.9676,
+      "step": 2893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.802603216792239e-05,
+      "loss": 0.8794,
+      "step": 2894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8024698518021658e-05,
+      "loss": 0.476,
+      "step": 2895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8023364449507332e-05,
+      "loss": 0.7521,
+      "step": 2896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8022029962422287e-05,
+      "loss": 0.3113,
+      "step": 2897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8020695056809426e-05,
+      "loss": 0.1755,
+      "step": 2898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8019359732711634e-05,
+      "loss": 0.3307,
+      "step": 2899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.801802399017184e-05,
+      "loss": 0.5215,
+      "step": 2900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8016687829232977e-05,
+      "loss": 1.4812,
+      "step": 2901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.801535124993798e-05,
+      "loss": 1.6432,
+      "step": 2902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.801401425232981e-05,
+      "loss": 0.6484,
+      "step": 2903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8012676836451444e-05,
+      "loss": 1.4771,
+      "step": 2904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8011339002345862e-05,
+      "loss": 0.3189,
+      "step": 2905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8010000750056058e-05,
+      "loss": 0.7672,
+      "step": 2906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8008662079625045e-05,
+      "loss": 0.687,
+      "step": 2907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8007322991095856e-05,
+      "loss": 0.5088,
+      "step": 2908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8005983484511515e-05,
+      "loss": 0.471,
+      "step": 2909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8004643559915094e-05,
+      "loss": 1.6588,
+      "step": 2910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.800330321734964e-05,
+      "loss": 0.6019,
+      "step": 2911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8001962456858238e-05,
+      "loss": 0.923,
+      "step": 2912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.800062127848398e-05,
+      "loss": 0.4084,
+      "step": 2913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.799927968226997e-05,
+      "loss": 1.2018,
+      "step": 2914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.799793766825933e-05,
+      "loss": 0.4208,
+      "step": 2915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.799659523649519e-05,
+      "loss": 0.9311,
+      "step": 2916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7995252387020692e-05,
+      "loss": 1.0723,
+      "step": 2917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7993909119879005e-05,
+      "loss": 0.623,
+      "step": 2918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7992565435113294e-05,
+      "loss": 0.2622,
+      "step": 2919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7991221332766747e-05,
+      "loss": 0.4999,
+      "step": 2920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7989876812882563e-05,
+      "loss": 0.2131,
+      "step": 2921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7988531875503957e-05,
+      "loss": 0.2704,
+      "step": 2922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7987186520674148e-05,
+      "loss": 0.899,
+      "step": 2923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7985840748436386e-05,
+      "loss": 0.3157,
+      "step": 2924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.798449455883392e-05,
+      "loss": 1.4176,
+      "step": 2925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7983147951910014e-05,
+      "loss": 1.0373,
+      "step": 2926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7981800927707954e-05,
+      "loss": 0.1939,
+      "step": 2927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.798045348627102e-05,
+      "loss": 0.3996,
+      "step": 2928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7979105627642532e-05,
+      "loss": 0.7293,
+      "step": 2929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7977757351865808e-05,
+      "loss": 0.2632,
+      "step": 2930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7976408658984176e-05,
+      "loss": 0.3725,
+      "step": 2931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.797505954904099e-05,
+      "loss": 0.9546,
+      "step": 2932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7973710022079605e-05,
+      "loss": 0.7265,
+      "step": 2933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.79723600781434e-05,
+      "loss": 0.6257,
+      "step": 2934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7971009717275755e-05,
+      "loss": 0.6005,
+      "step": 2935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7969658939520074e-05,
+      "loss": 0.432,
+      "step": 2936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.796830774491977e-05,
+      "loss": 0.3075,
+      "step": 2937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.796695613351828e-05,
+      "loss": 0.1705,
+      "step": 2938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.796560410535903e-05,
+      "loss": 1.1235,
+      "step": 2939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7964251660485476e-05,
+      "loss": 1.3844,
+      "step": 2940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7962898798941094e-05,
+      "loss": 0.6851,
+      "step": 2941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7961545520769364e-05,
+      "loss": 1.3324,
+      "step": 2942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7960191826013776e-05,
+      "loss": 0.7048,
+      "step": 2943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.795883771471784e-05,
+      "loss": 0.3563,
+      "step": 2944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7957483186925075e-05,
+      "loss": 0.5842,
+      "step": 2945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7956128242679017e-05,
+      "loss": 0.5109,
+      "step": 2946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7954772882023213e-05,
+      "loss": 1.6017,
+      "step": 2947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.795341710500123e-05,
+      "loss": 0.2499,
+      "step": 2948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.795206091165663e-05,
+      "loss": 0.7471,
+      "step": 2949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7950704302033014e-05,
+      "loss": 0.8749,
+      "step": 2950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7949347276173973e-05,
+      "loss": 0.4119,
+      "step": 2951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7947989834123132e-05,
+      "loss": 0.0543,
+      "step": 2952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7946631975924114e-05,
+      "loss": 1.292,
+      "step": 2953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7945273701620556e-05,
+      "loss": 1.0672,
+      "step": 2954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.794391501125612e-05,
+      "loss": 0.593,
+      "step": 2955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7942555904874475e-05,
+      "loss": 0.6477,
+      "step": 2956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7941196382519295e-05,
+      "loss": 0.4343,
+      "step": 2957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7939836444234278e-05,
+      "loss": 0.7116,
+      "step": 2958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7938476090063134e-05,
+      "loss": 1.3062,
+      "step": 2959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.793711532004959e-05,
+      "loss": 0.2404,
+      "step": 2960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7935754134237373e-05,
+      "loss": 0.3949,
+      "step": 2961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7934392532670234e-05,
+      "loss": 0.3566,
+      "step": 2962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7933030515391936e-05,
+      "loss": 0.9161,
+      "step": 2963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.793166808244625e-05,
+      "loss": 0.5058,
+      "step": 2964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.793030523387697e-05,
+      "loss": 0.7066,
+      "step": 2965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7928941969727894e-05,
+      "loss": 0.0851,
+      "step": 2966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.792757829004284e-05,
+      "loss": 1.3701,
+      "step": 2967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.792621419486564e-05,
+      "loss": 0.1745,
+      "step": 2968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7924849684240128e-05,
+      "loss": 0.6923,
+      "step": 2969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7923484758210163e-05,
+      "loss": 0.3877,
+      "step": 2970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7922119416819614e-05,
+      "loss": 0.3731,
+      "step": 2971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7920753660112363e-05,
+      "loss": 0.6689,
+      "step": 2972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7919387488132304e-05,
+      "loss": 1.6879,
+      "step": 2973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.791802090092335e-05,
+      "loss": 0.5658,
+      "step": 2974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7916653898529418e-05,
+      "loss": 1.4029,
+      "step": 2975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7915286480994444e-05,
+      "loss": 0.8392,
+      "step": 2976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.791391864836238e-05,
+      "loss": 1.2605,
+      "step": 2977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7912550400677187e-05,
+      "loss": 0.8282,
+      "step": 2978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.791118173798284e-05,
+      "loss": 0.3886,
+      "step": 2979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7909812660323322e-05,
+      "loss": 0.4224,
+      "step": 2980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7908443167742647e-05,
+      "loss": 1.5723,
+      "step": 2981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7907073260284824e-05,
+      "loss": 0.4186,
+      "step": 2982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.790570293799388e-05,
+      "loss": 1.0459,
+      "step": 2983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7904332200913854e-05,
+      "loss": 1.0076,
+      "step": 2984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7902961049088816e-05,
+      "loss": 0.6591,
+      "step": 2985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7901589482562822e-05,
+      "loss": 0.7847,
+      "step": 2986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7900217501379954e-05,
+      "loss": 1.8828,
+      "step": 2987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7898845105584318e-05,
+      "loss": 0.4861,
+      "step": 2988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7897472295220014e-05,
+      "loss": 0.6979,
+      "step": 2989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7896099070331158e-05,
+      "loss": 0.7429,
+      "step": 2990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.78947254309619e-05,
+      "loss": 0.3614,
+      "step": 2991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7893351377156384e-05,
+      "loss": 0.8422,
+      "step": 2992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7891976908958765e-05,
+      "loss": 1.0012,
+      "step": 2993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7890602026413224e-05,
+      "loss": 1.0494,
+      "step": 2994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.788922672956395e-05,
+      "loss": 1.5924,
+      "step": 2995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7887851018455147e-05,
+      "loss": 0.5704,
+      "step": 2996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7886474893131025e-05,
+      "loss": 0.562,
+      "step": 2997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7885098353635815e-05,
+      "loss": 0.7203,
+      "step": 2998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.788372140001376e-05,
+      "loss": 0.3462,
+      "step": 2999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7882344032309116e-05,
+      "loss": 0.635,
+      "step": 3000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7880966250566146e-05,
+      "loss": 1.8176,
+      "step": 3001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.787958805482914e-05,
+      "loss": 0.4313,
+      "step": 3002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7878209445142385e-05,
+      "loss": 1.1275,
+      "step": 3003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7876830421550194e-05,
+      "loss": 0.6803,
+      "step": 3004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7875450984096892e-05,
+      "loss": 0.5487,
+      "step": 3005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.787407113282681e-05,
+      "loss": 0.3388,
+      "step": 3006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7872690867784297e-05,
+      "loss": 0.0599,
+      "step": 3007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.787131018901371e-05,
+      "loss": 0.5315,
+      "step": 3008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.786992909655943e-05,
+      "loss": 0.4914,
+      "step": 3009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.786854759046585e-05,
+      "loss": 0.5293,
+      "step": 3010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7867165670777355e-05,
+      "loss": 1.4949,
+      "step": 3011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7865783337538378e-05,
+      "loss": 1.2931,
+      "step": 3012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7864400590793335e-05,
+      "loss": 0.1093,
+      "step": 3013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.786301743058668e-05,
+      "loss": 1.262,
+      "step": 3014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7861633856962848e-05,
+      "loss": 0.356,
+      "step": 3015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.786024986996633e-05,
+      "loss": 0.4168,
+      "step": 3016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7858865469641584e-05,
+      "loss": 0.3498,
+      "step": 3017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7857480656033124e-05,
+      "loss": 0.863,
+      "step": 3018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.785609542918545e-05,
+      "loss": 0.5772,
+      "step": 3019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7854709789143084e-05,
+      "loss": 0.5455,
+      "step": 3020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.785332373595056e-05,
+      "loss": 0.6241,
+      "step": 3021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7851937269652422e-05,
+      "loss": 1.1195,
+      "step": 3022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.785055039029324e-05,
+      "loss": 0.5356,
+      "step": 3023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7849163097917578e-05,
+      "loss": 0.6379,
+      "step": 3024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7847775392570036e-05,
+      "loss": 1.3581,
+      "step": 3025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.78463872742952e-05,
+      "loss": 0.6983,
+      "step": 3026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.78449987431377e-05,
+      "loss": 0.7302,
+      "step": 3027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.784360979914215e-05,
+      "loss": 0.6519,
+      "step": 3028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.784222044235319e-05,
+      "loss": 0.5562,
+      "step": 3029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7840830672815483e-05,
+      "loss": 0.4286,
+      "step": 3030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7839440490573696e-05,
+      "loss": 1.1536,
+      "step": 3031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7838049895672507e-05,
+      "loss": 1.783,
+      "step": 3032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7836658888156603e-05,
+      "loss": 1.371,
+      "step": 3033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7835267468070702e-05,
+      "loss": 0.459,
+      "step": 3034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7833875635459512e-05,
+      "loss": 1.024,
+      "step": 3035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.783248339036778e-05,
+      "loss": 1.0723,
+      "step": 3036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.783109073284025e-05,
+      "loss": 0.6674,
+      "step": 3037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7829697662921667e-05,
+      "loss": 0.4762,
+      "step": 3038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7828304180656824e-05,
+      "loss": 0.6852,
+      "step": 3039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7826910286090494e-05,
+      "loss": 0.5824,
+      "step": 3040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7825515979267487e-05,
+      "loss": 0.6977,
+      "step": 3041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7824121260232605e-05,
+      "loss": 1.0607,
+      "step": 3042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7822726129030685e-05,
+      "loss": 0.1706,
+      "step": 3043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7821330585706555e-05,
+      "loss": 1.0636,
+      "step": 3044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7819934630305077e-05,
+      "loss": 0.5394,
+      "step": 3045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7818538262871117e-05,
+      "loss": 0.2106,
+      "step": 3046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.781714148344955e-05,
+      "loss": 0.5615,
+      "step": 3047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.781574429208527e-05,
+      "loss": 0.783,
+      "step": 3048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.781434668882318e-05,
+      "loss": 1.3069,
+      "step": 3049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7812948673708204e-05,
+      "loss": 1.024,
+      "step": 3050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.781155024678527e-05,
+      "loss": 1.2271,
+      "step": 3051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7810151408099328e-05,
+      "loss": 0.0445,
+      "step": 3052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7808752157695328e-05,
+      "loss": 1.12,
+      "step": 3053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.780735249561825e-05,
+      "loss": 0.863,
+      "step": 3054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7805952421913077e-05,
+      "loss": 0.1083,
+      "step": 3055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.780455193662481e-05,
+      "loss": 0.4051,
+      "step": 3056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7803151039798457e-05,
+      "loss": 1.2858,
+      "step": 3057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.780174973147904e-05,
+      "loss": 1.5231,
+      "step": 3058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7800348011711606e-05,
+      "loss": 1.523,
+      "step": 3059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7798945880541197e-05,
+      "loss": 0.1268,
+      "step": 3060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.779754333801288e-05,
+      "loss": 0.7543,
+      "step": 3061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7796140384171735e-05,
+      "loss": 0.3073,
+      "step": 3062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.779473701906285e-05,
+      "loss": 0.6564,
+      "step": 3063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7793333242731332e-05,
+      "loss": 0.5067,
+      "step": 3064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.77919290552223e-05,
+      "loss": 0.3231,
+      "step": 3065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7790524456580882e-05,
+      "loss": 0.5237,
+      "step": 3066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7789119446852218e-05,
+      "loss": 0.6078,
+      "step": 3067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7787714026081472e-05,
+      "loss": 0.5305,
+      "step": 3068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7786308194313807e-05,
+      "loss": 0.1859,
+      "step": 3069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.778490195159441e-05,
+      "loss": 1.2962,
+      "step": 3070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7783495297968483e-05,
+      "loss": 0.4507,
+      "step": 3071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7782088233481227e-05,
+      "loss": 0.6119,
+      "step": 3072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7780680758177868e-05,
+      "loss": 1.3108,
+      "step": 3073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7779272872103645e-05,
+      "loss": 1.2562,
+      "step": 3074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7777864575303804e-05,
+      "loss": 0.1886,
+      "step": 3075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.777645586782361e-05,
+      "loss": 0.3093,
+      "step": 3076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7775046749708336e-05,
+      "loss": 0.7409,
+      "step": 3077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7773637221003275e-05,
+      "loss": 0.4266,
+      "step": 3078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7772227281753728e-05,
+      "loss": 1.2109,
+      "step": 3079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.777081693200501e-05,
+      "loss": 0.8789,
+      "step": 3080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7769406171802444e-05,
+      "loss": 1.5144,
+      "step": 3081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.776799500119138e-05,
+      "loss": 0.9248,
+      "step": 3082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.776658342021717e-05,
+      "loss": 0.4626,
+      "step": 3083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7765171428925183e-05,
+      "loss": 1.0445,
+      "step": 3084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.77637590273608e-05,
+      "loss": 1.4894,
+      "step": 3085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7762346215569413e-05,
+      "loss": 0.5577,
+      "step": 3086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7760932993596428e-05,
+      "loss": 1.2446,
+      "step": 3087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7759519361487278e-05,
+      "loss": 1.0606,
+      "step": 3088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7758105319287387e-05,
+      "loss": 1.3403,
+      "step": 3089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.77566908670422e-05,
+      "loss": 0.1665,
+      "step": 3090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.775527600479718e-05,
+      "loss": 0.3025,
+      "step": 3091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7753860732597813e-05,
+      "loss": 1.2467,
+      "step": 3092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7752445050489562e-05,
+      "loss": 0.6395,
+      "step": 3093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7751028958517946e-05,
+      "loss": 0.6569,
+      "step": 3094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7749612456728475e-05,
+      "loss": 0.9808,
+      "step": 3095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7748195545166667e-05,
+      "loss": 1.0186,
+      "step": 3096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7746778223878072e-05,
+      "loss": 0.4846,
+      "step": 3097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7745360492908236e-05,
+      "loss": 0.5447,
+      "step": 3098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7743942352302726e-05,
+      "loss": 0.4705,
+      "step": 3099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7742523802107126e-05,
+      "loss": 0.3789,
+      "step": 3100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7741104842367017e-05,
+      "loss": 0.1346,
+      "step": 3101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7739685473128017e-05,
+      "loss": 0.0529,
+      "step": 3102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7738265694435736e-05,
+      "loss": 1.509,
+      "step": 3103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7736845506335806e-05,
+      "loss": 0.1168,
+      "step": 3104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7735424908873875e-05,
+      "loss": 0.7746,
+      "step": 3105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7734003902095605e-05,
+      "loss": 0.0678,
+      "step": 3106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.773258248604666e-05,
+      "loss": 0.4418,
+      "step": 3107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7731160660772723e-05,
+      "loss": 0.435,
+      "step": 3108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7729738426319496e-05,
+      "loss": 1.4075,
+      "step": 3109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7728315782732692e-05,
+      "loss": 0.3235,
+      "step": 3110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7726892730058028e-05,
+      "loss": 0.4773,
+      "step": 3111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7725469268341243e-05,
+      "loss": 0.3802,
+      "step": 3112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7724045397628095e-05,
+      "loss": 1.4552,
+      "step": 3113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7722621117964332e-05,
+      "loss": 0.8861,
+      "step": 3114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7721196429395747e-05,
+      "loss": 0.4285,
+      "step": 3115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7719771331968115e-05,
+      "loss": 0.2424,
+      "step": 3116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7718345825727248e-05,
+      "loss": 0.3982,
+      "step": 3117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.771691991071896e-05,
+      "loss": 1.4855,
+      "step": 3118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7715493586989072e-05,
+      "loss": 0.992,
+      "step": 3119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7714066854583443e-05,
+      "loss": 1.4184,
+      "step": 3120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7712639713547907e-05,
+      "loss": 1.5479,
+      "step": 3121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.771121216392835e-05,
+      "loss": 0.2648,
+      "step": 3122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7709784205770642e-05,
+      "loss": 0.7784,
+      "step": 3123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.770835583912068e-05,
+      "loss": 0.515,
+      "step": 3124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7706927064024382e-05,
+      "loss": 0.5121,
+      "step": 3125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7705497880527655e-05,
+      "loss": 0.5695,
+      "step": 3126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7704068288676442e-05,
+      "loss": 0.6906,
+      "step": 3127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7702638288516683e-05,
+      "loss": 0.7945,
+      "step": 3128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7701207880094338e-05,
+      "loss": 1.5193,
+      "step": 3129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.769977706345539e-05,
+      "loss": 0.4742,
+      "step": 3130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7698345838645815e-05,
+      "loss": 1.2808,
+      "step": 3131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.769691420571162e-05,
+      "loss": 0.9139,
+      "step": 3132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7695482164698814e-05,
+      "loss": 2.2273,
+      "step": 3133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7694049715653424e-05,
+      "loss": 0.824,
+      "step": 3134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7692616858621486e-05,
+      "loss": 0.5464,
+      "step": 3135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.769118359364906e-05,
+      "loss": 0.3758,
+      "step": 3136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.76897499207822e-05,
+      "loss": 0.6523,
+      "step": 3137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.768831584006699e-05,
+      "loss": 0.1215,
+      "step": 3138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7686881351549517e-05,
+      "loss": 0.7051,
+      "step": 3139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.76854464552759e-05,
+      "loss": 0.4059,
+      "step": 3140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7684011151292232e-05,
+      "loss": 1.0312,
+      "step": 3141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.768257543964467e-05,
+      "loss": 0.2244,
+      "step": 3142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7681139320379335e-05,
+      "loss": 1.2941,
+      "step": 3143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7679702793542397e-05,
+      "loss": 0.8751,
+      "step": 3144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7678265859180026e-05,
+      "loss": 0.2285,
+      "step": 3145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7676828517338406e-05,
+      "loss": 0.2021,
+      "step": 3146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7675390768063722e-05,
+      "loss": 1.076,
+      "step": 3147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.767395261140219e-05,
+      "loss": 0.7726,
+      "step": 3148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7672514047400036e-05,
+      "loss": 0.887,
+      "step": 3149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7671075076103493e-05,
+      "loss": 1.0369,
+      "step": 3150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7669635697558807e-05,
+      "loss": 0.4402,
+      "step": 3151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.766819591181224e-05,
+      "loss": 1.3595,
+      "step": 3152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.766675571891007e-05,
+      "loss": 0.4069,
+      "step": 3153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7665315118898583e-05,
+      "loss": 0.2843,
+      "step": 3154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.766387411182408e-05,
+      "loss": 0.5638,
+      "step": 3155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7662432697732872e-05,
+      "loss": 0.4132,
+      "step": 3156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7660990876671292e-05,
+      "loss": 0.4933,
+      "step": 3157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7659548648685674e-05,
+      "loss": 1.3538,
+      "step": 3158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7658106013822373e-05,
+      "loss": 0.5673,
+      "step": 3159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.765666297212776e-05,
+      "loss": 0.3371,
+      "step": 3160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7655219523648207e-05,
+      "loss": 0.4345,
+      "step": 3161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7653775668430112e-05,
+      "loss": 0.354,
+      "step": 3162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.765233140651988e-05,
+      "loss": 0.236,
+      "step": 3163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7650886737963925e-05,
+      "loss": 0.3988,
+      "step": 3164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7649441662808677e-05,
+      "loss": 0.3269,
+      "step": 3165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7647996181100593e-05,
+      "loss": 0.8494,
+      "step": 3166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7646550292886118e-05,
+      "loss": 1.1874,
+      "step": 3167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.764510399821173e-05,
+      "loss": 1.2494,
+      "step": 3168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7643657297123915e-05,
+      "loss": 0.9732,
+      "step": 3169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.764221018966916e-05,
+      "loss": 0.3613,
+      "step": 3170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.764076267589398e-05,
+      "loss": 1.5017,
+      "step": 3171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7639314755844904e-05,
+      "loss": 0.5357,
+      "step": 3172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.763786642956846e-05,
+      "loss": 0.8609,
+      "step": 3173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7636417697111197e-05,
+      "loss": 0.6417,
+      "step": 3174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7634968558519683e-05,
+      "loss": 0.5814,
+      "step": 3175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7633519013840492e-05,
+      "loss": 0.9677,
+      "step": 3176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7632069063120212e-05,
+      "loss": 0.7347,
+      "step": 3177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7630618706405436e-05,
+      "loss": 0.6916,
+      "step": 3178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7629167943742794e-05,
+      "loss": 0.5227,
+      "step": 3179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.76277167751789e-05,
+      "loss": 0.9396,
+      "step": 3180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7626265200760403e-05,
+      "loss": 0.4152,
+      "step": 3181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.762481322053395e-05,
+      "loss": 1.2819,
+      "step": 3182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7623360834546216e-05,
+      "loss": 1.407,
+      "step": 3183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7621908042843874e-05,
+      "loss": 0.4152,
+      "step": 3184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7620454845473616e-05,
+      "loss": 0.4669,
+      "step": 3185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.761900124248215e-05,
+      "loss": 0.5565,
+      "step": 3186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7617547233916195e-05,
+      "loss": 0.6908,
+      "step": 3187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7616092819822484e-05,
+      "loss": 1.2223,
+      "step": 3188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.761463800024776e-05,
+      "loss": 1.3828,
+      "step": 3189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.761318277523878e-05,
+      "loss": 1.0368,
+      "step": 3190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7611727144842315e-05,
+      "loss": 0.8783,
+      "step": 3191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7610271109105155e-05,
+      "loss": 0.273,
+      "step": 3192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.760881466807409e-05,
+      "loss": 0.5983,
+      "step": 3193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7607357821795934e-05,
+      "loss": 0.9972,
+      "step": 3194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7605900570317505e-05,
+      "loss": 0.9154,
+      "step": 3195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7604442913685646e-05,
+      "loss": 0.6235,
+      "step": 3196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7602984851947198e-05,
+      "loss": 0.5225,
+      "step": 3197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.760152638514903e-05,
+      "loss": 0.7364,
+      "step": 3198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7600067513338012e-05,
+      "loss": 0.6614,
+      "step": 3199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.759860823656104e-05,
+      "loss": 0.2441,
+      "step": 3200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7597148554865004e-05,
+      "loss": 1.3158,
+      "step": 3201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7595688468296823e-05,
+      "loss": 1.5739,
+      "step": 3202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7594227976903435e-05,
+      "loss": 0.3518,
+      "step": 3203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.759276708073176e-05,
+      "loss": 0.6597,
+      "step": 3204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7591305779828765e-05,
+      "loss": 0.4458,
+      "step": 3205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7589844074241413e-05,
+      "loss": 0.6295,
+      "step": 3206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.758838196401668e-05,
+      "loss": 0.0933,
+      "step": 3207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7586919449201562e-05,
+      "loss": 0.1603,
+      "step": 3208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7585456529843068e-05,
+      "loss": 0.463,
+      "step": 3209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7583993205988207e-05,
+      "loss": 0.7831,
+      "step": 3210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7582529477684012e-05,
+      "loss": 0.0555,
+      "step": 3211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7581065344977533e-05,
+      "loss": 0.4675,
+      "step": 3212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7579600807915827e-05,
+      "loss": 0.5616,
+      "step": 3213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7578135866545956e-05,
+      "loss": 0.4159,
+      "step": 3214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7576670520915008e-05,
+      "loss": 0.1814,
+      "step": 3215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.757520477107008e-05,
+      "loss": 0.241,
+      "step": 3216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7573738617058285e-05,
+      "loss": 0.9637,
+      "step": 3217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7572272058926737e-05,
+      "loss": 0.6304,
+      "step": 3218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7570805096722578e-05,
+      "loss": 0.6853,
+      "step": 3219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7569337730492952e-05,
+      "loss": 0.2844,
+      "step": 3220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.756786996028502e-05,
+      "loss": 0.6982,
+      "step": 3221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7566401786145956e-05,
+      "loss": 0.6887,
+      "step": 3222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7564933208122952e-05,
+      "loss": 1.664,
+      "step": 3223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7563464226263205e-05,
+      "loss": 1.5652,
+      "step": 3224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7561994840613923e-05,
+      "loss": 0.6033,
+      "step": 3225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7560525051222342e-05,
+      "loss": 0.7462,
+      "step": 3226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7559054858135692e-05,
+      "loss": 0.7181,
+      "step": 3227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.755758426140123e-05,
+      "loss": 1.0443,
+      "step": 3228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7556113261066218e-05,
+      "loss": 0.7987,
+      "step": 3229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7554641857177936e-05,
+      "loss": 1.447,
+      "step": 3230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7553170049783677e-05,
+      "loss": 0.3496,
+      "step": 3231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.755169783893074e-05,
+      "loss": 0.5857,
+      "step": 3232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7550225224666444e-05,
+      "loss": 0.9702,
+      "step": 3233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.754875220703812e-05,
+      "loss": 0.2911,
+      "step": 3234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7547278786093107e-05,
+      "loss": 1.0665,
+      "step": 3235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.754580496187877e-05,
+      "loss": 0.1142,
+      "step": 3236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7544330734442464e-05,
+      "loss": 0.2891,
+      "step": 3237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7542856103831585e-05,
+      "loss": 1.148,
+      "step": 3238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7541381070093513e-05,
+      "loss": 1.2069,
+      "step": 3239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7539905633275663e-05,
+      "loss": 0.2043,
+      "step": 3240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7538429793425462e-05,
+      "loss": 0.8662,
+      "step": 3241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7536953550590334e-05,
+      "loss": 0.8856,
+      "step": 3242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7535476904817723e-05,
+      "loss": 0.9688,
+      "step": 3243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.75339998561551e-05,
+      "loss": 0.5455,
+      "step": 3244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7532522404649927e-05,
+      "loss": 0.4229,
+      "step": 3245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7531044550349695e-05,
+      "loss": 0.6138,
+      "step": 3246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7529566293301904e-05,
+      "loss": 0.6032,
+      "step": 3247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7528087633554057e-05,
+      "loss": 0.9253,
+      "step": 3248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7526608571153683e-05,
+      "loss": 0.6024,
+      "step": 3249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.752512910614832e-05,
+      "loss": 0.5752,
+      "step": 3250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7523649238585518e-05,
+      "loss": 1.8115,
+      "step": 3251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7522168968512835e-05,
+      "loss": 0.2848,
+      "step": 3252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7520688295977855e-05,
+      "loss": 0.3016,
+      "step": 3253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7519207221028157e-05,
+      "loss": 1.5374,
+      "step": 3254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.751772574371135e-05,
+      "loss": 1.1749,
+      "step": 3255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7516243864075047e-05,
+      "loss": 0.6341,
+      "step": 3256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.751476158216688e-05,
+      "loss": 0.4757,
+      "step": 3257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7513278898034478e-05,
+      "loss": 0.1499,
+      "step": 3258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7511795811725506e-05,
+      "loss": 0.7692,
+      "step": 3259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.751031232328762e-05,
+      "loss": 0.831,
+      "step": 3260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.750882843276851e-05,
+      "loss": 0.2552,
+      "step": 3261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.750734414021586e-05,
+      "loss": 0.5459,
+      "step": 3262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.750585944567738e-05,
+      "loss": 0.6744,
+      "step": 3263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7504374349200787e-05,
+      "loss": 0.3512,
+      "step": 3264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7502888850833812e-05,
+      "loss": 0.2917,
+      "step": 3265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.75014029506242e-05,
+      "loss": 0.7867,
+      "step": 3266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.74999166486197e-05,
+      "loss": 0.4978,
+      "step": 3267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7498429944868092e-05,
+      "loss": 0.8996,
+      "step": 3268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.749694283941715e-05,
+      "loss": 0.8913,
+      "step": 3269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7495455332314676e-05,
+      "loss": 0.2913,
+      "step": 3270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.749396742360848e-05,
+      "loss": 1.8164,
+      "step": 3271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7492479113346375e-05,
+      "loss": 1.7988,
+      "step": 3272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.74909904015762e-05,
+      "loss": 0.239,
+      "step": 3273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7489501288345805e-05,
+      "loss": 0.7672,
+      "step": 3274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7488011773703044e-05,
+      "loss": 0.279,
+      "step": 3275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7486521857695794e-05,
+      "loss": 1.6997,
+      "step": 3276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7485031540371943e-05,
+      "loss": 0.2696,
+      "step": 3277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7483540821779386e-05,
+      "loss": 0.4583,
+      "step": 3278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7482049701966035e-05,
+      "loss": 1.1996,
+      "step": 3279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.748055818097981e-05,
+      "loss": 0.5711,
+      "step": 3280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.747906625886866e-05,
+      "loss": 0.8223,
+      "step": 3281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7477573935680528e-05,
+      "loss": 0.4583,
+      "step": 3282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7476081211463375e-05,
+      "loss": 0.8547,
+      "step": 3283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7474588086265183e-05,
+      "loss": 1.5124,
+      "step": 3284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7473094560133936e-05,
+      "loss": 0.2149,
+      "step": 3285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7471600633117642e-05,
+      "loss": 1.6688,
+      "step": 3286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.747010630526431e-05,
+      "loss": 1.1247,
+      "step": 3287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7468611576621968e-05,
+      "loss": 0.6208,
+      "step": 3288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.746711644723866e-05,
+      "loss": 0.3757,
+      "step": 3289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.746562091716244e-05,
+      "loss": 0.2234,
+      "step": 3290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7464124986441366e-05,
+      "loss": 1.4114,
+      "step": 3291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7462628655123525e-05,
+      "loss": 0.9157,
+      "step": 3292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.746113192325701e-05,
+      "loss": 1.0825,
+      "step": 3293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.745963479088992e-05,
+      "loss": 0.7448,
+      "step": 3294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7458137258070378e-05,
+      "loss": 0.7909,
+      "step": 3295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7456639324846514e-05,
+      "loss": 0.1745,
+      "step": 3296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.745514099126647e-05,
+      "loss": 0.7637,
+      "step": 3297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7453642257378397e-05,
+      "loss": 0.6551,
+      "step": 3298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7452143123230475e-05,
+      "loss": 0.4174,
+      "step": 3299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.745064358887088e-05,
+      "loss": 0.5727,
+      "step": 3300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7449143654347807e-05,
+      "loss": 0.5019,
+      "step": 3301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7447643319709464e-05,
+      "loss": 0.1171,
+      "step": 3302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7446142585004077e-05,
+      "loss": 0.2376,
+      "step": 3303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7444641450279872e-05,
+      "loss": 0.5027,
+      "step": 3304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.74431399155851e-05,
+      "loss": 0.2421,
+      "step": 3305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.744163798096802e-05,
+      "loss": 0.1384,
+      "step": 3306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.74401356464769e-05,
+      "loss": 0.3784,
+      "step": 3307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.743863291216003e-05,
+      "loss": 1.2719,
+      "step": 3308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7437129778065703e-05,
+      "loss": 0.5457,
+      "step": 3309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7435626244242235e-05,
+      "loss": 1.13,
+      "step": 3310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.743412231073795e-05,
+      "loss": 0.0427,
+      "step": 3311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7432617977601174e-05,
+      "loss": 1.4069,
+      "step": 3312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.743111324488027e-05,
+      "loss": 1.1078,
+      "step": 3313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7429608112623596e-05,
+      "loss": 0.967,
+      "step": 3314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.742810258087952e-05,
+      "loss": 0.2509,
+      "step": 3315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7426596649696434e-05,
+      "loss": 1.2931,
+      "step": 3316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.742509031912274e-05,
+      "loss": 1.3752,
+      "step": 3317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7423583589206855e-05,
+      "loss": 0.4172,
+      "step": 3318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7422076459997197e-05,
+      "loss": 0.7776,
+      "step": 3319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.742056893154221e-05,
+      "loss": 0.3428,
+      "step": 3320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7419061003890344e-05,
+      "loss": 1.2059,
+      "step": 3321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7417552677090066e-05,
+      "loss": 1.311,
+      "step": 3322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7416043951189852e-05,
+      "loss": 0.9804,
+      "step": 3323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7414534826238196e-05,
+      "loss": 0.8105,
+      "step": 3324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7413025302283594e-05,
+      "loss": 0.4209,
+      "step": 3325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7411515379374566e-05,
+      "loss": 1.6393,
+      "step": 3326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7410005057559643e-05,
+      "loss": 1.2428,
+      "step": 3327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7408494336887365e-05,
+      "loss": 0.8342,
+      "step": 3328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7406983217406284e-05,
+      "loss": 0.5609,
+      "step": 3329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.740547169916497e-05,
+      "loss": 0.771,
+      "step": 3330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7403959782212008e-05,
+      "loss": 1.018,
+      "step": 3331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7402447466595975e-05,
+      "loss": 1.6077,
+      "step": 3332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.74009347523655e-05,
+      "loss": 0.2117,
+      "step": 3333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.739942163956918e-05,
+      "loss": 0.5397,
+      "step": 3334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7397908128255658e-05,
+      "loss": 0.6128,
+      "step": 3335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.739639421847358e-05,
+      "loss": 0.4518,
+      "step": 3336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7394879910271594e-05,
+      "loss": 1.4155,
+      "step": 3337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7393365203698378e-05,
+      "loss": 1.3752,
+      "step": 3338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7391850098802606e-05,
+      "loss": 0.1898,
+      "step": 3339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7390334595632986e-05,
+      "loss": 0.6759,
+      "step": 3340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7388818694238216e-05,
+      "loss": 0.1002,
+      "step": 3341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.738730239466702e-05,
+      "loss": 1.0092,
+      "step": 3342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.738578569696813e-05,
+      "loss": 0.6546,
+      "step": 3343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7384268601190298e-05,
+      "loss": 1.34,
+      "step": 3344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7382751107382274e-05,
+      "loss": 0.706,
+      "step": 3345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7381233215592845e-05,
+      "loss": 0.8713,
+      "step": 3346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.737971492587078e-05,
+      "loss": 0.547,
+      "step": 3347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7378196238264887e-05,
+      "loss": 0.5431,
+      "step": 3348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7376677152823973e-05,
+      "loss": 0.3168,
+      "step": 3349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7375157669596862e-05,
+      "loss": 0.3083,
+      "step": 3350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.737363778863239e-05,
+      "loss": 0.267,
+      "step": 3351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7372117509979406e-05,
+      "loss": 0.2551,
+      "step": 3352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7370596833686775e-05,
+      "loss": 0.0633,
+      "step": 3353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7369075759803367e-05,
+      "loss": 1.0779,
+      "step": 3354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.736755428837807e-05,
+      "loss": 1.7885,
+      "step": 3355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.736603241945978e-05,
+      "loss": 0.4452,
+      "step": 3356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7364510153097423e-05,
+      "loss": 1.0724,
+      "step": 3357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7362987489339912e-05,
+      "loss": 0.5584,
+      "step": 3358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.736146442823619e-05,
+      "loss": 0.4384,
+      "step": 3359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.735994096983521e-05,
+      "loss": 0.2176,
+      "step": 3360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.735841711418593e-05,
+      "loss": 0.7928,
+      "step": 3361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7356892861337332e-05,
+      "loss": 0.1775,
+      "step": 3362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7355368211338405e-05,
+      "loss": 0.5677,
+      "step": 3363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7353843164238147e-05,
+      "loss": 0.6095,
+      "step": 3364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7352317720085577e-05,
+      "loss": 0.58,
+      "step": 3365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7350791878929722e-05,
+      "loss": 1.1101,
+      "step": 3366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7349265640819625e-05,
+      "loss": 1.5901,
+      "step": 3367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7347739005804333e-05,
+      "loss": 1.2181,
+      "step": 3368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.734621197393292e-05,
+      "loss": 0.5669,
+      "step": 3369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7344684545254453e-05,
+      "loss": 1.5628,
+      "step": 3370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7343156719818037e-05,
+      "loss": 0.4028,
+      "step": 3371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7341628497672767e-05,
+      "loss": 0.7448,
+      "step": 3372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7340099878867766e-05,
+      "loss": 1.0912,
+      "step": 3373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.733857086345216e-05,
+      "loss": 0.3554,
+      "step": 3374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7337041451475095e-05,
+      "loss": 0.1601,
+      "step": 3375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.733551164298572e-05,
+      "loss": 0.3726,
+      "step": 3376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7333981438033203e-05,
+      "loss": 0.8085,
+      "step": 3377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.733245083666674e-05,
+      "loss": 0.2961,
+      "step": 3378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7330919838935504e-05,
+      "loss": 0.4586,
+      "step": 3379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7329388444888714e-05,
+      "loss": 0.1109,
+      "step": 3380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7327856654575585e-05,
+      "loss": 0.5821,
+      "step": 3381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7326324468045347e-05,
+      "loss": 0.3037,
+      "step": 3382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7324791885347248e-05,
+      "loss": 1.2202,
+      "step": 3383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7323258906530545e-05,
+      "loss": 0.3202,
+      "step": 3384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7321725531644507e-05,
+      "loss": 0.5994,
+      "step": 3385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.732019176073841e-05,
+      "loss": 0.2051,
+      "step": 3386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.731865759386156e-05,
+      "loss": 0.2701,
+      "step": 3387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.731712303106326e-05,
+      "loss": 0.8494,
+      "step": 3388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7315588072392833e-05,
+      "loss": 0.1487,
+      "step": 3389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.731405271789961e-05,
+      "loss": 0.9613,
+      "step": 3390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.731251696763294e-05,
+      "loss": 0.8188,
+      "step": 3391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7310980821642178e-05,
+      "loss": 0.5742,
+      "step": 3392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.73094442799767e-05,
+      "loss": 0.5775,
+      "step": 3393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.730790734268589e-05,
+      "loss": 0.791,
+      "step": 3394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.730637000981914e-05,
+      "loss": 0.6055,
+      "step": 3395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.730483228142586e-05,
+      "loss": 1.1699,
+      "step": 3396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.730329415755548e-05,
+      "loss": 0.3139,
+      "step": 3397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7301755638257436e-05,
+      "loss": 0.4689,
+      "step": 3398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7300216723581165e-05,
+      "loss": 0.3261,
+      "step": 3399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7298677413576136e-05,
+      "loss": 0.4239,
+      "step": 3400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.729713770829182e-05,
+      "loss": 0.1538,
+      "step": 3401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7295597607777702e-05,
+      "loss": 0.3561,
+      "step": 3402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7294057112083288e-05,
+      "loss": 1.4303,
+      "step": 3403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7292516221258078e-05,
+      "loss": 0.2943,
+      "step": 3404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.72909749353516e-05,
+      "loss": 0.4399,
+      "step": 3405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.72894332544134e-05,
+      "loss": 0.3858,
+      "step": 3406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.728789117849302e-05,
+      "loss": 0.2513,
+      "step": 3407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7286348707640016e-05,
+      "loss": 1.0503,
+      "step": 3408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7284805841903973e-05,
+      "loss": 0.4664,
+      "step": 3409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7283262581334476e-05,
+      "loss": 2.3905,
+      "step": 3410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7281718925981127e-05,
+      "loss": 0.6354,
+      "step": 3411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7280174875893534e-05,
+      "loss": 0.8076,
+      "step": 3412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7278630431121327e-05,
+      "loss": 0.5562,
+      "step": 3413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7277085591714143e-05,
+      "loss": 0.2869,
+      "step": 3414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7275540357721634e-05,
+      "loss": 0.3675,
+      "step": 3415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.727399472919346e-05,
+      "loss": 0.5376,
+      "step": 3416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.72724487061793e-05,
+      "loss": 0.598,
+      "step": 3417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7270902288728848e-05,
+      "loss": 1.1153,
+      "step": 3418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7269355476891798e-05,
+      "loss": 0.4111,
+      "step": 3419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.726780827071787e-05,
+      "loss": 0.1338,
+      "step": 3420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.726626067025679e-05,
+      "loss": 0.4678,
+      "step": 3421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.726471267555829e-05,
+      "loss": 0.1543,
+      "step": 3422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7263164286672136e-05,
+      "loss": 1.0702,
+      "step": 3423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7261615503648085e-05,
+      "loss": 0.6026,
+      "step": 3424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7260066326535918e-05,
+      "loss": 0.6129,
+      "step": 3425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.725851675538542e-05,
+      "loss": 0.7191,
+      "step": 3426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7256966790246402e-05,
+      "loss": 0.2667,
+      "step": 3427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7255416431168675e-05,
+      "loss": 0.1195,
+      "step": 3428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7253865678202067e-05,
+      "loss": 1.6963,
+      "step": 3429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7252314531396424e-05,
+      "loss": 0.8212,
+      "step": 3430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7250762990801597e-05,
+      "loss": 0.9465,
+      "step": 3431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7249211056467448e-05,
+      "loss": 0.5482,
+      "step": 3432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7247658728443864e-05,
+      "loss": 0.7713,
+      "step": 3433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7246106006780732e-05,
+      "loss": 0.7094,
+      "step": 3434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7244552891527956e-05,
+      "loss": 0.5822,
+      "step": 3435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7242999382735453e-05,
+      "loss": 1.7273,
+      "step": 3436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7241445480453157e-05,
+      "loss": 0.3199,
+      "step": 3437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.723989118473101e-05,
+      "loss": 0.1512,
+      "step": 3438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7238336495618962e-05,
+      "loss": 0.5762,
+      "step": 3439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7236781413166982e-05,
+      "loss": 0.2607,
+      "step": 3440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7235225937425056e-05,
+      "loss": 1.1068,
+      "step": 3441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7233670068443167e-05,
+      "loss": 0.8571,
+      "step": 3442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.723211380627133e-05,
+      "loss": 0.5544,
+      "step": 3443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.723055715095956e-05,
+      "loss": 0.6637,
+      "step": 3444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.722900010255789e-05,
+      "loss": 0.7175,
+      "step": 3445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7227442661116355e-05,
+      "loss": 0.4472,
+      "step": 3446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.722588482668502e-05,
+      "loss": 0.6256,
+      "step": 3447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7224326599313948e-05,
+      "loss": 1.148,
+      "step": 3448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7222767979053226e-05,
+      "loss": 0.1713,
+      "step": 3449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7221208965952946e-05,
+      "loss": 0.3265,
+      "step": 3450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7219649560063213e-05,
+      "loss": 1.6536,
+      "step": 3451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7218089761434145e-05,
+      "loss": 1.4645,
+      "step": 3452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.721652957011588e-05,
+      "loss": 1.9692,
+      "step": 3453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7214968986158558e-05,
+      "loss": 0.0402,
+      "step": 3454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7213408009612338e-05,
+      "loss": 1.6948,
+      "step": 3455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7211846640527384e-05,
+      "loss": 0.4197,
+      "step": 3456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7210284878953888e-05,
+      "loss": 0.5185,
+      "step": 3457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7208722724942034e-05,
+      "loss": 1.0664,
+      "step": 3458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.720716017854204e-05,
+      "loss": 0.2388,
+      "step": 3459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7205597239804122e-05,
+      "loss": 0.4775,
+      "step": 3460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7204033908778514e-05,
+      "loss": 0.1229,
+      "step": 3461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7202470185515456e-05,
+      "loss": 1.2698,
+      "step": 3462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7200906070065212e-05,
+      "loss": 1.218,
+      "step": 3463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7199341562478047e-05,
+      "loss": 0.6306,
+      "step": 3464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7197776662804254e-05,
+      "loss": 0.6624,
+      "step": 3465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.719621137109412e-05,
+      "loss": 1.2483,
+      "step": 3466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7194645687397955e-05,
+      "loss": 0.376,
+      "step": 3467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.719307961176608e-05,
+      "loss": 0.7651,
+      "step": 3468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7191513144248838e-05,
+      "loss": 0.2726,
+      "step": 3469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7189946284896558e-05,
+      "loss": 1.6639,
+      "step": 3470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7188379033759614e-05,
+      "loss": 0.8013,
+      "step": 3471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.718681139088837e-05,
+      "loss": 0.7893,
+      "step": 3472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7185243356333208e-05,
+      "loss": 0.5364,
+      "step": 3473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7183674930144532e-05,
+      "loss": 1.2606,
+      "step": 3474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7182106112372742e-05,
+      "loss": 0.5346,
+      "step": 3475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.718053690306827e-05,
+      "loss": 1.2242,
+      "step": 3476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7178967302281544e-05,
+      "loss": 1.0211,
+      "step": 3477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7177397310063014e-05,
+      "loss": 0.5855,
+      "step": 3478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7175826926463136e-05,
+      "loss": 0.3808,
+      "step": 3479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7174256151532383e-05,
+      "loss": 0.464,
+      "step": 3480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7172684985321245e-05,
+      "loss": 0.5238,
+      "step": 3481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.717111342788021e-05,
+      "loss": 0.2235,
+      "step": 3482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7169541479259797e-05,
+      "loss": 0.4078,
+      "step": 3483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.716796913951052e-05,
+      "loss": 0.7605,
+      "step": 3484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7166396408682918e-05,
+      "loss": 0.4645,
+      "step": 3485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7164823286827546e-05,
+      "loss": 0.3381,
+      "step": 3486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7163249773994948e-05,
+      "loss": 1.2476,
+      "step": 3487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.716167587023571e-05,
+      "loss": 0.476,
+      "step": 3488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.716010157560041e-05,
+      "loss": 0.9733,
+      "step": 3489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7158526890139652e-05,
+      "loss": 0.7808,
+      "step": 3490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7156951813904042e-05,
+      "loss": 0.9645,
+      "step": 3491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.71553763469442e-05,
+      "loss": 0.9575,
+      "step": 3492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7153800489310765e-05,
+      "loss": 0.1862,
+      "step": 3493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7152224241054392e-05,
+      "loss": 0.4968,
+      "step": 3494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.715064760222573e-05,
+      "loss": 0.929,
+      "step": 3495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.714907057287546e-05,
+      "loss": 1.0061,
+      "step": 3496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.714749315305426e-05,
+      "loss": 0.9241,
+      "step": 3497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7145915342812836e-05,
+      "loss": 0.8279,
+      "step": 3498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7144337142201895e-05,
+      "loss": 0.5063,
+      "step": 3499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.714275855127216e-05,
+      "loss": 0.3798,
+      "step": 3500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7141179570074373e-05,
+      "loss": 1.3977,
+      "step": 3501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7139600198659272e-05,
+      "loss": 1.3389,
+      "step": 3502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7138020437077625e-05,
+      "loss": 0.5697,
+      "step": 3503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7136440285380205e-05,
+      "loss": 0.2786,
+      "step": 3504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7134859743617795e-05,
+      "loss": 1.5853,
+      "step": 3505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7133278811841198e-05,
+      "loss": 0.9751,
+      "step": 3506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.713169749010122e-05,
+      "loss": 0.4387,
+      "step": 3507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7130115778448694e-05,
+      "loss": 0.0286,
+      "step": 3508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7128533676934437e-05,
+      "loss": 0.8273,
+      "step": 3509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7126951185609322e-05,
+      "loss": 0.3626,
+      "step": 3510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.712536830452419e-05,
+      "loss": 0.2132,
+      "step": 3511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7123785033729928e-05,
+      "loss": 0.4033,
+      "step": 3512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7122201373277417e-05,
+      "loss": 0.969,
+      "step": 3513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7120617323217556e-05,
+      "loss": 0.781,
+      "step": 3514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7119032883601257e-05,
+      "loss": 1.2832,
+      "step": 3515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7117448054479445e-05,
+      "loss": 0.4026,
+      "step": 3516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.711586283590305e-05,
+      "loss": 0.7258,
+      "step": 3517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.711427722792303e-05,
+      "loss": 0.2986,
+      "step": 3518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7112691230590338e-05,
+      "loss": 0.5391,
+      "step": 3519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7111104843955957e-05,
+      "loss": 0.4041,
+      "step": 3520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.710951806807087e-05,
+      "loss": 0.3167,
+      "step": 3521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7107930902986067e-05,
+      "loss": 0.2755,
+      "step": 3522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7106343348752574e-05,
+      "loss": 0.5734,
+      "step": 3523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7104755405421402e-05,
+      "loss": 1.6802,
+      "step": 3524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.71031670730436e-05,
+      "loss": 0.7807,
+      "step": 3525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.710157835167021e-05,
+      "loss": 0.4086,
+      "step": 3526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.709998924135229e-05,
+      "loss": 0.9833,
+      "step": 3527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7098399742140916e-05,
+      "loss": 0.1951,
+      "step": 3528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7096809854087182e-05,
+      "loss": 1.128,
+      "step": 3529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7095219577242176e-05,
+      "loss": 1.2873,
+      "step": 3530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.709362891165702e-05,
+      "loss": 0.5388,
+      "step": 3531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.709203785738283e-05,
+      "loss": 0.9188,
+      "step": 3532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7090446414470744e-05,
+      "loss": 0.5623,
+      "step": 3533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7088854582971915e-05,
+      "loss": 0.6193,
+      "step": 3534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7087262362937504e-05,
+      "loss": 0.7978,
+      "step": 3535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7085669754418676e-05,
+      "loss": 0.5057,
+      "step": 3536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7084076757466628e-05,
+      "loss": 0.1143,
+      "step": 3537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7082483372132555e-05,
+      "loss": 1.171,
+      "step": 3538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7080889598467666e-05,
+      "loss": 0.2968,
+      "step": 3539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.707929543652319e-05,
+      "loss": 1.0027,
+      "step": 3540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7077700886350357e-05,
+      "loss": 0.5791,
+      "step": 3541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7076105948000423e-05,
+      "loss": 0.8345,
+      "step": 3542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7074510621524642e-05,
+      "loss": 0.7386,
+      "step": 3543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7072914906974292e-05,
+      "loss": 0.5015,
+      "step": 3544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7071318804400663e-05,
+      "loss": 1.2676,
+      "step": 3545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.706972231385504e-05,
+      "loss": 0.7035,
+      "step": 3546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7068125435388753e-05,
+      "loss": 0.325,
+      "step": 3547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.706652816905311e-05,
+      "loss": 1.1495,
+      "step": 3548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7064930514899458e-05,
+      "loss": 0.6886,
+      "step": 3549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.706333247297914e-05,
+      "loss": 0.3122,
+      "step": 3550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7061734043343515e-05,
+      "loss": 0.0285,
+      "step": 3551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7060135226043957e-05,
+      "loss": 0.2506,
+      "step": 3552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.705853602113186e-05,
+      "loss": 0.1698,
+      "step": 3553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7056936428658614e-05,
+      "loss": 0.9256,
+      "step": 3554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7055336448675632e-05,
+      "loss": 1.6854,
+      "step": 3555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7053736081234336e-05,
+      "loss": 0.4849,
+      "step": 3556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.705213532638617e-05,
+      "loss": 1.1644,
+      "step": 3557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7050534184182575e-05,
+      "loss": 1.0303,
+      "step": 3558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7048932654675006e-05,
+      "loss": 1.5267,
+      "step": 3559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7047330737914947e-05,
+      "loss": 0.6116,
+      "step": 3560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.704572843395388e-05,
+      "loss": 0.6106,
+      "step": 3561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7044125742843296e-05,
+      "loss": 0.617,
+      "step": 3562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.704252266463472e-05,
+      "loss": 1.3504,
+      "step": 3563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7040919199379666e-05,
+      "loss": 0.3809,
+      "step": 3564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.703931534712967e-05,
+      "loss": 0.2859,
+      "step": 3565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7037711107936277e-05,
+      "loss": 1.4586,
+      "step": 3566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7036106481851047e-05,
+      "loss": 0.8816,
+      "step": 3567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7034501468925557e-05,
+      "loss": 0.9301,
+      "step": 3568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.70328960692114e-05,
+      "loss": 0.8647,
+      "step": 3569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7031290282760156e-05,
+      "loss": 2.0969,
+      "step": 3570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7029684109623444e-05,
+      "loss": 1.0954,
+      "step": 3571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7028077549852888e-05,
+      "loss": 0.8221,
+      "step": 3572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7026470603500117e-05,
+      "loss": 0.6887,
+      "step": 3573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7024863270616782e-05,
+      "loss": 1.3075,
+      "step": 3574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7023255551254547e-05,
+      "loss": 0.7394,
+      "step": 3575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7021647445465077e-05,
+      "loss": 1.4729,
+      "step": 3576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7020038953300057e-05,
+      "loss": 0.66,
+      "step": 3577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7018430074811187e-05,
+      "loss": 1.0061,
+      "step": 3578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7016820810050178e-05,
+      "loss": 0.8017,
+      "step": 3579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7015211159068744e-05,
+      "loss": 0.7353,
+      "step": 3580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7013601121918627e-05,
+      "loss": 1.4322,
+      "step": 3581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.701199069865157e-05,
+      "loss": 0.4083,
+      "step": 3582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7010379889319333e-05,
+      "loss": 0.7987,
+      "step": 3583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7008768693973684e-05,
+      "loss": 0.364,
+      "step": 3584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7007157112666414e-05,
+      "loss": 0.8379,
+      "step": 3585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7005545145449314e-05,
+      "loss": 0.5722,
+      "step": 3586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7003932792374194e-05,
+      "loss": 0.2573,
+      "step": 3587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7002320053492874e-05,
+      "loss": 1.1085,
+      "step": 3588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7000706928857184e-05,
+      "loss": 0.466,
+      "step": 3589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6999093418518977e-05,
+      "loss": 0.8929,
+      "step": 3590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6997479522530108e-05,
+      "loss": 1.4213,
+      "step": 3591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6995865240942446e-05,
+      "loss": 0.3438,
+      "step": 3592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6994250573807877e-05,
+      "loss": 0.5049,
+      "step": 3593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6992635521178294e-05,
+      "loss": 1.2355,
+      "step": 3594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6991020083105612e-05,
+      "loss": 0.4469,
+      "step": 3595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6989404259641735e-05,
+      "loss": 0.3784,
+      "step": 3596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6987788050838612e-05,
+      "loss": 0.4523,
+      "step": 3597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6986171456748178e-05,
+      "loss": 0.4777,
+      "step": 3598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6984554477422396e-05,
+      "loss": 0.7862,
+      "step": 3599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.698293711291323e-05,
+      "loss": 0.5871,
+      "step": 3600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.698131936327267e-05,
+      "loss": 0.3773,
+      "step": 3601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6979701228552703e-05,
+      "loss": 0.1721,
+      "step": 3602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6978082708805338e-05,
+      "loss": 1.5888,
+      "step": 3603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6976463804082594e-05,
+      "loss": 0.4086,
+      "step": 3604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6974844514436504e-05,
+      "loss": 1.3716,
+      "step": 3605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6973224839919114e-05,
+      "loss": 1.3248,
+      "step": 3606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6971604780582476e-05,
+      "loss": 1.5791,
+      "step": 3607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6969984336478662e-05,
+      "loss": 0.5072,
+      "step": 3608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6968363507659747e-05,
+      "loss": 0.1608,
+      "step": 3609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6966742294177835e-05,
+      "loss": 0.2453,
+      "step": 3610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6965120696085017e-05,
+      "loss": 0.285,
+      "step": 3611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6963498713433427e-05,
+      "loss": 0.2546,
+      "step": 3612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.696187634627518e-05,
+      "loss": 0.198,
+      "step": 3613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.696025359466244e-05,
+      "loss": 0.6959,
+      "step": 3614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6958630458647336e-05,
+      "loss": 1.5009,
+      "step": 3615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6957006938282058e-05,
+      "loss": 1.1384,
+      "step": 3616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.695538303361877e-05,
+      "loss": 0.3425,
+      "step": 3617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6953758744709678e-05,
+      "loss": 0.4464,
+      "step": 3618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6952134071606975e-05,
+      "loss": 0.1862,
+      "step": 3619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6950509014362887e-05,
+      "loss": 0.1341,
+      "step": 3620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6948883573029633e-05,
+      "loss": 1.2884,
+      "step": 3621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6947257747659466e-05,
+      "loss": 0.82,
+      "step": 3622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.694563153830463e-05,
+      "loss": 0.7713,
+      "step": 3623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6944004945017396e-05,
+      "loss": 1.504,
+      "step": 3624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6942377967850042e-05,
+      "loss": 0.1985,
+      "step": 3625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6940750606854866e-05,
+      "loss": 0.1546,
+      "step": 3626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6939122862084158e-05,
+      "loss": 1.969,
+      "step": 3627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6937494733590246e-05,
+      "loss": 0.4069,
+      "step": 3628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.693586622142545e-05,
+      "loss": 0.8443,
+      "step": 3629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.693423732564211e-05,
+      "loss": 0.5106,
+      "step": 3630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.693260804629259e-05,
+      "loss": 0.4415,
+      "step": 3631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.693097838342924e-05,
+      "loss": 0.5248,
+      "step": 3632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6929348337104444e-05,
+      "loss": 0.3746,
+      "step": 3633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6927717907370593e-05,
+      "loss": 0.4872,
+      "step": 3634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6926087094280087e-05,
+      "loss": 1.0067,
+      "step": 3635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6924455897885342e-05,
+      "loss": 0.9622,
+      "step": 3636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6922824318238784e-05,
+      "loss": 0.9782,
+      "step": 3637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6921192355392853e-05,
+      "loss": 0.9086,
+      "step": 3638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6919560009399995e-05,
+      "loss": 0.4968,
+      "step": 3639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.691792728031268e-05,
+      "loss": 0.4747,
+      "step": 3640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.691629416818338e-05,
+      "loss": 1.7374,
+      "step": 3641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6914660673064583e-05,
+      "loss": 0.8613,
+      "step": 3642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6913026795008796e-05,
+      "loss": 1.0214,
+      "step": 3643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6911392534068523e-05,
+      "loss": 1.0877,
+      "step": 3644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6909757890296297e-05,
+      "loss": 1.4818,
+      "step": 3645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6908122863744647e-05,
+      "loss": 0.3967,
+      "step": 3646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6906487454466128e-05,
+      "loss": 0.8462,
+      "step": 3647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.69048516625133e-05,
+      "loss": 0.8056,
+      "step": 3648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.690321548793874e-05,
+      "loss": 0.6471,
+      "step": 3649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6901578930795032e-05,
+      "loss": 0.4266,
+      "step": 3650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6899941991134778e-05,
+      "loss": 1.2571,
+      "step": 3651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6898304669010583e-05,
+      "loss": 0.0807,
+      "step": 3652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6896666964475078e-05,
+      "loss": 0.2425,
+      "step": 3653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6895028877580895e-05,
+      "loss": 1.2916,
+      "step": 3654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.689339040838068e-05,
+      "loss": 0.9055,
+      "step": 3655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6891751556927102e-05,
+      "loss": 0.4823,
+      "step": 3656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6890112323272823e-05,
+      "loss": 0.4351,
+      "step": 3657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6888472707470536e-05,
+      "loss": 0.0943,
+      "step": 3658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6886832709572933e-05,
+      "loss": 0.2369,
+      "step": 3659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6885192329632724e-05,
+      "loss": 0.5922,
+      "step": 3660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.688355156770263e-05,
+      "loss": 0.454,
+      "step": 3661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6881910423835393e-05,
+      "loss": 0.1116,
+      "step": 3662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6880268898083748e-05,
+      "loss": 0.2427,
+      "step": 3663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6878626990500462e-05,
+      "loss": 0.2451,
+      "step": 3664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.68769847011383e-05,
+      "loss": 1.1908,
+      "step": 3665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6875342030050054e-05,
+      "loss": 0.3915,
+      "step": 3666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.687369897728851e-05,
+      "loss": 0.2452,
+      "step": 3667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6872055542906476e-05,
+      "loss": 0.688,
+      "step": 3668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.687041172695678e-05,
+      "loss": 0.064,
+      "step": 3669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6868767529492248e-05,
+      "loss": 0.7176,
+      "step": 3670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6867122950565723e-05,
+      "loss": 0.0821,
+      "step": 3671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.686547799023007e-05,
+      "loss": 0.5641,
+      "step": 3672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6863832648538148e-05,
+      "loss": 0.4095,
+      "step": 3673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6862186925542846e-05,
+      "loss": 0.2749,
+      "step": 3674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.686054082129705e-05,
+      "loss": 0.5213,
+      "step": 3675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6858894335853674e-05,
+      "loss": 0.2283,
+      "step": 3676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6857247469265634e-05,
+      "loss": 0.6053,
+      "step": 3677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.685560022158585e-05,
+      "loss": 0.3416,
+      "step": 3678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.685395259286728e-05,
+      "loss": 0.8148,
+      "step": 3679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6852304583162873e-05,
+      "loss": 0.2407,
+      "step": 3680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.685065619252559e-05,
+      "loss": 0.8744,
+      "step": 3681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6849007421008415e-05,
+      "loss": 0.6293,
+      "step": 3682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6847358268664345e-05,
+      "loss": 0.8758,
+      "step": 3683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6845708735546374e-05,
+      "loss": 1.0442,
+      "step": 3684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.684405882170752e-05,
+      "loss": 0.8783,
+      "step": 3685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6842408527200815e-05,
+      "loss": 0.8728,
+      "step": 3686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6840757852079296e-05,
+      "loss": 1.0199,
+      "step": 3687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.683910679639602e-05,
+      "loss": 0.9914,
+      "step": 3688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6837455360204053e-05,
+      "loss": 0.7656,
+      "step": 3689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6835803543556465e-05,
+      "loss": 1.2671,
+      "step": 3690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6834151346506343e-05,
+      "loss": 0.2524,
+      "step": 3691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6832498769106803e-05,
+      "loss": 0.4704,
+      "step": 3692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6830845811410947e-05,
+      "loss": 0.2658,
+      "step": 3693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6829192473471904e-05,
+      "loss": 0.2806,
+      "step": 3694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.682753875534281e-05,
+      "loss": 1.0381,
+      "step": 3695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6825884657076823e-05,
+      "loss": 1.1593,
+      "step": 3696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6824230178727098e-05,
+      "loss": 0.858,
+      "step": 3697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6822575320346815e-05,
+      "loss": 0.5072,
+      "step": 3698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6820920081989155e-05,
+      "loss": 0.1138,
+      "step": 3699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.681926446370732e-05,
+      "loss": 0.5026,
+      "step": 3700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6817608465554526e-05,
+      "loss": 1.2399,
+      "step": 3701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6815952087583994e-05,
+      "loss": 0.5916,
+      "step": 3702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.681429532984896e-05,
+      "loss": 0.937,
+      "step": 3703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6812638192402665e-05,
+      "loss": 0.4001,
+      "step": 3704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.681098067529838e-05,
+      "loss": 0.4638,
+      "step": 3705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6809322778589376e-05,
+      "loss": 0.9971,
+      "step": 3706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6807664502328928e-05,
+      "loss": 0.3925,
+      "step": 3707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6806005846570346e-05,
+      "loss": 0.0495,
+      "step": 3708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.680434681136693e-05,
+      "loss": 1.4351,
+      "step": 3709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.680268739677201e-05,
+      "loss": 0.0663,
+      "step": 3710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.680102760283891e-05,
+      "loss": 0.0492,
+      "step": 3711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6799367429620982e-05,
+      "loss": 0.2449,
+      "step": 3712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6797706877171578e-05,
+      "loss": 0.5693,
+      "step": 3713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6796045945544077e-05,
+      "loss": 0.1396,
+      "step": 3714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.679438463479185e-05,
+      "loss": 1.2959,
+      "step": 3715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6792722944968305e-05,
+      "loss": 0.9458,
+      "step": 3716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6791060876126837e-05,
+      "loss": 0.6183,
+      "step": 3717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.678939842832087e-05,
+      "loss": 0.239,
+      "step": 3718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6787735601603836e-05,
+      "loss": 0.6735,
+      "step": 3719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6786072396029178e-05,
+      "loss": 0.0779,
+      "step": 3720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6784408811650347e-05,
+      "loss": 0.3011,
+      "step": 3721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6782744848520812e-05,
+      "loss": 0.6361,
+      "step": 3722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.678108050669406e-05,
+      "loss": 1.0826,
+      "step": 3723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6779415786223572e-05,
+      "loss": 0.7789,
+      "step": 3724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.677775068716286e-05,
+      "loss": 0.9301,
+      "step": 3725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.677608520956544e-05,
+      "loss": 0.1764,
+      "step": 3726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6774419353484834e-05,
+      "loss": 1.0291,
+      "step": 3727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6772753118974587e-05,
+      "loss": 0.2962,
+      "step": 3728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.677108650608825e-05,
+      "loss": 0.4363,
+      "step": 3729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6769419514879392e-05,
+      "loss": 1.2487,
+      "step": 3730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6767752145401588e-05,
+      "loss": 0.5567,
+      "step": 3731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6766084397708427e-05,
+      "loss": 1.043,
+      "step": 3732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6764416271853508e-05,
+      "loss": 0.7929,
+      "step": 3733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6762747767890447e-05,
+      "loss": 0.8472,
+      "step": 3734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.676107888587287e-05,
+      "loss": 0.991,
+      "step": 3735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6759409625854414e-05,
+      "loss": 0.9694,
+      "step": 3736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.675773998788873e-05,
+      "loss": 0.7299,
+      "step": 3737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6756069972029475e-05,
+      "loss": 1.2238,
+      "step": 3738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.675439957833033e-05,
+      "loss": 0.8835,
+      "step": 3739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.675272880684498e-05,
+      "loss": 0.8265,
+      "step": 3740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6751057657627118e-05,
+      "loss": 0.2166,
+      "step": 3741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6749386130730465e-05,
+      "loss": 0.9143,
+      "step": 3742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6747714226208734e-05,
+      "loss": 0.2682,
+      "step": 3743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6746041944115666e-05,
+      "loss": 0.2177,
+      "step": 3744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6744369284505007e-05,
+      "loss": 0.3468,
+      "step": 3745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6742696247430517e-05,
+      "loss": 0.5388,
+      "step": 3746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.674102283294596e-05,
+      "loss": 0.7501,
+      "step": 3747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6739349041105127e-05,
+      "loss": 1.0182,
+      "step": 3748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6737674871961813e-05,
+      "loss": 0.6783,
+      "step": 3749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6736000325569825e-05,
+      "loss": 0.4678,
+      "step": 3750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6734325401982983e-05,
+      "loss": 0.4694,
+      "step": 3751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6732650101255117e-05,
+      "loss": 0.0812,
+      "step": 3752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6730974423440076e-05,
+      "loss": 1.2342,
+      "step": 3753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.672929836859171e-05,
+      "loss": 1.5434,
+      "step": 3754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.672762193676389e-05,
+      "loss": 1.3129,
+      "step": 3755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6725945128010498e-05,
+      "loss": 0.3388,
+      "step": 3756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6724267942385425e-05,
+      "loss": 0.4487,
+      "step": 3757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6722590379942575e-05,
+      "loss": 1.9124,
+      "step": 3758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.672091244073587e-05,
+      "loss": 0.4523,
+      "step": 3759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6719234124819232e-05,
+      "loss": 0.4831,
+      "step": 3760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6717555432246605e-05,
+      "loss": 0.3704,
+      "step": 3761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6715876363071942e-05,
+      "loss": 0.3284,
+      "step": 3762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.671419691734921e-05,
+      "loss": 0.4994,
+      "step": 3763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6712517095132383e-05,
+      "loss": 1.4433,
+      "step": 3764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.671083689647545e-05,
+      "loss": 0.2595,
+      "step": 3765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6709156321432423e-05,
+      "loss": 0.2758,
+      "step": 3766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6707475370057297e-05,
+      "loss": 1.8288,
+      "step": 3767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6705794042404114e-05,
+      "loss": 0.8516,
+      "step": 3768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6704112338526906e-05,
+      "loss": 1.6458,
+      "step": 3769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6702430258479723e-05,
+      "loss": 1.3239,
+      "step": 3770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6700747802316623e-05,
+      "loss": 1.8139,
+      "step": 3771
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6699064970091688e-05,
+      "loss": 1.3617,
+      "step": 3772
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6697381761859e-05,
+      "loss": 1.9514,
+      "step": 3773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6695698177672652e-05,
+      "loss": 1.224,
+      "step": 3774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6694014217586766e-05,
+      "loss": 0.7297,
+      "step": 3775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.669232988165546e-05,
+      "loss": 0.4582,
+      "step": 3776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6690645169932863e-05,
+      "loss": 0.3378,
+      "step": 3777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6688960082473127e-05,
+      "loss": 1.839,
+      "step": 3778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.668727461933041e-05,
+      "loss": 0.3973,
+      "step": 3779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.668558878055888e-05,
+      "loss": 0.698,
+      "step": 3780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6683902566212725e-05,
+      "loss": 0.1044,
+      "step": 3781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6682215976346135e-05,
+      "loss": 0.4147,
+      "step": 3782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6680529011013315e-05,
+      "loss": 0.397,
+      "step": 3783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6678841670268493e-05,
+      "loss": 0.9225,
+      "step": 3784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.667715395416589e-05,
+      "loss": 1.2023,
+      "step": 3785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6675465862759757e-05,
+      "loss": 1.1632,
+      "step": 3786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6673777396104345e-05,
+      "loss": 1.1383,
+      "step": 3787
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6672088554253924e-05,
+      "loss": 0.7191,
+      "step": 3788
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6670399337262772e-05,
+      "loss": 0.6222,
+      "step": 3789
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.666870974518518e-05,
+      "loss": 0.4995,
+      "step": 3790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6667019778075445e-05,
+      "loss": 2.1366,
+      "step": 3791
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.66653294359879e-05,
+      "loss": 0.8036,
+      "step": 3792
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6663638718976857e-05,
+      "loss": 1.3674,
+      "step": 3793
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6661947627096656e-05,
+      "loss": 1.4435,
+      "step": 3794
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.666025616040166e-05,
+      "loss": 1.115,
+      "step": 3795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6658564318946222e-05,
+      "loss": 1.0144,
+      "step": 3796
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6656872102784723e-05,
+      "loss": 1.5892,
+      "step": 3797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6655179511971546e-05,
+      "loss": 0.8084,
+      "step": 3798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.66534865465611e-05,
+      "loss": 0.276,
+      "step": 3799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6651793206607787e-05,
+      "loss": 0.2976,
+      "step": 3800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6650099492166036e-05,
+      "loss": 0.2706,
+      "step": 3801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.664840540329028e-05,
+      "loss": 0.2995,
+      "step": 3802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6646710940034974e-05,
+      "loss": 0.1419,
+      "step": 3803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6645016102454568e-05,
+      "loss": 1.2401,
+      "step": 3804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.664332089060354e-05,
+      "loss": 1.0762,
+      "step": 3805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6641625304536375e-05,
+      "loss": 0.4105,
+      "step": 3806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6639929344307568e-05,
+      "loss": 0.8869,
+      "step": 3807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6638233009971623e-05,
+      "loss": 0.5229,
+      "step": 3808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6636536301583062e-05,
+      "loss": 1.6375,
+      "step": 3809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6634839219196424e-05,
+      "loss": 1.733,
+      "step": 3810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6633141762866244e-05,
+      "loss": 0.4855,
+      "step": 3811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6631443932647085e-05,
+      "loss": 0.6973,
+      "step": 3812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.662974572859351e-05,
+      "loss": 0.1115,
+      "step": 3813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6628047150760103e-05,
+      "loss": 0.6995,
+      "step": 3814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6626348199201452e-05,
+      "loss": 0.8098,
+      "step": 3815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6624648873972167e-05,
+      "loss": 0.2142,
+      "step": 3816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.662294917512686e-05,
+      "loss": 0.4279,
+      "step": 3817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6621249102720162e-05,
+      "loss": 0.4823,
+      "step": 3818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.661954865680671e-05,
+      "loss": 0.8469,
+      "step": 3819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.661784783744116e-05,
+      "loss": 2.104,
+      "step": 3820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6616146644678175e-05,
+      "loss": 0.1368,
+      "step": 3821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6614445078572423e-05,
+      "loss": 0.1943,
+      "step": 3822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.661274313917861e-05,
+      "loss": 0.9119,
+      "step": 3823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6611040826551422e-05,
+      "loss": 1.1303,
+      "step": 3824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.660933814074558e-05,
+      "loss": 0.5846,
+      "step": 3825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.66076350818158e-05,
+      "loss": 0.3804,
+      "step": 3826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6605931649816823e-05,
+      "loss": 0.3343,
+      "step": 3827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.66042278448034e-05,
+      "loss": 0.2233,
+      "step": 3828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.660252366683028e-05,
+      "loss": 1.3339,
+      "step": 3829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.660081911595225e-05,
+      "loss": 0.8948,
+      "step": 3830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6599114192224084e-05,
+      "loss": 0.4125,
+      "step": 3831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6597408895700583e-05,
+      "loss": 0.7823,
+      "step": 3832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6595703226436557e-05,
+      "loss": 1.6369,
+      "step": 3833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.659399718448682e-05,
+      "loss": 0.3231,
+      "step": 3834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.659229076990621e-05,
+      "loss": 0.2675,
+      "step": 3835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.659058398274956e-05,
+      "loss": 0.6105,
+      "step": 3836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6588876823071744e-05,
+      "loss": 0.5438,
+      "step": 3837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.658716929092761e-05,
+      "loss": 0.979,
+      "step": 3838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6585461386372058e-05,
+      "loss": 0.1849,
+      "step": 3839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.658375310945997e-05,
+      "loss": 0.9796,
+      "step": 3840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.658204446024625e-05,
+      "loss": 1.3553,
+      "step": 3841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6580335438785807e-05,
+      "loss": 1.7203,
+      "step": 3842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6578626045133586e-05,
+      "loss": 0.5719,
+      "step": 3843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.657691627934451e-05,
+      "loss": 0.5966,
+      "step": 3844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6575206141473542e-05,
+      "loss": 0.9742,
+      "step": 3845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6573495631575638e-05,
+      "loss": 0.6885,
+      "step": 3846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6571784749705783e-05,
+      "loss": 0.397,
+      "step": 3847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.657007349591896e-05,
+      "loss": 0.9525,
+      "step": 3848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6568361870270166e-05,
+      "loss": 0.6829,
+      "step": 3849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6566649872814416e-05,
+      "loss": 0.5627,
+      "step": 3850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6564937503606727e-05,
+      "loss": 1.5479,
+      "step": 3851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6563224762702143e-05,
+      "loss": 0.9331,
+      "step": 3852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.656151165015571e-05,
+      "loss": 0.3299,
+      "step": 3853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6559798166022483e-05,
+      "loss": 0.2806,
+      "step": 3854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6558084310357536e-05,
+      "loss": 0.4384,
+      "step": 3855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6556370083215952e-05,
+      "loss": 0.1628,
+      "step": 3856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.655465548465283e-05,
+      "loss": 0.7727,
+      "step": 3857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6552940514723274e-05,
+      "loss": 0.0355,
+      "step": 3858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.65512251734824e-05,
+      "loss": 0.1234,
+      "step": 3859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6549509460985344e-05,
+      "loss": 0.1321,
+      "step": 3860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6547793377287243e-05,
+      "loss": 2.0076,
+      "step": 3861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.654607692244326e-05,
+      "loss": 0.1899,
+      "step": 3862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6544360096508552e-05,
+      "loss": 0.1241,
+      "step": 3863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.654264289953831e-05,
+      "loss": 0.6442,
+      "step": 3864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6540925331587714e-05,
+      "loss": 0.3974,
+      "step": 3865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6539207392711973e-05,
+      "loss": 0.5803,
+      "step": 3866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6537489082966296e-05,
+      "loss": 0.414,
+      "step": 3867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6535770402405914e-05,
+      "loss": 0.6132,
+      "step": 3868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6534051351086066e-05,
+      "loss": 1.5141,
+      "step": 3869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6532331929061998e-05,
+      "loss": 0.3215,
+      "step": 3870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6530612136388973e-05,
+      "loss": 0.9616,
+      "step": 3871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6528891973122273e-05,
+      "loss": 0.3905,
+      "step": 3872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.652717143931717e-05,
+      "loss": 0.3377,
+      "step": 3873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6525450535028973e-05,
+      "loss": 0.6667,
+      "step": 3874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.652372926031299e-05,
+      "loss": 1.4193,
+      "step": 3875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.652200761522454e-05,
+      "loss": 0.2825,
+      "step": 3876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6520285599818955e-05,
+      "loss": 0.6341,
+      "step": 3877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.651856321415159e-05,
+      "loss": 0.8087,
+      "step": 3878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6516840458277784e-05,
+      "loss": 1.2823,
+      "step": 3879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6515117332252925e-05,
+      "loss": 1.2527,
+      "step": 3880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6513393836132392e-05,
+      "loss": 0.3196,
+      "step": 3881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6511669969971568e-05,
+      "loss": 0.9349,
+      "step": 3882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.650994573382586e-05,
+      "loss": 0.7307,
+      "step": 3883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6508221127750697e-05,
+      "loss": 0.088,
+      "step": 3884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.650649615180149e-05,
+      "loss": 0.3586,
+      "step": 3885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6504770806033695e-05,
+      "loss": 1.2877,
+      "step": 3886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6503045090502756e-05,
+      "loss": 1.4716,
+      "step": 3887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6501319005264142e-05,
+      "loss": 0.4419,
+      "step": 3888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6499592550373326e-05,
+      "loss": 0.7158,
+      "step": 3889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6497865725885796e-05,
+      "loss": 0.8136,
+      "step": 3890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6496138531857055e-05,
+      "loss": 0.8161,
+      "step": 3891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6494410968342613e-05,
+      "loss": 0.3229,
+      "step": 3892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6492683035397996e-05,
+      "loss": 1.5276,
+      "step": 3893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6490954733078737e-05,
+      "loss": 0.9958,
+      "step": 3894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6489226061440387e-05,
+      "loss": 0.4218,
+      "step": 3895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.64874970205385e-05,
+      "loss": 0.6974,
+      "step": 3896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6485767610428652e-05,
+      "loss": 1.2972,
+      "step": 3897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6484037831166428e-05,
+      "loss": 0.7506,
+      "step": 3898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6482307682807412e-05,
+      "loss": 0.4066,
+      "step": 3899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.648057716540723e-05,
+      "loss": 0.5835,
+      "step": 3900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.647884627902148e-05,
+      "loss": 1.3521,
+      "step": 3901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6477115023705805e-05,
+      "loss": 0.2706,
+      "step": 3902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6475383399515843e-05,
+      "loss": 0.4488,
+      "step": 3903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.647365140650726e-05,
+      "loss": 0.2409,
+      "step": 3904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.64719190447357e-05,
+      "loss": 0.316,
+      "step": 3905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6470186314256858e-05,
+      "loss": 0.2441,
+      "step": 3906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6468453215126423e-05,
+      "loss": 0.0524,
+      "step": 3907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.646671974740009e-05,
+      "loss": 0.4843,
+      "step": 3908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6464985911133576e-05,
+      "loss": 0.5602,
+      "step": 3909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.646325170638261e-05,
+      "loss": 0.5776,
+      "step": 3910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6461517133202918e-05,
+      "loss": 0.3256,
+      "step": 3911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6459782191650262e-05,
+      "loss": 1.2238,
+      "step": 3912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6458046881780396e-05,
+      "loss": 0.894,
+      "step": 3913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6456311203649095e-05,
+      "loss": 0.8814,
+      "step": 3914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6454575157312146e-05,
+      "loss": 0.0346,
+      "step": 3915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6452838742825342e-05,
+      "loss": 0.2234,
+      "step": 3916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.645110196024449e-05,
+      "loss": 0.8155,
+      "step": 3917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6449364809625414e-05,
+      "loss": 0.4025,
+      "step": 3918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6447627291023944e-05,
+      "loss": 0.0509,
+      "step": 3919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6445889404495922e-05,
+      "loss": 0.183,
+      "step": 3920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6444151150097206e-05,
+      "loss": 0.6848,
+      "step": 3921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6442412527883664e-05,
+      "loss": 1.6637,
+      "step": 3922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6440673537911175e-05,
+      "loss": 0.4389,
+      "step": 3923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6438934180235626e-05,
+      "loss": 0.3028,
+      "step": 3924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.643719445491293e-05,
+      "loss": 1.1697,
+      "step": 3925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6435454361998994e-05,
+      "loss": 0.4883,
+      "step": 3926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6433713901549742e-05,
+      "loss": 0.7355,
+      "step": 3927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.643197307362112e-05,
+      "loss": 1.6862,
+      "step": 3928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6430231878269067e-05,
+      "loss": 0.1287,
+      "step": 3929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6428490315549563e-05,
+      "loss": 0.5563,
+      "step": 3930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6426748385518566e-05,
+      "loss": 0.8606,
+      "step": 3931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6425006088232063e-05,
+      "loss": 1.1527,
+      "step": 3932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6423263423746057e-05,
+      "loss": 0.8553,
+      "step": 3933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.642152039211656e-05,
+      "loss": 0.1672,
+      "step": 3934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6419776993399585e-05,
+      "loss": 1.2227,
+      "step": 3935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6418033227651167e-05,
+      "loss": 0.6523,
+      "step": 3936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6416289094927355e-05,
+      "loss": 0.9722,
+      "step": 3937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.64145445952842e-05,
+      "loss": 0.7186,
+      "step": 3938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6412799728777776e-05,
+      "loss": 0.357,
+      "step": 3939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6411054495464153e-05,
+      "loss": 0.3486,
+      "step": 3940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6409308895399436e-05,
+      "loss": 0.915,
+      "step": 3941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.640756292863972e-05,
+      "loss": 0.8193,
+      "step": 3942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6405816595241115e-05,
+      "loss": 0.1246,
+      "step": 3943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6404069895259763e-05,
+      "loss": 0.7471,
+      "step": 3944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.640232282875179e-05,
+      "loss": 0.2654,
+      "step": 3945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6400575395773357e-05,
+      "loss": 0.1323,
+      "step": 3946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6398827596380617e-05,
+      "loss": 1.2352,
+      "step": 3947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6397079430629752e-05,
+      "loss": 1.2158,
+      "step": 3948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.639533089857694e-05,
+      "loss": 1.2602,
+      "step": 3949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6393582000278387e-05,
+      "loss": 0.7604,
+      "step": 3950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6391832735790295e-05,
+      "loss": 0.3665,
+      "step": 3951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.639008310516889e-05,
+      "loss": 1.4275,
+      "step": 3952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.638833310847041e-05,
+      "loss": 0.9202,
+      "step": 3953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.638658274575109e-05,
+      "loss": 0.4293,
+      "step": 3954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6384832017067188e-05,
+      "loss": 1.6285,
+      "step": 3955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.638308092247498e-05,
+      "loss": 0.0644,
+      "step": 3956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6381329462030737e-05,
+      "loss": 1.1402,
+      "step": 3957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6379577635790758e-05,
+      "loss": 1.2495,
+      "step": 3958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6377825443811344e-05,
+      "loss": 0.645,
+      "step": 3959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.637607288614881e-05,
+      "loss": 0.1665,
+      "step": 3960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.637431996285948e-05,
+      "loss": 0.0785,
+      "step": 3961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.63725666739997e-05,
+      "loss": 0.051,
+      "step": 3962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.637081301962582e-05,
+      "loss": 1.8203,
+      "step": 3963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6369058999794194e-05,
+      "loss": 0.3548,
+      "step": 3964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6367304614561202e-05,
+      "loss": 0.4936,
+      "step": 3965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6365549863983233e-05,
+      "loss": 1.3776,
+      "step": 3966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6363794748116677e-05,
+      "loss": 1.8327,
+      "step": 3967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.636203926701795e-05,
+      "loss": 0.2749,
+      "step": 3968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6360283420743473e-05,
+      "loss": 0.2133,
+      "step": 3969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.635852720934967e-05,
+      "loss": 0.0973,
+      "step": 3970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6356770632893002e-05,
+      "loss": 0.6898,
+      "step": 3971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6355013691429905e-05,
+      "loss": 0.1031,
+      "step": 3972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6353256385016865e-05,
+      "loss": 1.9106,
+      "step": 3973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6351498713710352e-05,
+      "loss": 0.5105,
+      "step": 3974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.634974067756686e-05,
+      "loss": 0.9562,
+      "step": 3975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6347982276642895e-05,
+      "loss": 0.164,
+      "step": 3976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.634622351099496e-05,
+      "loss": 0.6001,
+      "step": 3977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.63444643806796e-05,
+      "loss": 0.6113,
+      "step": 3978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6342704885753343e-05,
+      "loss": 0.1903,
+      "step": 3979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.634094502627274e-05,
+      "loss": 0.6698,
+      "step": 3980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.633918480229435e-05,
+      "loss": 0.2232,
+      "step": 3981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6337424213874754e-05,
+      "loss": 0.9943,
+      "step": 3982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.633566326107053e-05,
+      "loss": 0.4556,
+      "step": 3983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.633390194393828e-05,
+      "loss": 0.4927,
+      "step": 3984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.633214026253461e-05,
+      "loss": 0.7165,
+      "step": 3985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6330378216916137e-05,
+      "loss": 0.4298,
+      "step": 3986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6328615807139506e-05,
+      "loss": 0.3899,
+      "step": 3987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6326853033261346e-05,
+      "loss": 1.4781,
+      "step": 3988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6325089895338324e-05,
+      "loss": 0.6757,
+      "step": 3989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.63233263934271e-05,
+      "loss": 0.2479,
+      "step": 3990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6321562527584353e-05,
+      "loss": 0.7119,
+      "step": 3991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6319798297866772e-05,
+      "loss": 0.8624,
+      "step": 3992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.631803370433107e-05,
+      "loss": 0.2909,
+      "step": 3993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.631626874703395e-05,
+      "loss": 0.7324,
+      "step": 3994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6314503426032143e-05,
+      "loss": 1.0138,
+      "step": 3995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6312737741382383e-05,
+      "loss": 0.4742,
+      "step": 3996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.631097169314142e-05,
+      "loss": 0.6762,
+      "step": 3997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.630920528136602e-05,
+      "loss": 0.4893,
+      "step": 3998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.630743850611295e-05,
+      "loss": 0.8535,
+      "step": 3999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.630567136743899e-05,
+      "loss": 0.9932,
+      "step": 4000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6303903865400947e-05,
+      "loss": 1.5587,
+      "step": 4001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6302136000055618e-05,
+      "loss": 1.361,
+      "step": 4002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.630036777145983e-05,
+      "loss": 0.1914,
+      "step": 4003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6298599179670416e-05,
+      "loss": 1.4676,
+      "step": 4004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.62968302247442e-05,
+      "loss": 1.4253,
+      "step": 4005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.629506090673806e-05,
+      "loss": 0.3248,
+      "step": 4006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.629329122570885e-05,
+      "loss": 0.309,
+      "step": 4007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6291521181713444e-05,
+      "loss": 0.0288,
+      "step": 4008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.628975077480874e-05,
+      "loss": 1.6804,
+      "step": 4009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6287980005051628e-05,
+      "loss": 1.1254,
+      "step": 4010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6286208872499035e-05,
+      "loss": 0.4964,
+      "step": 4011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6284437377207874e-05,
+      "loss": 1.2538,
+      "step": 4012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6282665519235082e-05,
+      "loss": 0.6653,
+      "step": 4013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6280893298637612e-05,
+      "loss": 0.5153,
+      "step": 4014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.627912071547242e-05,
+      "loss": 0.5218,
+      "step": 4015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6277347769796473e-05,
+      "loss": 0.23,
+      "step": 4016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.627557446166676e-05,
+      "loss": 0.7812,
+      "step": 4017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6273800791140275e-05,
+      "loss": 0.2062,
+      "step": 4018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6272026758274014e-05,
+      "loss": 0.3281,
+      "step": 4019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6270252363125007e-05,
+      "loss": 1.0403,
+      "step": 4020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6268477605750274e-05,
+      "loss": 0.1759,
+      "step": 4021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6266702486206862e-05,
+      "loss": 0.7464,
+      "step": 4022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.626492700455182e-05,
+      "loss": 1.4661,
+      "step": 4023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6263151160842204e-05,
+      "loss": 1.0794,
+      "step": 4024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6261374955135108e-05,
+      "loss": 1.0091,
+      "step": 4025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6259598387487605e-05,
+      "loss": 0.9648,
+      "step": 4026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.62578214579568e-05,
+      "loss": 1.1518,
+      "step": 4027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6256044166599798e-05,
+      "loss": 0.602,
+      "step": 4028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6254266513473728e-05,
+      "loss": 0.4261,
+      "step": 4029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6252488498635716e-05,
+      "loss": 0.615,
+      "step": 4030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6250710122142917e-05,
+      "loss": 1.5947,
+      "step": 4031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6248931384052477e-05,
+      "loss": 0.9029,
+      "step": 4032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6247152284421575e-05,
+      "loss": 1.6021,
+      "step": 4033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6245372823307383e-05,
+      "loss": 0.3445,
+      "step": 4034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6243593000767098e-05,
+      "loss": 0.2651,
+      "step": 4035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6241812816857923e-05,
+      "loss": 1.1978,
+      "step": 4036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.624003227163707e-05,
+      "loss": 0.116,
+      "step": 4037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6238251365161767e-05,
+      "loss": 0.0999,
+      "step": 4038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6236470097489254e-05,
+      "loss": 0.1776,
+      "step": 4039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6234688468676776e-05,
+      "loss": 0.9314,
+      "step": 4040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6232906478781606e-05,
+      "loss": 0.3578,
+      "step": 4041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6231124127861005e-05,
+      "loss": 0.6244,
+      "step": 4042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6229341415972265e-05,
+      "loss": 0.3875,
+      "step": 4043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6227558343172672e-05,
+      "loss": 1.131,
+      "step": 4044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6225774909519548e-05,
+      "loss": 0.6404,
+      "step": 4045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6223991115070206e-05,
+      "loss": 0.905,
+      "step": 4046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.622220695988197e-05,
+      "loss": 0.6035,
+      "step": 4047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6220422444012195e-05,
+      "loss": 0.389,
+      "step": 4048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6218637567518232e-05,
+      "loss": 0.3112,
+      "step": 4049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6216852330457445e-05,
+      "loss": 0.7649,
+      "step": 4050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6215066732887207e-05,
+      "loss": 0.2978,
+      "step": 4051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6213280774864914e-05,
+      "loss": 1.5632,
+      "step": 4052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6211494456447964e-05,
+      "loss": 1.4399,
+      "step": 4053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6209707777693773e-05,
+      "loss": 0.0588,
+      "step": 4054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6207920738659756e-05,
+      "loss": 0.3996,
+      "step": 4055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.620613333940335e-05,
+      "loss": 0.6299,
+      "step": 4056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6204345579982018e-05,
+      "loss": 0.5167,
+      "step": 4057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6202557460453196e-05,
+      "loss": 1.3294,
+      "step": 4058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6200768980874365e-05,
+      "loss": 0.3665,
+      "step": 4059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.619898014130301e-05,
+      "loss": 0.3513,
+      "step": 4060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.619719094179662e-05,
+      "loss": 1.3336,
+      "step": 4061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.61954013824127e-05,
+      "loss": 0.1944,
+      "step": 4062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6193611463208762e-05,
+      "loss": 0.203,
+      "step": 4063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6191821184242347e-05,
+      "loss": 0.654,
+      "step": 4064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.619003054557098e-05,
+      "loss": 0.1941,
+      "step": 4065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.618823954725222e-05,
+      "loss": 0.2027,
+      "step": 4066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6186448189343626e-05,
+      "loss": 0.3806,
+      "step": 4067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6184656471902777e-05,
+      "loss": 0.5589,
+      "step": 4068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6182864394987253e-05,
+      "loss": 0.1559,
+      "step": 4069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6181071958654657e-05,
+      "loss": 0.4963,
+      "step": 4070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6179279162962588e-05,
+      "loss": 0.2732,
+      "step": 4071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6177486007968682e-05,
+      "loss": 0.9475,
+      "step": 4072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.617569249373056e-05,
+      "loss": 0.7459,
+      "step": 4073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6173898620305864e-05,
+      "loss": 1.2401,
+      "step": 4074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.617210438775226e-05,
+      "loss": 0.3834,
+      "step": 4075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.61703097961274e-05,
+      "loss": 0.1412,
+      "step": 4076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.616851484548898e-05,
+      "loss": 1.5376,
+      "step": 4077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6166719535894666e-05,
+      "loss": 0.1527,
+      "step": 4078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6164923867402183e-05,
+      "loss": 0.33,
+      "step": 4079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6163127840069226e-05,
+      "loss": 0.3488,
+      "step": 4080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6161331453953533e-05,
+      "loss": 0.8781,
+      "step": 4081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.615953470911283e-05,
+      "loss": 0.5369,
+      "step": 4082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6157737605604867e-05,
+      "loss": 1.6653,
+      "step": 4083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.615594014348741e-05,
+      "loss": 0.6422,
+      "step": 4084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6154142322818214e-05,
+      "loss": 0.9096,
+      "step": 4085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6152344143655078e-05,
+      "loss": 0.5332,
+      "step": 4086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6150545606055783e-05,
+      "loss": 1.0682,
+      "step": 4087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6148746710078143e-05,
+      "loss": 1.0898,
+      "step": 4088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6146947455779963e-05,
+      "loss": 0.1468,
+      "step": 4089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6145147843219083e-05,
+      "loss": 0.694,
+      "step": 4090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6143347872453337e-05,
+      "loss": 0.4148,
+      "step": 4091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6141547543540574e-05,
+      "loss": 0.425,
+      "step": 4092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.613974685653866e-05,
+      "loss": 1.1151,
+      "step": 4093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6137945811505467e-05,
+      "loss": 0.2172,
+      "step": 4094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6136144408498887e-05,
+      "loss": 0.564,
+      "step": 4095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.613434264757681e-05,
+      "loss": 0.2288,
+      "step": 4096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6132540528797143e-05,
+      "loss": 0.383,
+      "step": 4097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.613073805221781e-05,
+      "loss": 0.4165,
+      "step": 4098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6128935217896744e-05,
+      "loss": 0.5611,
+      "step": 4099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6127132025891886e-05,
+      "loss": 0.5504,
+      "step": 4100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6125328476261187e-05,
+      "loss": 0.429,
+      "step": 4101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6123524569062623e-05,
+      "loss": 1.525,
+      "step": 4102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6121720304354156e-05,
+      "loss": 0.1177,
+      "step": 4103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.611991568219379e-05,
+      "loss": 1.5974,
+      "step": 4104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.611811070263952e-05,
+      "loss": 1.2085,
+      "step": 4105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6116305365749355e-05,
+      "loss": 1.5173,
+      "step": 4106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6114499671581327e-05,
+      "loss": 0.6767,
+      "step": 4107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.611269362019346e-05,
+      "loss": 0.3803,
+      "step": 4108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6110887211643808e-05,
+      "loss": 0.3352,
+      "step": 4109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6109080445990425e-05,
+      "loss": 1.64,
+      "step": 4110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6107273323291384e-05,
+      "loss": 0.6463,
+      "step": 4111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.610546584360476e-05,
+      "loss": 1.2815,
+      "step": 4112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6103658006988653e-05,
+      "loss": 1.3216,
+      "step": 4113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.610184981350116e-05,
+      "loss": 0.2526,
+      "step": 4114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.61000412632004e-05,
+      "loss": 0.4339,
+      "step": 4115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.60982323561445e-05,
+      "loss": 0.7585,
+      "step": 4116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.60964230923916e-05,
+      "loss": 0.5294,
+      "step": 4117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6094613471999845e-05,
+      "loss": 1.6278,
+      "step": 4118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6092803495027396e-05,
+      "loss": 0.2123,
+      "step": 4119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.609099316153243e-05,
+      "loss": 0.1928,
+      "step": 4120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.608918247157313e-05,
+      "loss": 0.7016,
+      "step": 4121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.608737142520769e-05,
+      "loss": 0.48,
+      "step": 4122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6085560022494316e-05,
+      "loss": 0.0771,
+      "step": 4123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6083748263491235e-05,
+      "loss": 0.7138,
+      "step": 4124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.608193614825666e-05,
+      "loss": 0.2872,
+      "step": 4125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6080123676848847e-05,
+      "loss": 1.5172,
+      "step": 4126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6078310849326046e-05,
+      "loss": 0.8168,
+      "step": 4127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6076497665746516e-05,
+      "loss": 0.7598,
+      "step": 4128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6074684126168538e-05,
+      "loss": 1.2174,
+      "step": 4129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.60728702306504e-05,
+      "loss": 1.45,
+      "step": 4130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.607105597925039e-05,
+      "loss": 0.6454,
+      "step": 4131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.606924137202683e-05,
+      "loss": 1.9236,
+      "step": 4132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.606742640903804e-05,
+      "loss": 0.7795,
+      "step": 4133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6065611090342345e-05,
+      "loss": 0.5029,
+      "step": 4134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6063795415998096e-05,
+      "loss": 0.5708,
+      "step": 4135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.606197938606365e-05,
+      "loss": 1.4325,
+      "step": 4136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.606016300059737e-05,
+      "loss": 0.2274,
+      "step": 4137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6058346259657635e-05,
+      "loss": 0.7052,
+      "step": 4138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6056529163302836e-05,
+      "loss": 1.0264,
+      "step": 4139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6054711711591373e-05,
+      "loss": 0.7884,
+      "step": 4140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.605289390458166e-05,
+      "loss": 1.13,
+      "step": 4141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6051075742332126e-05,
+      "loss": 0.9398,
+      "step": 4142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6049257224901197e-05,
+      "loss": 0.298,
+      "step": 4143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6047438352347327e-05,
+      "loss": 0.1123,
+      "step": 4144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6045619124728975e-05,
+      "loss": 0.783,
+      "step": 4145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6043799542104607e-05,
+      "loss": 0.8341,
+      "step": 4146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.604197960453271e-05,
+      "loss": 0.9952,
+      "step": 4147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6040159312071766e-05,
+      "loss": 0.6095,
+      "step": 4148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6038338664780292e-05,
+      "loss": 0.1214,
+      "step": 4149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6036517662716793e-05,
+      "loss": 0.3209,
+      "step": 4150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6034696305939805e-05,
+      "loss": 1.004,
+      "step": 4151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.603287459450786e-05,
+      "loss": 1.0859,
+      "step": 4152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6031052528479506e-05,
+      "loss": 1.0949,
+      "step": 4153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6029230107913316e-05,
+      "loss": 0.0493,
+      "step": 4154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6027407332867848e-05,
+      "loss": 0.2007,
+      "step": 4155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6025584203401694e-05,
+      "loss": 1.1299,
+      "step": 4156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.602376071957345e-05,
+      "loss": 0.2981,
+      "step": 4157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.602193688144172e-05,
+      "loss": 1.1977,
+      "step": 4158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6020112689065122e-05,
+      "loss": 0.3035,
+      "step": 4159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6018288142502284e-05,
+      "loss": 0.4999,
+      "step": 4160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6016463241811853e-05,
+      "loss": 0.3048,
+      "step": 4161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6014637987052475e-05,
+      "loss": 0.3763,
+      "step": 4162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.601281237828282e-05,
+      "loss": 0.0474,
+      "step": 4163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6010986415561554e-05,
+      "loss": 0.7172,
+      "step": 4164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6009160098947372e-05,
+      "loss": 0.3929,
+      "step": 4165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6007333428498966e-05,
+      "loss": 1.4405,
+      "step": 4166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6005506404275046e-05,
+      "loss": 0.5007,
+      "step": 4167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6003679026334338e-05,
+      "loss": 0.0904,
+      "step": 4168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6001851294735568e-05,
+      "loss": 0.4397,
+      "step": 4169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.600002320953748e-05,
+      "loss": 0.4936,
+      "step": 4170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.599819477079883e-05,
+      "loss": 0.7145,
+      "step": 4171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5996365978578392e-05,
+      "loss": 1.0368,
+      "step": 4172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.599453683293493e-05,
+      "loss": 0.4601,
+      "step": 4173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.599270733392723e-05,
+      "loss": 0.7084,
+      "step": 4174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5990877481614107e-05,
+      "loss": 0.6567,
+      "step": 4175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5989047276054366e-05,
+      "loss": 0.3714,
+      "step": 4176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5987216717306827e-05,
+      "loss": 0.3263,
+      "step": 4177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5985385805430327e-05,
+      "loss": 0.3332,
+      "step": 4178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5983554540483708e-05,
+      "loss": 0.3024,
+      "step": 4179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5981722922525832e-05,
+      "loss": 0.622,
+      "step": 4180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5979890951615564e-05,
+      "loss": 1.732,
+      "step": 4181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5978058627811784e-05,
+      "loss": 1.1641,
+      "step": 4182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5976225951173387e-05,
+      "loss": 1.0767,
+      "step": 4183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5974392921759266e-05,
+      "loss": 0.1812,
+      "step": 4184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5972559539628343e-05,
+      "loss": 0.7359,
+      "step": 4185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.597072580483954e-05,
+      "loss": 1.4127,
+      "step": 4186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.596889171745179e-05,
+      "loss": 0.9438,
+      "step": 4187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5967057277524043e-05,
+      "loss": 0.5041,
+      "step": 4188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5965222485115264e-05,
+      "loss": 0.9037,
+      "step": 4189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.596338734028441e-05,
+      "loss": 0.7629,
+      "step": 4190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5961551843090478e-05,
+      "loss": 0.2315,
+      "step": 4191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.595971599359245e-05,
+      "loss": 0.7284,
+      "step": 4192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5957879791849327e-05,
+      "loss": 0.5981,
+      "step": 4193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5956043237920135e-05,
+      "loss": 0.5883,
+      "step": 4194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5954206331863893e-05,
+      "loss": 0.8175,
+      "step": 4195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.595236907373965e-05,
+      "loss": 0.4869,
+      "step": 4196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5950531463606445e-05,
+      "loss": 0.7583,
+      "step": 4197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5948693501523335e-05,
+      "loss": 0.1497,
+      "step": 4198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5946855187549406e-05,
+      "loss": 0.6752,
+      "step": 4199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5945016521743734e-05,
+      "loss": 0.4784,
+      "step": 4200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.594317750416541e-05,
+      "loss": 0.1876,
+      "step": 4201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.594133813487355e-05,
+      "loss": 0.1306,
+      "step": 4202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5939498413927257e-05,
+      "loss": 0.4528,
+      "step": 4203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5937658341385675e-05,
+      "loss": 1.4379,
+      "step": 4204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.593581791730793e-05,
+      "loss": 1.6238,
+      "step": 4205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5933977141753186e-05,
+      "loss": 1.465,
+      "step": 4206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5932136014780593e-05,
+      "loss": 0.479,
+      "step": 4207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.593029453644934e-05,
+      "loss": 0.3609,
+      "step": 4208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5928452706818596e-05,
+      "loss": 0.3396,
+      "step": 4209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5926610525947562e-05,
+      "loss": 0.4257,
+      "step": 4210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5924767993895456e-05,
+      "loss": 0.4263,
+      "step": 4211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5922925110721484e-05,
+      "loss": 1.4129,
+      "step": 4212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5921081876484882e-05,
+      "loss": 0.3603,
+      "step": 4213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5919238291244894e-05,
+      "loss": 0.2422,
+      "step": 4214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.591739435506077e-05,
+      "loss": 2.1341,
+      "step": 4215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5915550067991768e-05,
+      "loss": 0.1044,
+      "step": 4216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5913705430097173e-05,
+      "loss": 0.1289,
+      "step": 4217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5911860441436264e-05,
+      "loss": 0.125,
+      "step": 4218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5910015102068346e-05,
+      "loss": 0.268,
+      "step": 4219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5908169412052725e-05,
+      "loss": 1.5852,
+      "step": 4220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5906323371448724e-05,
+      "loss": 0.139,
+      "step": 4221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.590447698031567e-05,
+      "loss": 1.0212,
+      "step": 4222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5902630238712903e-05,
+      "loss": 0.2057,
+      "step": 4223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.590078314669979e-05,
+      "loss": 0.2545,
+      "step": 4224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5898935704335684e-05,
+      "loss": 0.9637,
+      "step": 4225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.589708791167997e-05,
+      "loss": 0.9147,
+      "step": 4226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.589523976879203e-05,
+      "loss": 0.4646,
+      "step": 4227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.589339127573126e-05,
+      "loss": 0.8501,
+      "step": 4228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5891542432557086e-05,
+      "loss": 1.0184,
+      "step": 4229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5889693239328917e-05,
+      "loss": 0.6107,
+      "step": 4230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5887843696106188e-05,
+      "loss": 0.6693,
+      "step": 4231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5885993802948345e-05,
+      "loss": 1.8793,
+      "step": 4232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5884143559914842e-05,
+      "loss": 0.2374,
+      "step": 4233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.588229296706515e-05,
+      "loss": 0.9127,
+      "step": 4234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5880442024458742e-05,
+      "loss": 0.4425,
+      "step": 4235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5878590732155106e-05,
+      "loss": 0.4609,
+      "step": 4236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.587673909021375e-05,
+      "loss": 0.4237,
+      "step": 4237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.587488709869418e-05,
+      "loss": 1.1005,
+      "step": 4238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5873034757655922e-05,
+      "loss": 0.1249,
+      "step": 4239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5871182067158506e-05,
+      "loss": 0.6054,
+      "step": 4240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.586932902726148e-05,
+      "loss": 0.8649,
+      "step": 4241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5867475638024404e-05,
+      "loss": 0.178,
+      "step": 4242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5865621899506834e-05,
+      "loss": 1.1926,
+      "step": 4243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.586376781176836e-05,
+      "loss": 0.6153,
+      "step": 4244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.586191337486858e-05,
+      "loss": 0.6987,
+      "step": 4245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5860058588867075e-05,
+      "loss": 0.6099,
+      "step": 4246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5858203453823472e-05,
+      "loss": 0.8908,
+      "step": 4247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5856347969797393e-05,
+      "loss": 0.2446,
+      "step": 4248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5854492136848472e-05,
+      "loss": 0.6263,
+      "step": 4249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.585263595503635e-05,
+      "loss": 0.718,
+      "step": 4250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5850779424420697e-05,
+      "loss": 0.2022,
+      "step": 4251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5848922545061168e-05,
+      "loss": 0.941,
+      "step": 4252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5847065317017452e-05,
+      "loss": 1.2478,
+      "step": 4253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.584520774034924e-05,
+      "loss": 0.4864,
+      "step": 4254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5843349815116225e-05,
+      "loss": 1.3611,
+      "step": 4255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5841491541378135e-05,
+      "loss": 0.3303,
+      "step": 4256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5839632919194686e-05,
+      "loss": 0.8272,
+      "step": 4257
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.583777394862562e-05,
+      "loss": 0.8891,
+      "step": 4258
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5835914629730672e-05,
+      "loss": 0.6782,
+      "step": 4259
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5834054962569615e-05,
+      "loss": 0.6281,
+      "step": 4260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.583219494720221e-05,
+      "loss": 0.5792,
+      "step": 4261
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.583033458368824e-05,
+      "loss": 0.2067,
+      "step": 4262
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5828473872087502e-05,
+      "loss": 0.3005,
+      "step": 4263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.582661281245979e-05,
+      "loss": 0.2412,
+      "step": 4264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5824751404864923e-05,
+      "loss": 0.1376,
+      "step": 4265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.582288964936273e-05,
+      "loss": 0.2025,
+      "step": 4266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5821027546013043e-05,
+      "loss": 0.9495,
+      "step": 4267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5819165094875712e-05,
+      "loss": 0.1994,
+      "step": 4268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.581730229601059e-05,
+      "loss": 1.2726,
+      "step": 4269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5815439149477558e-05,
+      "loss": 0.4207,
+      "step": 4270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5813575655336494e-05,
+      "loss": 1.6388,
+      "step": 4271
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5811711813647288e-05,
+      "loss": 1.2125,
+      "step": 4272
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5809847624469843e-05,
+      "loss": 0.6815,
+      "step": 4273
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5807983087864074e-05,
+      "loss": 0.908,
+      "step": 4274
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5806118203889913e-05,
+      "loss": 1.1844,
+      "step": 4275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.580425297260729e-05,
+      "loss": 1.8949,
+      "step": 4276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.580238739407616e-05,
+      "loss": 0.6802,
+      "step": 4277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.580052146835648e-05,
+      "loss": 0.9487,
+      "step": 4278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.579865519550822e-05,
+      "loss": 0.9436,
+      "step": 4279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5796788575591357e-05,
+      "loss": 0.3839,
+      "step": 4280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5794921608665896e-05,
+      "loss": 0.5621,
+      "step": 4281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5793054294791833e-05,
+      "loss": 0.8157,
+      "step": 4282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5791186634029185e-05,
+      "loss": 0.6706,
+      "step": 4283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5789318626437978e-05,
+      "loss": 0.2811,
+      "step": 4284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5787450272078255e-05,
+      "loss": 0.9431,
+      "step": 4285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5785581571010054e-05,
+      "loss": 0.8958,
+      "step": 4286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5783712523293446e-05,
+      "loss": 0.9515,
+      "step": 4287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5781843128988493e-05,
+      "loss": 0.4534,
+      "step": 4288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5779973388155284e-05,
+      "loss": 0.4798,
+      "step": 4289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.577810330085391e-05,
+      "loss": 0.622,
+      "step": 4290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.577623286714448e-05,
+      "loss": 0.5253,
+      "step": 4291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5774362087087104e-05,
+      "loss": 0.865,
+      "step": 4292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5772490960741906e-05,
+      "loss": 0.6773,
+      "step": 4293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5770619488169037e-05,
+      "loss": 0.4715,
+      "step": 4294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.576874766942863e-05,
+      "loss": 0.5304,
+      "step": 4295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5766875504580854e-05,
+      "loss": 0.955,
+      "step": 4296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.576500299368588e-05,
+      "loss": 0.5691,
+      "step": 4297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5763130136803888e-05,
+      "loss": 0.1771,
+      "step": 4298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.576125693399508e-05,
+      "loss": 0.188,
+      "step": 4299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5759383385319644e-05,
+      "loss": 0.2179,
+      "step": 4300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5757509490837815e-05,
+      "loss": 0.3088,
+      "step": 4301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5755635250609802e-05,
+      "loss": 0.1823,
+      "step": 4302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.575376066469586e-05,
+      "loss": 0.0501,
+      "step": 4303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.575188573315622e-05,
+      "loss": 1.044,
+      "step": 4304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.575001045605116e-05,
+      "loss": 0.3086,
+      "step": 4305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5748134833440943e-05,
+      "loss": 1.5181,
+      "step": 4306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.574625886538585e-05,
+      "loss": 0.7664,
+      "step": 4307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.574438255194618e-05,
+      "loss": 0.7659,
+      "step": 4308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.574250589318223e-05,
+      "loss": 0.567,
+      "step": 4309
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.574062888915432e-05,
+      "loss": 0.1675,
+      "step": 4310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.573875153992278e-05,
+      "loss": 0.4344,
+      "step": 4311
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.573687384554794e-05,
+      "loss": 1.5135,
+      "step": 4312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.573499580609016e-05,
+      "loss": 0.0398,
+      "step": 4313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5733117421609788e-05,
+      "loss": 0.63,
+      "step": 4314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5731238692167202e-05,
+      "loss": 0.7851,
+      "step": 4315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.572935961782279e-05,
+      "loss": 0.7461,
+      "step": 4316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5727480198636933e-05,
+      "loss": 0.2666,
+      "step": 4317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5725600434670044e-05,
+      "loss": 1.7926,
+      "step": 4318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5723720325982538e-05,
+      "loss": 0.7335,
+      "step": 4319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5721839872634835e-05,
+      "loss": 0.0962,
+      "step": 4320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.571995907468738e-05,
+      "loss": 1.1387,
+      "step": 4321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.571807793220062e-05,
+      "loss": 0.3693,
+      "step": 4322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5716196445235014e-05,
+      "loss": 0.7741,
+      "step": 4323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5714314613851035e-05,
+      "loss": 1.7291,
+      "step": 4324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5712432438109163e-05,
+      "loss": 0.5331,
+      "step": 4325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.571054991806989e-05,
+      "loss": 0.2004,
+      "step": 4326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.570866705379372e-05,
+      "loss": 0.3106,
+      "step": 4327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5706783845341168e-05,
+      "loss": 0.7888,
+      "step": 4328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5704900292772766e-05,
+      "loss": 0.3537,
+      "step": 4329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.570301639614905e-05,
+      "loss": 0.513,
+      "step": 4330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.570113215553056e-05,
+      "loss": 0.2549,
+      "step": 4331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5699247570977867e-05,
+      "loss": 0.8934,
+      "step": 4332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5697362642551528e-05,
+      "loss": 0.6849,
+      "step": 4333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.569547737031214e-05,
+      "loss": 1.1926,
+      "step": 4334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5693591754320288e-05,
+      "loss": 0.7101,
+      "step": 4335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.569170579463657e-05,
+      "loss": 0.7781,
+      "step": 4336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5689819491321608e-05,
+      "loss": 0.6594,
+      "step": 4337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5687932844436025e-05,
+      "loss": 0.9567,
+      "step": 4338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5686045854040465e-05,
+      "loss": 0.4145,
+      "step": 4339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5684158520195563e-05,
+      "loss": 1.6024,
+      "step": 4340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.568227084296199e-05,
+      "loss": 0.5289,
+      "step": 4341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5680382822400406e-05,
+      "loss": 1.0464,
+      "step": 4342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5678494458571496e-05,
+      "loss": 0.5094,
+      "step": 4343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5676605751535957e-05,
+      "loss": 0.6858,
+      "step": 4344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5674716701354485e-05,
+      "loss": 0.4887,
+      "step": 4345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5672827308087797e-05,
+      "loss": 0.4634,
+      "step": 4346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5670937571796618e-05,
+      "loss": 0.6547,
+      "step": 4347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.566904749254168e-05,
+      "loss": 0.5825,
+      "step": 4348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5667157070383738e-05,
+      "loss": 0.3323,
+      "step": 4349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5665266305383543e-05,
+      "loss": 0.7841,
+      "step": 4350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5663375197601874e-05,
+      "loss": 1.4664,
+      "step": 4351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5661483747099497e-05,
+      "loss": 1.2678,
+      "step": 4352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5659591953937212e-05,
+      "loss": 0.8637,
+      "step": 4353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5657699818175823e-05,
+      "loss": 0.5737,
+      "step": 4354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5655807339876132e-05,
+      "loss": 0.4458,
+      "step": 4355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.565391451909898e-05,
+      "loss": 0.223,
+      "step": 4356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5652021355905188e-05,
+      "loss": 0.3174,
+      "step": 4357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5650127850355608e-05,
+      "loss": 0.9147,
+      "step": 4358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.56482340025111e-05,
+      "loss": 0.0949,
+      "step": 4359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5646339812432528e-05,
+      "loss": 0.6644,
+      "step": 4360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.564444528018077e-05,
+      "loss": 0.1937,
+      "step": 4361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5642550405816718e-05,
+      "loss": 0.0568,
+      "step": 4362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5640655189401273e-05,
+      "loss": 0.5159,
+      "step": 4363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5638759630995355e-05,
+      "loss": 1.7363,
+      "step": 4364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5636863730659874e-05,
+      "loss": 0.6493,
+      "step": 4365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5634967488455773e-05,
+      "loss": 0.6254,
+      "step": 4366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.563307090444399e-05,
+      "loss": 1.4327,
+      "step": 4367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.563117397868549e-05,
+      "loss": 0.2127,
+      "step": 4368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5629276711241236e-05,
+      "loss": 0.1819,
+      "step": 4369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5627379102172205e-05,
+      "loss": 0.2626,
+      "step": 4370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5625481151539387e-05,
+      "loss": 0.358,
+      "step": 4371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5623582859403786e-05,
+      "loss": 0.223,
+      "step": 4372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5621684225826402e-05,
+      "loss": 0.0796,
+      "step": 4373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5619785250868266e-05,
+      "loss": 0.183,
+      "step": 4374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5617885934590414e-05,
+      "loss": 0.0611,
+      "step": 4375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5615986277053884e-05,
+      "loss": 0.7801,
+      "step": 4376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5614086278319733e-05,
+      "loss": 2.0288,
+      "step": 4377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5612185938449023e-05,
+      "loss": 0.2835,
+      "step": 4378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5610285257502837e-05,
+      "loss": 0.8613,
+      "step": 4379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5608384235542258e-05,
+      "loss": 0.2516,
+      "step": 4380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5606482872628386e-05,
+      "loss": 0.1182,
+      "step": 4381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5604581168822333e-05,
+      "loss": 1.1069,
+      "step": 4382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.560267912418522e-05,
+      "loss": 0.7802,
+      "step": 4383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5600776738778174e-05,
+      "loss": 0.4744,
+      "step": 4384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5598874012662345e-05,
+      "loss": 1.5663,
+      "step": 4385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.559697094589888e-05,
+      "loss": 0.4163,
+      "step": 4386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5595067538548947e-05,
+      "loss": 1.4316,
+      "step": 4387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5593163790673716e-05,
+      "loss": 0.3113,
+      "step": 4388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5591259702334385e-05,
+      "loss": 0.3276,
+      "step": 4389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.558935527359214e-05,
+      "loss": 1.1673,
+      "step": 4390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.558745050450819e-05,
+      "loss": 1.2558,
+      "step": 4391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.558554539514376e-05,
+      "loss": 1.056,
+      "step": 4392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5583639945560077e-05,
+      "loss": 1.0167,
+      "step": 4393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5581734155818384e-05,
+      "loss": 0.4374,
+      "step": 4394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5579828025979936e-05,
+      "loss": 0.8971,
+      "step": 4395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.557792155610599e-05,
+      "loss": 0.5767,
+      "step": 4396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5576014746257818e-05,
+      "loss": 0.3094,
+      "step": 4397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.557410759649671e-05,
+      "loss": 0.7753,
+      "step": 4398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5572200106883966e-05,
+      "loss": 1.161,
+      "step": 4399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.557029227748088e-05,
+      "loss": 0.5495,
+      "step": 4400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5568384108348775e-05,
+      "loss": 0.3581,
+      "step": 4401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.556647559954899e-05,
+      "loss": 0.3444,
+      "step": 4402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5564566751142854e-05,
+      "loss": 0.051,
+      "step": 4403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5562657563191713e-05,
+      "loss": 1.7048,
+      "step": 4404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.556074803575694e-05,
+      "loss": 0.0362,
+      "step": 4405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5558838168899895e-05,
+      "loss": 0.6995,
+      "step": 4406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5556927962681977e-05,
+      "loss": 0.2389,
+      "step": 4407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5555017417164565e-05,
+      "loss": 1.7882,
+      "step": 4408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.555310653240907e-05,
+      "loss": 1.8771,
+      "step": 4409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.555119530847691e-05,
+      "loss": 0.3037,
+      "step": 4410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5549283745429505e-05,
+      "loss": 0.3908,
+      "step": 4411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5547371843328296e-05,
+      "loss": 0.3165,
+      "step": 4412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5545459602234736e-05,
+      "loss": 0.1793,
+      "step": 4413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5543547022210277e-05,
+      "loss": 1.2169,
+      "step": 4414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5541634103316392e-05,
+      "loss": 0.4845,
+      "step": 4415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.553972084561457e-05,
+      "loss": 0.2007,
+      "step": 4416
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5537807249166292e-05,
+      "loss": 1.6198,
+      "step": 4417
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5535893314033062e-05,
+      "loss": 1.3457,
+      "step": 4418
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5533979040276402e-05,
+      "loss": 0.264,
+      "step": 4419
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5532064427957827e-05,
+      "loss": 0.9057,
+      "step": 4420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.553014947713888e-05,
+      "loss": 0.1631,
+      "step": 4421
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5528234187881103e-05,
+      "loss": 0.2031,
+      "step": 4422
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5526318560246056e-05,
+      "loss": 0.4829,
+      "step": 4423
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.552440259429531e-05,
+      "loss": 0.129,
+      "step": 4424
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5522486290090434e-05,
+      "loss": 0.2288,
+      "step": 4425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.552056964769303e-05,
+      "loss": 0.2271,
+      "step": 4426
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5518652667164687e-05,
+      "loss": 0.3031,
+      "step": 4427
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5516735348567026e-05,
+      "loss": 0.5653,
+      "step": 4428
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.551481769196167e-05,
+      "loss": 0.3284,
+      "step": 4429
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5512899697410247e-05,
+      "loss": 0.5258,
+      "step": 4430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.55109813649744e-05,
+      "loss": 0.5121,
+      "step": 4431
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.550906269471579e-05,
+      "loss": 1.1128,
+      "step": 4432
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.550714368669608e-05,
+      "loss": 0.4333,
+      "step": 4433
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5505224340976948e-05,
+      "loss": 0.5433,
+      "step": 4434
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.550330465762008e-05,
+      "loss": 1.0185,
+      "step": 4435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5501384636687178e-05,
+      "loss": 0.115,
+      "step": 4436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.549946427823995e-05,
+      "loss": 1.3788,
+      "step": 4437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.549754358234011e-05,
+      "loss": 0.724,
+      "step": 4438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5495622549049403e-05,
+      "loss": 1.6613,
+      "step": 4439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5493701178429556e-05,
+      "loss": 1.127,
+      "step": 4440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5491779470542328e-05,
+      "loss": 0.4802,
+      "step": 4441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5489857425449485e-05,
+      "loss": 0.9461,
+      "step": 4442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5487935043212798e-05,
+      "loss": 0.652,
+      "step": 4443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5486012323894058e-05,
+      "loss": 0.4951,
+      "step": 4444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5484089267555054e-05,
+      "loss": 1.2877,
+      "step": 4445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5482165874257598e-05,
+      "loss": 0.8394,
+      "step": 4446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.54802421440635e-05,
+      "loss": 0.7758,
+      "step": 4447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5478318077034603e-05,
+      "loss": 0.7584,
+      "step": 4448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5476393673232733e-05,
+      "loss": 0.3297,
+      "step": 4449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5474468932719746e-05,
+      "loss": 0.5991,
+      "step": 4450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5472543855557502e-05,
+      "loss": 0.231,
+      "step": 4451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5470618441807878e-05,
+      "loss": 0.2504,
+      "step": 4452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5468692691532745e-05,
+      "loss": 1.5666,
+      "step": 4453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.546676660479401e-05,
+      "loss": 0.133,
+      "step": 4454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5464840181653568e-05,
+      "loss": 1.4883,
+      "step": 4455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.546291342217334e-05,
+      "loss": 0.2328,
+      "step": 4456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5460986326415243e-05,
+      "loss": 0.9778,
+      "step": 4457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5459058894441225e-05,
+      "loss": 0.0837,
+      "step": 4458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5457131126313228e-05,
+      "loss": 1.025,
+      "step": 4459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5455203022093213e-05,
+      "loss": 0.0567,
+      "step": 4460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5453274581843145e-05,
+      "loss": 1.4757,
+      "step": 4461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.545134580562501e-05,
+      "loss": 1.387,
+      "step": 4462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.544941669350079e-05,
+      "loss": 1.0505,
+      "step": 4463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5447487245532498e-05,
+      "loss": 0.1368,
+      "step": 4464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5445557461782134e-05,
+      "loss": 0.6217,
+      "step": 4465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5443627342311735e-05,
+      "loss": 0.2253,
+      "step": 4466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.544169688718332e-05,
+      "loss": 1.1412,
+      "step": 4467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5439766096458947e-05,
+      "loss": 0.3187,
+      "step": 4468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5437834970200664e-05,
+      "loss": 1.0812,
+      "step": 4469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5435903508470537e-05,
+      "loss": 0.2499,
+      "step": 4470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.543397171133065e-05,
+      "loss": 0.7854,
+      "step": 4471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5432039578843085e-05,
+      "loss": 0.496,
+      "step": 4472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.543010711106994e-05,
+      "loss": 0.6691,
+      "step": 4473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5428174308073327e-05,
+      "loss": 2.1734,
+      "step": 4474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.542624116991537e-05,
+      "loss": 0.6681,
+      "step": 4475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.542430769665819e-05,
+      "loss": 1.4507,
+      "step": 4476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5422373888363942e-05,
+      "loss": 0.6554,
+      "step": 4477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5420439745094766e-05,
+      "loss": 1.5305,
+      "step": 4478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5418505266912832e-05,
+      "loss": 1.6928,
+      "step": 4479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5416570453880312e-05,
+      "loss": 1.6902,
+      "step": 4480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5414635306059392e-05,
+      "loss": 0.5797,
+      "step": 4481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.541269982351227e-05,
+      "loss": 1.7033,
+      "step": 4482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5410764006301148e-05,
+      "loss": 1.2116,
+      "step": 4483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.540882785448824e-05,
+      "loss": 0.739,
+      "step": 4484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5406891368135787e-05,
+      "loss": 0.4127,
+      "step": 4485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.540495454730601e-05,
+      "loss": 0.2372,
+      "step": 4486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5403017392061174e-05,
+      "loss": 0.5875,
+      "step": 4487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5401079902463534e-05,
+      "loss": 0.9611,
+      "step": 4488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5399142078575357e-05,
+      "loss": 0.6059,
+      "step": 4489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5397203920458926e-05,
+      "loss": 0.5132,
+      "step": 4490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5395265428176535e-05,
+      "loss": 1.1682,
+      "step": 4491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.539332660179049e-05,
+      "loss": 0.973,
+      "step": 4492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5391387441363095e-05,
+      "loss": 0.7275,
+      "step": 4493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.538944794695669e-05,
+      "loss": 0.458,
+      "step": 4494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5387508118633594e-05,
+      "loss": 0.6716,
+      "step": 4495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5385567956456166e-05,
+      "loss": 0.8871,
+      "step": 4496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.538362746048675e-05,
+      "loss": 0.8188,
+      "step": 4497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.538168663078773e-05,
+      "loss": 0.3849,
+      "step": 4498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5379745467421468e-05,
+      "loss": 1.1269,
+      "step": 4499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5377803970450365e-05,
+      "loss": 0.2927,
+      "step": 4500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.537586213993681e-05,
+      "loss": 0.0727,
+      "step": 4501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5373919975943227e-05,
+      "loss": 0.2949,
+      "step": 4502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5371977478532027e-05,
+      "loss": 0.0528,
+      "step": 4503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.537003464776564e-05,
+      "loss": 1.0265,
+      "step": 4504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5368091483706514e-05,
+      "loss": 0.2196,
+      "step": 4505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.53661479864171e-05,
+      "loss": 0.2471,
+      "step": 4506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5364204155959868e-05,
+      "loss": 0.7654,
+      "step": 4507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5362259992397283e-05,
+      "loss": 0.8994,
+      "step": 4508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5360315495791836e-05,
+      "loss": 1.0782,
+      "step": 4509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.535837066620602e-05,
+      "loss": 0.8901,
+      "step": 4510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5356425503702347e-05,
+      "loss": 0.6268,
+      "step": 4511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5354480008343333e-05,
+      "loss": 0.1396,
+      "step": 4512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.53525341801915e-05,
+      "loss": 0.4704,
+      "step": 4513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.535058801930939e-05,
+      "loss": 0.5974,
+      "step": 4514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5348641525759563e-05,
+      "loss": 0.6008,
+      "step": 4515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.534669469960456e-05,
+      "loss": 0.314,
+      "step": 4516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5344747540906966e-05,
+      "loss": 0.4442,
+      "step": 4517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.534280004972936e-05,
+      "loss": 0.1367,
+      "step": 4518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5340852226134333e-05,
+      "loss": 0.4197,
+      "step": 4519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5338904070184486e-05,
+      "loss": 1.2371,
+      "step": 4520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5336955581942435e-05,
+      "loss": 0.2407,
+      "step": 4521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.533500676147081e-05,
+      "loss": 0.731,
+      "step": 4522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5333057608832235e-05,
+      "loss": 0.1759,
+      "step": 4523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5331108124089362e-05,
+      "loss": 0.2613,
+      "step": 4524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5329158307304847e-05,
+      "loss": 0.6734,
+      "step": 4525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5327208158541355e-05,
+      "loss": 0.6836,
+      "step": 4526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5325257677861567e-05,
+      "loss": 0.62,
+      "step": 4527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5323306865328166e-05,
+      "loss": 0.7796,
+      "step": 4528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.532135572100386e-05,
+      "loss": 0.1234,
+      "step": 4529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.531940424495135e-05,
+      "loss": 0.261,
+      "step": 4530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.531745243723336e-05,
+      "loss": 0.3812,
+      "step": 4531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5315500297912626e-05,
+      "loss": 0.8964,
+      "step": 4532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.531354782705188e-05,
+      "loss": 0.3575,
+      "step": 4533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.531159502471388e-05,
+      "loss": 1.0682,
+      "step": 4534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.530964189096139e-05,
+      "loss": 0.3378,
+      "step": 4535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5307688425857177e-05,
+      "loss": 0.4523,
+      "step": 4536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.530573462946403e-05,
+      "loss": 0.9167,
+      "step": 4537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5303780501844753e-05,
+      "loss": 0.3286,
+      "step": 4538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.530182604306214e-05,
+      "loss": 0.955,
+      "step": 4539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5299871253179e-05,
+      "loss": 0.5115,
+      "step": 4540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5297916132258182e-05,
+      "loss": 0.2935,
+      "step": 4541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5295960680362505e-05,
+      "loss": 1.0165,
+      "step": 4542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5294004897554823e-05,
+      "loss": 1.0888,
+      "step": 4543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5292048783898e-05,
+      "loss": 1.5653,
+      "step": 4544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.52900923394549e-05,
+      "loss": 0.4616,
+      "step": 4545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.52881355642884e-05,
+      "loss": 0.3533,
+      "step": 4546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5286178458461402e-05,
+      "loss": 1.2022,
+      "step": 4547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5284221022036794e-05,
+      "loss": 1.0067,
+      "step": 4548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5282263255077496e-05,
+      "loss": 0.466,
+      "step": 4549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5280305157646434e-05,
+      "loss": 0.2479,
+      "step": 4550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.527834672980653e-05,
+      "loss": 0.0521,
+      "step": 4551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.527638797162073e-05,
+      "loss": 0.0449,
+      "step": 4552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5274428883151997e-05,
+      "loss": 0.3798,
+      "step": 4553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5272469464463294e-05,
+      "loss": 0.0908,
+      "step": 4554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.527050971561759e-05,
+      "loss": 0.0784,
+      "step": 4555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5268549636677872e-05,
+      "loss": 0.3202,
+      "step": 4556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.526658922770715e-05,
+      "loss": 0.4308,
+      "step": 4557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5264628488768418e-05,
+      "loss": 0.8314,
+      "step": 4558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5262667419924692e-05,
+      "loss": 0.1719,
+      "step": 4559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5260706021239012e-05,
+      "loss": 0.6746,
+      "step": 4560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.525874429277441e-05,
+      "loss": 0.2518,
+      "step": 4561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.525678223459394e-05,
+      "loss": 1.3355,
+      "step": 4562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5254819846760656e-05,
+      "loss": 0.3937,
+      "step": 4563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5252857129337637e-05,
+      "loss": 0.7261,
+      "step": 4564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5250894082387963e-05,
+      "loss": 0.5486,
+      "step": 4565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.524893070597472e-05,
+      "loss": 0.7781,
+      "step": 4566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5246967000161018e-05,
+      "loss": 0.6576,
+      "step": 4567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5245002965009965e-05,
+      "loss": 0.267,
+      "step": 4568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.524303860058469e-05,
+      "loss": 1.5481,
+      "step": 4569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.524107390694832e-05,
+      "loss": 0.4695,
+      "step": 4570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5239108884164018e-05,
+      "loss": 0.1349,
+      "step": 4571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5237143532294915e-05,
+      "loss": 0.2815,
+      "step": 4572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5235177851404195e-05,
+      "loss": 1.6757,
+      "step": 4573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.523321184155503e-05,
+      "loss": 1.816,
+      "step": 4574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.523124550281061e-05,
+      "loss": 0.2957,
+      "step": 4575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.522927883523413e-05,
+      "loss": 0.5827,
+      "step": 4576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5227311838888795e-05,
+      "loss": 0.8023,
+      "step": 4577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.522534451383783e-05,
+      "loss": 0.2944,
+      "step": 4578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5223376860144462e-05,
+      "loss": 0.9918,
+      "step": 4579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5221408877871934e-05,
+      "loss": 0.1488,
+      "step": 4580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5219440567083497e-05,
+      "loss": 1.4877,
+      "step": 4581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5217471927842405e-05,
+      "loss": 1.1392,
+      "step": 4582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.521550296021194e-05,
+      "loss": 0.7327,
+      "step": 4583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5213533664255382e-05,
+      "loss": 1.2791,
+      "step": 4584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.521156404003602e-05,
+      "loss": 0.6456,
+      "step": 4585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5209594087617163e-05,
+      "loss": 0.4503,
+      "step": 4586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5207623807062122e-05,
+      "loss": 1.21,
+      "step": 4587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5205653198434222e-05,
+      "loss": 0.5844,
+      "step": 4588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5203682261796798e-05,
+      "loss": 0.5425,
+      "step": 4589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5201710997213194e-05,
+      "loss": 0.2245,
+      "step": 4590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5199739404746774e-05,
+      "loss": 0.3111,
+      "step": 4591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5197767484460895e-05,
+      "loss": 1.4099,
+      "step": 4592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5195795236418944e-05,
+      "loss": 1.3248,
+      "step": 4593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.51938226606843e-05,
+      "loss": 0.6774,
+      "step": 4594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5191849757320366e-05,
+      "loss": 0.7811,
+      "step": 4595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5189876526390555e-05,
+      "loss": 0.9483,
+      "step": 4596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.518790296795828e-05,
+      "loss": 0.8047,
+      "step": 4597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5185929082086978e-05,
+      "loss": 0.5656,
+      "step": 4598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.518395486884008e-05,
+      "loss": 0.4276,
+      "step": 4599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5181980328281045e-05,
+      "loss": 0.4763,
+      "step": 4600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.518000546047333e-05,
+      "loss": 1.6527,
+      "step": 4601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5178030265480414e-05,
+      "loss": 1.8072,
+      "step": 4602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5176054743365773e-05,
+      "loss": 1.5029,
+      "step": 4603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5174078894192902e-05,
+      "loss": 0.5336,
+      "step": 4604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5172102718025304e-05,
+      "loss": 0.8606,
+      "step": 4605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5170126214926495e-05,
+      "loss": 0.2798,
+      "step": 4606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5168149384960003e-05,
+      "loss": 0.3456,
+      "step": 4607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5166172228189358e-05,
+      "loss": 0.5287,
+      "step": 4608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5164194744678105e-05,
+      "loss": 1.2625,
+      "step": 4609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5162216934489802e-05,
+      "loss": 0.446,
+      "step": 4610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5160238797688017e-05,
+      "loss": 0.5605,
+      "step": 4611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.515826033433633e-05,
+      "loss": 0.3418,
+      "step": 4612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.515628154449832e-05,
+      "loss": 0.3217,
+      "step": 4613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.51543024282376e-05,
+      "loss": 1.2543,
+      "step": 4614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5152322985617763e-05,
+      "loss": 0.568,
+      "step": 4615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5150343216702432e-05,
+      "loss": 0.1971,
+      "step": 4616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5148363121555245e-05,
+      "loss": 0.6054,
+      "step": 4617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.514638270023984e-05,
+      "loss": 1.7146,
+      "step": 4618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5144401952819856e-05,
+      "loss": 0.8123,
+      "step": 4619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5142420879358967e-05,
+      "loss": 0.7251,
+      "step": 4620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5140439479920842e-05,
+      "loss": 0.7162,
+      "step": 4621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.513845775456916e-05,
+      "loss": 1.5002,
+      "step": 4622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5136475703367615e-05,
+      "loss": 1.199,
+      "step": 4623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5134493326379906e-05,
+      "loss": 0.6329,
+      "step": 4624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5132510623669757e-05,
+      "loss": 1.2481,
+      "step": 4625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5130527595300884e-05,
+      "loss": 0.5012,
+      "step": 4626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5128544241337022e-05,
+      "loss": 0.3395,
+      "step": 4627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5126560561841916e-05,
+      "loss": 0.8706,
+      "step": 4628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.512457655687933e-05,
+      "loss": 0.7668,
+      "step": 4629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5122592226513016e-05,
+      "loss": 0.6443,
+      "step": 4630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.512060757080676e-05,
+      "loss": 0.3184,
+      "step": 4631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.511862258982434e-05,
+      "loss": 0.8258,
+      "step": 4632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.511663728362956e-05,
+      "loss": 0.6699,
+      "step": 4633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.511465165228623e-05,
+      "loss": 0.5381,
+      "step": 4634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5112665695858168e-05,
+      "loss": 1.4967,
+      "step": 4635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.511067941440919e-05,
+      "loss": 0.2435,
+      "step": 4636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5108692808003154e-05,
+      "loss": 0.763,
+      "step": 4637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5106705876703896e-05,
+      "loss": 0.723,
+      "step": 4638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.510471862057528e-05,
+      "loss": 0.8889,
+      "step": 4639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5102731039681177e-05,
+      "loss": 0.1168,
+      "step": 4640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5100743134085467e-05,
+      "loss": 0.2109,
+      "step": 4641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.509875490385204e-05,
+      "loss": 0.4597,
+      "step": 4642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5096766349044802e-05,
+      "loss": 0.784,
+      "step": 4643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.509477746972766e-05,
+      "loss": 0.6621,
+      "step": 4644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5092788265964536e-05,
+      "loss": 0.7674,
+      "step": 4645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.509079873781937e-05,
+      "loss": 0.7008,
+      "step": 4646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5088808885356097e-05,
+      "loss": 0.6668,
+      "step": 4647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5086818708638678e-05,
+      "loss": 0.6494,
+      "step": 4648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5084828207731077e-05,
+      "loss": 0.3464,
+      "step": 4649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5082837382697266e-05,
+      "loss": 0.693,
+      "step": 4650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5080846233601222e-05,
+      "loss": 0.8045,
+      "step": 4651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.507885476050696e-05,
+      "loss": 1.0335,
+      "step": 4652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5076862963478463e-05,
+      "loss": 0.5426,
+      "step": 4653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5074870842579764e-05,
+      "loss": 0.3353,
+      "step": 4654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.507287839787488e-05,
+      "loss": 0.5742,
+      "step": 4655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.507088562942786e-05,
+      "loss": 0.7128,
+      "step": 4656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.506889253730274e-05,
+      "loss": 0.6365,
+      "step": 4657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.506689912156358e-05,
+      "loss": 1.5408,
+      "step": 4658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5064905382274455e-05,
+      "loss": 0.8382,
+      "step": 4659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5062911319499435e-05,
+      "loss": 0.5409,
+      "step": 4660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.506091693330262e-05,
+      "loss": 1.3025,
+      "step": 4661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.505892222374809e-05,
+      "loss": 0.0839,
+      "step": 4662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.505692719089998e-05,
+      "loss": 1.0409,
+      "step": 4663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5054931834822392e-05,
+      "loss": 0.1364,
+      "step": 4664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5052936155579468e-05,
+      "loss": 0.7032,
+      "step": 4665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5050940153235335e-05,
+      "loss": 0.8278,
+      "step": 4666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.504894382785416e-05,
+      "loss": 0.7189,
+      "step": 4667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.50469471795001e-05,
+      "loss": 0.7012,
+      "step": 4668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.504495020823732e-05,
+      "loss": 0.6767,
+      "step": 4669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.504295291413001e-05,
+      "loss": 1.0425,
+      "step": 4670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5040955297242364e-05,
+      "loss": 1.0947,
+      "step": 4671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.503895735763858e-05,
+      "loss": 0.9514,
+      "step": 4672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5036959095382875e-05,
+      "loss": 0.7799,
+      "step": 4673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.503496051053947e-05,
+      "loss": 0.6415,
+      "step": 4674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5032961603172605e-05,
+      "loss": 0.2593,
+      "step": 4675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.503096237334652e-05,
+      "loss": 1.5211,
+      "step": 4676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5028962821125472e-05,
+      "loss": 0.9849,
+      "step": 4677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5026962946573724e-05,
+      "loss": 0.9845,
+      "step": 4678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.502496274975556e-05,
+      "loss": 0.8051,
+      "step": 4679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5022962230735258e-05,
+      "loss": 0.4438,
+      "step": 4680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5020961389577117e-05,
+      "loss": 0.5701,
+      "step": 4681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.501896022634544e-05,
+      "loss": 0.5007,
+      "step": 4682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5016958741104555e-05,
+      "loss": 0.7516,
+      "step": 4683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.501495693391878e-05,
+      "loss": 0.7013,
+      "step": 4684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5012954804852458e-05,
+      "loss": 0.5256,
+      "step": 4685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5010952353969936e-05,
+      "loss": 1.7263,
+      "step": 4686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5008949581335576e-05,
+      "loss": 0.1078,
+      "step": 4687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5006946487013735e-05,
+      "loss": 0.5749,
+      "step": 4688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5004943071068808e-05,
+      "loss": 0.4671,
+      "step": 4689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5002939333565174e-05,
+      "loss": 0.2889,
+      "step": 4690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.500093527456724e-05,
+      "loss": 0.6828,
+      "step": 4691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.499893089413941e-05,
+      "loss": 0.713,
+      "step": 4692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.499692619234611e-05,
+      "loss": 1.0146,
+      "step": 4693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.499492116925177e-05,
+      "loss": 0.8008,
+      "step": 4694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.499291582492083e-05,
+      "loss": 0.6564,
+      "step": 4695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.499091015941774e-05,
+      "loss": 0.1616,
+      "step": 4696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4988904172806964e-05,
+      "loss": 0.5911,
+      "step": 4697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.498689786515298e-05,
+      "loss": 0.6387,
+      "step": 4698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4984891236520263e-05,
+      "loss": 0.4933,
+      "step": 4699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4982884286973306e-05,
+      "loss": 0.2573,
+      "step": 4700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4980877016576615e-05,
+      "loss": 0.4888,
+      "step": 4701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4978869425394706e-05,
+      "loss": 0.0421,
+      "step": 4702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.49768615134921e-05,
+      "loss": 1.5294,
+      "step": 4703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4974853280933332e-05,
+      "loss": 0.2454,
+      "step": 4704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4972844727782946e-05,
+      "loss": 1.1026,
+      "step": 4705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4970835854105493e-05,
+      "loss": 0.1411,
+      "step": 4706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.496882665996555e-05,
+      "loss": 1.332,
+      "step": 4707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.496681714542768e-05,
+      "loss": 0.6658,
+      "step": 4708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.496480731055647e-05,
+      "loss": 0.1387,
+      "step": 4709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4962797155416522e-05,
+      "loss": 1.7728,
+      "step": 4710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.496078668007244e-05,
+      "loss": 0.6568,
+      "step": 4711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4958775884588844e-05,
+      "loss": 0.4682,
+      "step": 4712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4956764769030353e-05,
+      "loss": 0.8054,
+      "step": 4713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.495475333346161e-05,
+      "loss": 1.6298,
+      "step": 4714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4952741577947264e-05,
+      "loss": 0.7117,
+      "step": 4715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4950729502551963e-05,
+      "loss": 0.8509,
+      "step": 4716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4948717107340386e-05,
+      "loss": 0.9394,
+      "step": 4717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4946704392377212e-05,
+      "loss": 0.0562,
+      "step": 4718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4944691357727117e-05,
+      "loss": 0.0378,
+      "step": 4719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4942678003454814e-05,
+      "loss": 0.652,
+      "step": 4720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4940664329625004e-05,
+      "loss": 0.4098,
+      "step": 4721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.493865033630241e-05,
+      "loss": 0.1041,
+      "step": 4722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4936636023551757e-05,
+      "loss": 0.379,
+      "step": 4723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.493462139143779e-05,
+      "loss": 0.1162,
+      "step": 4724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4932606440025256e-05,
+      "loss": 0.0912,
+      "step": 4725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4930591169378915e-05,
+      "loss": 2.2302,
+      "step": 4726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4928575579563543e-05,
+      "loss": 1.9052,
+      "step": 4727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4926559670643917e-05,
+      "loss": 0.8532,
+      "step": 4728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4924543442684832e-05,
+      "loss": 0.4394,
+      "step": 4729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4922526895751082e-05,
+      "loss": 1.1046,
+      "step": 4730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4920510029907487e-05,
+      "loss": 0.6945,
+      "step": 4731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4918492845218864e-05,
+      "loss": 0.1603,
+      "step": 4732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.491647534175004e-05,
+      "loss": 0.6995,
+      "step": 4733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4914457519565875e-05,
+      "loss": 0.8699,
+      "step": 4734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4912439378731208e-05,
+      "loss": 1.3742,
+      "step": 4735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.49104209193109e-05,
+      "loss": 0.3303,
+      "step": 4736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4908402141369832e-05,
+      "loss": 0.459,
+      "step": 4737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4906383044972885e-05,
+      "loss": 1.3422,
+      "step": 4738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.490436363018495e-05,
+      "loss": 0.1968,
+      "step": 4739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4902343897070936e-05,
+      "loss": 0.1021,
+      "step": 4740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4900323845695756e-05,
+      "loss": 1.0175,
+      "step": 4741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4898303476124335e-05,
+      "loss": 1.1535,
+      "step": 4742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4896282788421596e-05,
+      "loss": 0.319,
+      "step": 4743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48942617826525e-05,
+      "loss": 0.5945,
+      "step": 4744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4892240458881998e-05,
+      "loss": 0.5548,
+      "step": 4745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.489021881717505e-05,
+      "loss": 0.6298,
+      "step": 4746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.488819685759663e-05,
+      "loss": 1.4082,
+      "step": 4747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4886174580211736e-05,
+      "loss": 0.5904,
+      "step": 4748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4884151985085352e-05,
+      "loss": 0.3197,
+      "step": 4749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4882129072282488e-05,
+      "loss": 0.4661,
+      "step": 4750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.488010584186816e-05,
+      "loss": 0.091,
+      "step": 4751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4878082293907398e-05,
+      "loss": 0.2003,
+      "step": 4752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4876058428465232e-05,
+      "loss": 1.0787,
+      "step": 4753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.487403424560672e-05,
+      "loss": 0.0698,
+      "step": 4754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4872009745396905e-05,
+      "loss": 0.2721,
+      "step": 4755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4869984927900864e-05,
+      "loss": 0.1824,
+      "step": 4756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4867959793183676e-05,
+      "loss": 0.7771,
+      "step": 4757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.486593434131042e-05,
+      "loss": 0.3707,
+      "step": 4758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4863908572346203e-05,
+      "loss": 0.105,
+      "step": 4759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4861882486356124e-05,
+      "loss": 0.3041,
+      "step": 4760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4859856083405314e-05,
+      "loss": 1.4803,
+      "step": 4761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.485782936355889e-05,
+      "loss": 0.0544,
+      "step": 4762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4855802326881994e-05,
+      "loss": 0.2329,
+      "step": 4763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4853774973439775e-05,
+      "loss": 0.2905,
+      "step": 4764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48517473032974e-05,
+      "loss": 0.2585,
+      "step": 4765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4849719316520023e-05,
+      "loss": 0.5614,
+      "step": 4766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4847691013172834e-05,
+      "loss": 0.1814,
+      "step": 4767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4845662393321024e-05,
+      "loss": 0.7359,
+      "step": 4768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4843633457029787e-05,
+      "loss": 0.3041,
+      "step": 4769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.484160420436433e-05,
+      "loss": 0.6988,
+      "step": 4770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4839574635389884e-05,
+      "loss": 0.8458,
+      "step": 4771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4837544750171675e-05,
+      "loss": 0.5469,
+      "step": 4772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4835514548774938e-05,
+      "loss": 0.1313,
+      "step": 4773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4833484031264926e-05,
+      "loss": 0.6956,
+      "step": 4774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4831453197706905e-05,
+      "loss": 0.4346,
+      "step": 4775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4829422048166146e-05,
+      "loss": 1.3215,
+      "step": 4776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4827390582707922e-05,
+      "loss": 1.0025,
+      "step": 4777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4825358801397526e-05,
+      "loss": 0.4411,
+      "step": 4778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4823326704300267e-05,
+      "loss": 0.8853,
+      "step": 4779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4821294291481448e-05,
+      "loss": 0.1134,
+      "step": 4780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48192615630064e-05,
+      "loss": 0.3887,
+      "step": 4781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4817228518940444e-05,
+      "loss": 0.8983,
+      "step": 4782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4815195159348933e-05,
+      "loss": 1.0017,
+      "step": 4783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4813161484297208e-05,
+      "loss": 0.6703,
+      "step": 4784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.481112749385064e-05,
+      "loss": 0.4984,
+      "step": 4785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4809093188074596e-05,
+      "loss": 0.7706,
+      "step": 4786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4807058567034463e-05,
+      "loss": 0.4931,
+      "step": 4787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.480502363079563e-05,
+      "loss": 0.8284,
+      "step": 4788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48029883794235e-05,
+      "loss": 0.269,
+      "step": 4789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4800952812983494e-05,
+      "loss": 0.8691,
+      "step": 4790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.479891693154102e-05,
+      "loss": 0.4373,
+      "step": 4791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4796880735161523e-05,
+      "loss": 1.2804,
+      "step": 4792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.479484422391044e-05,
+      "loss": 0.7122,
+      "step": 4793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.479280739785323e-05,
+      "loss": 0.5251,
+      "step": 4794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4790770257055358e-05,
+      "loss": 0.2236,
+      "step": 4795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4788732801582287e-05,
+      "loss": 0.8585,
+      "step": 4796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.478669503149951e-05,
+      "loss": 0.3185,
+      "step": 4797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4784656946872517e-05,
+      "loss": 0.3783,
+      "step": 4798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4782618547766817e-05,
+      "loss": 0.2707,
+      "step": 4799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.478057983424791e-05,
+      "loss": 0.5609,
+      "step": 4800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.477854080638134e-05,
+      "loss": 0.0599,
+      "step": 4801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4776501464232628e-05,
+      "loss": 1.3556,
+      "step": 4802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4774461807867324e-05,
+      "loss": 0.3454,
+      "step": 4803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.477242183735098e-05,
+      "loss": 0.3578,
+      "step": 4804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4770381552749162e-05,
+      "loss": 0.9837,
+      "step": 4805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4768340954127447e-05,
+      "loss": 1.8487,
+      "step": 4806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.476630004155141e-05,
+      "loss": 0.4849,
+      "step": 4807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.476425881508666e-05,
+      "loss": 0.4704,
+      "step": 4808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.476221727479879e-05,
+      "loss": 0.7163,
+      "step": 4809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.476017542075342e-05,
+      "loss": 0.2856,
+      "step": 4810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.475813325301618e-05,
+      "loss": 0.4531,
+      "step": 4811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4756090771652694e-05,
+      "loss": 0.306,
+      "step": 4812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4754047976728615e-05,
+      "loss": 0.4396,
+      "step": 4813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.47520048683096e-05,
+      "loss": 1.1833,
+      "step": 4814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4749961446461307e-05,
+      "loss": 0.2397,
+      "step": 4815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.474791771124942e-05,
+      "loss": 1.1447,
+      "step": 4816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.474587366273962e-05,
+      "loss": 0.8105,
+      "step": 4817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.47438293009976e-05,
+      "loss": 0.2698,
+      "step": 4818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4741784626089072e-05,
+      "loss": 1.4142,
+      "step": 4819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4739739638079744e-05,
+      "loss": 1.5441,
+      "step": 4820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4737694337035354e-05,
+      "loss": 1.2469,
+      "step": 4821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.473564872302163e-05,
+      "loss": 0.4098,
+      "step": 4822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4733602796104314e-05,
+      "loss": 0.7512,
+      "step": 4823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4731556556349168e-05,
+      "loss": 1.6334,
+      "step": 4824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4729510003821957e-05,
+      "loss": 0.4641,
+      "step": 4825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.472746313858846e-05,
+      "loss": 0.5807,
+      "step": 4826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4725415960714455e-05,
+      "loss": 0.2202,
+      "step": 4827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4723368470265744e-05,
+      "loss": 0.1991,
+      "step": 4828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4721320667308136e-05,
+      "loss": 0.2032,
+      "step": 4829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4719272551907447e-05,
+      "loss": 0.3307,
+      "step": 4830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4717224124129494e-05,
+      "loss": 0.2494,
+      "step": 4831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4715175384040122e-05,
+      "loss": 1.5403,
+      "step": 4832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4713126331705177e-05,
+      "loss": 0.4514,
+      "step": 4833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4711076967190513e-05,
+      "loss": 1.0663,
+      "step": 4834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4709027290562e-05,
+      "loss": 0.5284,
+      "step": 4835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4706977301885507e-05,
+      "loss": 0.2585,
+      "step": 4836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.470492700122693e-05,
+      "loss": 0.3144,
+      "step": 4837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.470287638865216e-05,
+      "loss": 1.0558,
+      "step": 4838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.47008254642271e-05,
+      "loss": 0.6058,
+      "step": 4839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4698774228017675e-05,
+      "loss": 1.4791,
+      "step": 4840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4696722680089808e-05,
+      "loss": 1.038,
+      "step": 4841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.469467082050944e-05,
+      "loss": 0.3661,
+      "step": 4842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4692618649342512e-05,
+      "loss": 0.3517,
+      "step": 4843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.469056616665498e-05,
+      "loss": 0.6686,
+      "step": 4844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4688513372512816e-05,
+      "loss": 0.4494,
+      "step": 4845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4686460266981995e-05,
+      "loss": 1.2197,
+      "step": 4846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.46844068501285e-05,
+      "loss": 0.1473,
+      "step": 4847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.468235312201833e-05,
+      "loss": 0.7786,
+      "step": 4848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4680299082717496e-05,
+      "loss": 1.4329,
+      "step": 4849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4678244732292005e-05,
+      "loss": 0.9429,
+      "step": 4850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4676190070807897e-05,
+      "loss": 1.2303,
+      "step": 4851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4674135098331195e-05,
+      "loss": 0.1966,
+      "step": 4852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4672079814927962e-05,
+      "loss": 0.8927,
+      "step": 4853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4670024220664234e-05,
+      "loss": 0.704,
+      "step": 4854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4667968315606096e-05,
+      "loss": 0.2231,
+      "step": 4855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4665912099819617e-05,
+      "loss": 1.5067,
+      "step": 4856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4663855573370886e-05,
+      "loss": 0.8095,
+      "step": 4857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4661798736325995e-05,
+      "loss": 1.8923,
+      "step": 4858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4659741588751056e-05,
+      "loss": 0.1965,
+      "step": 4859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4657684130712182e-05,
+      "loss": 0.3499,
+      "step": 4860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4655626362275503e-05,
+      "loss": 1.1888,
+      "step": 4861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4653568283507155e-05,
+      "loss": 2.053,
+      "step": 4862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.465150989447328e-05,
+      "loss": 0.3196,
+      "step": 4863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.464945119524004e-05,
+      "loss": 0.2914,
+      "step": 4864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.46473921858736e-05,
+      "loss": 0.1294,
+      "step": 4865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4645332866440136e-05,
+      "loss": 0.3478,
+      "step": 4866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4643273237005834e-05,
+      "loss": 1.239,
+      "step": 4867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.464121329763689e-05,
+      "loss": 0.6723,
+      "step": 4868
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4639153048399518e-05,
+      "loss": 1.0428,
+      "step": 4869
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4637092489359922e-05,
+      "loss": 0.5946,
+      "step": 4870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4635031620584335e-05,
+      "loss": 0.4978,
+      "step": 4871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4632970442138992e-05,
+      "loss": 0.2498,
+      "step": 4872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4630908954090143e-05,
+      "loss": 0.5258,
+      "step": 4873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.462884715650404e-05,
+      "loss": 0.6915,
+      "step": 4874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.462678504944695e-05,
+      "loss": 0.6783,
+      "step": 4875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.462472263298515e-05,
+      "loss": 0.4229,
+      "step": 4876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4622659907184926e-05,
+      "loss": 0.758,
+      "step": 4877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4620596872112568e-05,
+      "loss": 0.285,
+      "step": 4878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4618533527834397e-05,
+      "loss": 0.3345,
+      "step": 4879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.461646987441671e-05,
+      "loss": 0.7003,
+      "step": 4880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4614405911925845e-05,
+      "loss": 1.1266,
+      "step": 4881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4612341640428134e-05,
+      "loss": 0.2892,
+      "step": 4882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4610277059989924e-05,
+      "loss": 1.1576,
+      "step": 4883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.460821217067757e-05,
+      "loss": 0.4012,
+      "step": 4884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.460614697255744e-05,
+      "loss": 0.613,
+      "step": 4885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4604081465695904e-05,
+      "loss": 1.2587,
+      "step": 4886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.460201565015935e-05,
+      "loss": 0.7171,
+      "step": 4887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.459994952601417e-05,
+      "loss": 0.4417,
+      "step": 4888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.459788309332678e-05,
+      "loss": 0.6547,
+      "step": 4889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4595816352163583e-05,
+      "loss": 0.6056,
+      "step": 4890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.459374930259101e-05,
+      "loss": 1.2093,
+      "step": 4891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4591681944675492e-05,
+      "loss": 1.0579,
+      "step": 4892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4589614278483475e-05,
+      "loss": 1.0182,
+      "step": 4893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4587546304081423e-05,
+      "loss": 0.8353,
+      "step": 4894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.458547802153579e-05,
+      "loss": 0.393,
+      "step": 4895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.458340943091305e-05,
+      "loss": 0.5807,
+      "step": 4896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4581340532279692e-05,
+      "loss": 0.4639,
+      "step": 4897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.457927132570221e-05,
+      "loss": 1.3999,
+      "step": 4898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.457720181124711e-05,
+      "loss": 0.6635,
+      "step": 4899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4575131988980903e-05,
+      "loss": 0.477,
+      "step": 4900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.457306185897011e-05,
+      "loss": 0.035,
+      "step": 4901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4570991421281277e-05,
+      "loss": 0.1386,
+      "step": 4902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.456892067598093e-05,
+      "loss": 0.2538,
+      "step": 4903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4566849623135638e-05,
+      "loss": 0.0687,
+      "step": 4904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4564778262811954e-05,
+      "loss": 0.2673,
+      "step": 4905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4562706595076462e-05,
+      "loss": 0.2128,
+      "step": 4906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.456063461999574e-05,
+      "loss": 0.4713,
+      "step": 4907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4558562337636377e-05,
+      "loss": 1.746,
+      "step": 4908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4556489748064982e-05,
+      "loss": 0.353,
+      "step": 4909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4554416851348167e-05,
+      "loss": 0.9341,
+      "step": 4910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.455234364755255e-05,
+      "loss": 0.1712,
+      "step": 4911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4550270136744774e-05,
+      "loss": 0.8476,
+      "step": 4912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4548196318991472e-05,
+      "loss": 0.0366,
+      "step": 4913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4546122194359298e-05,
+      "loss": 0.6789,
+      "step": 4914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4544047762914918e-05,
+      "loss": 0.4167,
+      "step": 4915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4541973024725003e-05,
+      "loss": 1.0206,
+      "step": 4916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4539897979856228e-05,
+      "loss": 0.2158,
+      "step": 4917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4537822628375303e-05,
+      "loss": 0.0933,
+      "step": 4918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4535746970348912e-05,
+      "loss": 0.1398,
+      "step": 4919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4533671005843768e-05,
+      "loss": 0.1058,
+      "step": 4920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4531594734926603e-05,
+      "loss": 1.1387,
+      "step": 4921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.452951815766414e-05,
+      "loss": 0.157,
+      "step": 4922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4527441274123125e-05,
+      "loss": 0.6418,
+      "step": 4923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4525364084370305e-05,
+      "loss": 0.6365,
+      "step": 4924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.452328658847244e-05,
+      "loss": 0.7196,
+      "step": 4925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.452120878649631e-05,
+      "loss": 0.2348,
+      "step": 4926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.451913067850868e-05,
+      "loss": 0.2981,
+      "step": 4927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4517052264576357e-05,
+      "loss": 0.6377,
+      "step": 4928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4514973544766127e-05,
+      "loss": 1.0484,
+      "step": 4929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4512894519144807e-05,
+      "loss": 0.4037,
+      "step": 4930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.451081518777922e-05,
+      "loss": 1.9402,
+      "step": 4931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4508735550736188e-05,
+      "loss": 1.6295,
+      "step": 4932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4506655608082555e-05,
+      "loss": 0.3482,
+      "step": 4933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.450457535988517e-05,
+      "loss": 1.1241,
+      "step": 4934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4502494806210897e-05,
+      "loss": 2.0154,
+      "step": 4935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.450041394712659e-05,
+      "loss": 0.2669,
+      "step": 4936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4498332782699145e-05,
+      "loss": 0.5308,
+      "step": 4937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4496251312995442e-05,
+      "loss": 1.1649,
+      "step": 4938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4494169538082374e-05,
+      "loss": 1.4735,
+      "step": 4939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4492087458026863e-05,
+      "loss": 0.4873,
+      "step": 4940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4490005072895816e-05,
+      "loss": 0.7432,
+      "step": 4941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4487922382756164e-05,
+      "loss": 0.1858,
+      "step": 4942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4485839387674846e-05,
+      "loss": 0.5944,
+      "step": 4943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.448375608771881e-05,
+      "loss": 0.8277,
+      "step": 4944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.448167248295501e-05,
+      "loss": 0.8669,
+      "step": 4945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4479588573450415e-05,
+      "loss": 0.2115,
+      "step": 4946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4477504359272e-05,
+      "loss": 0.6799,
+      "step": 4947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.447541984048676e-05,
+      "loss": 0.7033,
+      "step": 4948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4473335017161674e-05,
+      "loss": 0.5019,
+      "step": 4949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.447124988936377e-05,
+      "loss": 0.4244,
+      "step": 4950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4469164457160047e-05,
+      "loss": 1.2008,
+      "step": 4951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4467078720617533e-05,
+      "loss": 1.2605,
+      "step": 4952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4464992679803275e-05,
+      "loss": 0.1057,
+      "step": 4953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4462906334784303e-05,
+      "loss": 1.2837,
+      "step": 4954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4460819685627685e-05,
+      "loss": 1.289,
+      "step": 4955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4458732732400476e-05,
+      "loss": 0.0411,
+      "step": 4956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4456645475169758e-05,
+      "loss": 0.5485,
+      "step": 4957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.445455791400261e-05,
+      "loss": 0.2539,
+      "step": 4958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.445247004896613e-05,
+      "loss": 1.6749,
+      "step": 4959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.445038188012742e-05,
+      "loss": 1.0152,
+      "step": 4960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4448293407553594e-05,
+      "loss": 1.4149,
+      "step": 4961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4446204631311783e-05,
+      "loss": 0.4692,
+      "step": 4962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.44441155514691e-05,
+      "loss": 0.4262,
+      "step": 4963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4442026168092708e-05,
+      "loss": 1.5858,
+      "step": 4964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4439936481249753e-05,
+      "loss": 0.9769,
+      "step": 4965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4437846491007396e-05,
+      "loss": 1.3051,
+      "step": 4966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4435756197432813e-05,
+      "loss": 0.6426,
+      "step": 4967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4433665600593177e-05,
+      "loss": 0.2647,
+      "step": 4968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4431574700555693e-05,
+      "loss": 1.7198,
+      "step": 4969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.442948349738755e-05,
+      "loss": 0.1301,
+      "step": 4970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.442739199115597e-05,
+      "loss": 0.5545,
+      "step": 4971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4425300181928164e-05,
+      "loss": 1.3251,
+      "step": 4972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4423208069771373e-05,
+      "loss": 0.1386,
+      "step": 4973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4421115654752826e-05,
+      "loss": 1.0238,
+      "step": 4974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4419022936939784e-05,
+      "loss": 0.3146,
+      "step": 4975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.44169299163995e-05,
+      "loss": 0.8504,
+      "step": 4976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4414836593199247e-05,
+      "loss": 1.0303,
+      "step": 4977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.44127429674063e-05,
+      "loss": 0.8002,
+      "step": 4978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.441064903908796e-05,
+      "loss": 1.0986,
+      "step": 4979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.440855480831151e-05,
+      "loss": 0.3507,
+      "step": 4980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.440646027514426e-05,
+      "loss": 0.4557,
+      "step": 4981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4404365439653543e-05,
+      "loss": 2.1381,
+      "step": 4982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4402270301906676e-05,
+      "loss": 0.8889,
+      "step": 4983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4400174861970996e-05,
+      "loss": 0.9172,
+      "step": 4984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.439807911991386e-05,
+      "loss": 0.2316,
+      "step": 4985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4395983075802606e-05,
+      "loss": 1.3092,
+      "step": 4986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4393886729704622e-05,
+      "loss": 1.3478,
+      "step": 4987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4391790081687273e-05,
+      "loss": 1.4526,
+      "step": 4988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4389693131817946e-05,
+      "loss": 0.4574,
+      "step": 4989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.438759588016404e-05,
+      "loss": 0.1382,
+      "step": 4990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.438549832679296e-05,
+      "loss": 0.3434,
+      "step": 4991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4383400471772125e-05,
+      "loss": 0.5877,
+      "step": 4992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4381302315168944e-05,
+      "loss": 0.4344,
+      "step": 4993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4379203857050874e-05,
+      "loss": 0.9286,
+      "step": 4994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.437710509748534e-05,
+      "loss": 0.587,
+      "step": 4995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4375006036539815e-05,
+      "loss": 0.6847,
+      "step": 4996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4372906674281745e-05,
+      "loss": 0.9899,
+      "step": 4997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.437080701077861e-05,
+      "loss": 0.6734,
+      "step": 4998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4368707046097896e-05,
+      "loss": 0.2665,
+      "step": 4999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.43666067803071e-05,
+      "loss": 0.2723,
+      "step": 5000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4364506213473712e-05,
+      "loss": 0.2727,
+      "step": 5001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.436240534566525e-05,
+      "loss": 0.0381,
+      "step": 5002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4360304176949244e-05,
+      "loss": 1.7367,
+      "step": 5003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4358202707393212e-05,
+      "loss": 0.4679,
+      "step": 5004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.43561009370647e-05,
+      "loss": 0.3003,
+      "step": 5005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4353998866031265e-05,
+      "loss": 0.0583,
+      "step": 5006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.435189649436046e-05,
+      "loss": 0.0577,
+      "step": 5007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4349793822119858e-05,
+      "loss": 0.2144,
+      "step": 5008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4347690849377036e-05,
+      "loss": 1.0747,
+      "step": 5009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4345587576199594e-05,
+      "loss": 1.4196,
+      "step": 5010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4343484002655114e-05,
+      "loss": 1.1294,
+      "step": 5011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4341380128811223e-05,
+      "loss": 0.0993,
+      "step": 5012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4339275954735525e-05,
+      "loss": 1.1274,
+      "step": 5013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4337171480495654e-05,
+      "loss": 0.7755,
+      "step": 5014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.433506670615925e-05,
+      "loss": 0.1373,
+      "step": 5015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4332961631793953e-05,
+      "loss": 1.4216,
+      "step": 5016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4330856257467432e-05,
+      "loss": 1.1232,
+      "step": 5017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.432875058324734e-05,
+      "loss": 0.3894,
+      "step": 5018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4326644609201368e-05,
+      "loss": 0.5275,
+      "step": 5019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4324538335397188e-05,
+      "loss": 0.3403,
+      "step": 5020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4322431761902505e-05,
+      "loss": 0.6132,
+      "step": 5021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4320324888785018e-05,
+      "loss": 2.2496,
+      "step": 5022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.431821771611245e-05,
+      "loss": 1.128,
+      "step": 5023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4316110243952516e-05,
+      "loss": 0.4537,
+      "step": 5024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.431400247237295e-05,
+      "loss": 0.3859,
+      "step": 5025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4311894401441508e-05,
+      "loss": 1.4677,
+      "step": 5026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4309786031225928e-05,
+      "loss": 0.4389,
+      "step": 5027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.430767736179399e-05,
+      "loss": 0.8524,
+      "step": 5028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4305568393213452e-05,
+      "loss": 0.5403,
+      "step": 5029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.43034591255521e-05,
+      "loss": 0.9414,
+      "step": 5030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4301349558877727e-05,
+      "loss": 0.6703,
+      "step": 5031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4299239693258133e-05,
+      "loss": 1.3682,
+      "step": 5032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4297129528761135e-05,
+      "loss": 0.8004,
+      "step": 5033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.429501906545455e-05,
+      "loss": 0.4019,
+      "step": 5034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4292908303406207e-05,
+      "loss": 0.4862,
+      "step": 5035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.429079724268394e-05,
+      "loss": 1.276,
+      "step": 5036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.428868588335561e-05,
+      "loss": 0.6459,
+      "step": 5037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.428657422548907e-05,
+      "loss": 0.5206,
+      "step": 5038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4284462269152187e-05,
+      "loss": 1.0281,
+      "step": 5039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4282350014412845e-05,
+      "loss": 0.1521,
+      "step": 5040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.428023746133893e-05,
+      "loss": 0.526,
+      "step": 5041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.427812460999834e-05,
+      "loss": 1.6621,
+      "step": 5042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.427601146045897e-05,
+      "loss": 0.7997,
+      "step": 5043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4273898012788757e-05,
+      "loss": 0.8443,
+      "step": 5044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.427178426705561e-05,
+      "loss": 0.4557,
+      "step": 5045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4269670223327477e-05,
+      "loss": 0.5653,
+      "step": 5046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.426755588167229e-05,
+      "loss": 1.1275,
+      "step": 5047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4265441242158017e-05,
+      "loss": 0.8133,
+      "step": 5048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.426332630485262e-05,
+      "loss": 0.2374,
+      "step": 5049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4261211069824064e-05,
+      "loss": 0.195,
+      "step": 5050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.425909553714034e-05,
+      "loss": 1.7751,
+      "step": 5051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4256979706869443e-05,
+      "loss": 0.3036,
+      "step": 5052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4254863579079373e-05,
+      "loss": 0.1232,
+      "step": 5053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.425274715383814e-05,
+      "loss": 0.4603,
+      "step": 5054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4250630431213767e-05,
+      "loss": 0.417,
+      "step": 5055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.424851341127429e-05,
+      "loss": 0.282,
+      "step": 5056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4246396094087743e-05,
+      "loss": 1.2969,
+      "step": 5057
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4244278479722182e-05,
+      "loss": 0.2986,
+      "step": 5058
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4242160568245667e-05,
+      "loss": 1.4764,
+      "step": 5059
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4240042359726263e-05,
+      "loss": 0.1962,
+      "step": 5060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.423792385423205e-05,
+      "loss": 0.8961,
+      "step": 5061
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4235805051831125e-05,
+      "loss": 0.2716,
+      "step": 5062
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4233685952591576e-05,
+      "loss": 1.5537,
+      "step": 5063
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.423156655658151e-05,
+      "loss": 0.1842,
+      "step": 5064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4229446863869057e-05,
+      "loss": 1.1129,
+      "step": 5065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4227326874522333e-05,
+      "loss": 0.1031,
+      "step": 5066
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4225206588609476e-05,
+      "loss": 0.3051,
+      "step": 5067
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4223086006198638e-05,
+      "loss": 0.1643,
+      "step": 5068
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4220965127357965e-05,
+      "loss": 1.7293,
+      "step": 5069
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4218843952155628e-05,
+      "loss": 0.9098,
+      "step": 5070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4216722480659805e-05,
+      "loss": 1.614,
+      "step": 5071
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4214600712938667e-05,
+      "loss": 0.317,
+      "step": 5072
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4212478649060418e-05,
+      "loss": 0.1996,
+      "step": 5073
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.421035628909326e-05,
+      "loss": 1.8801,
+      "step": 5074
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4208233633105408e-05,
+      "loss": 0.4923,
+      "step": 5075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4206110681165078e-05,
+      "loss": 1.4874,
+      "step": 5076
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4203987433340506e-05,
+      "loss": 0.7856,
+      "step": 5077
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.420186388969993e-05,
+      "loss": 1.0012,
+      "step": 5078
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4199740050311606e-05,
+      "loss": 0.4499,
+      "step": 5079
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4197615915243787e-05,
+      "loss": 0.8762,
+      "step": 5080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4195491484564748e-05,
+      "loss": 1.3327,
+      "step": 5081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4193366758342765e-05,
+      "loss": 0.7716,
+      "step": 5082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4191241736646125e-05,
+      "loss": 0.2163,
+      "step": 5083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4189116419543133e-05,
+      "loss": 1.0227,
+      "step": 5084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4186990807102093e-05,
+      "loss": 0.1679,
+      "step": 5085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4184864899391318e-05,
+      "loss": 1.0387,
+      "step": 5086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.418273869647914e-05,
+      "loss": 0.101,
+      "step": 5087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4180612198433897e-05,
+      "loss": 0.2649,
+      "step": 5088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.417848540532393e-05,
+      "loss": 0.8696,
+      "step": 5089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4176358317217592e-05,
+      "loss": 1.1656,
+      "step": 5090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4174230934183254e-05,
+      "loss": 0.1824,
+      "step": 5091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.417210325628929e-05,
+      "loss": 0.3371,
+      "step": 5092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4169975283604072e-05,
+      "loss": 1.7357,
+      "step": 5093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.416784701619601e-05,
+      "loss": 0.7871,
+      "step": 5094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4165718454133493e-05,
+      "loss": 1.359,
+      "step": 5095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.416358959748494e-05,
+      "loss": 0.7163,
+      "step": 5096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.416146044631877e-05,
+      "loss": 0.9091,
+      "step": 5097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4159331000703412e-05,
+      "loss": 0.9088,
+      "step": 5098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.415720126070731e-05,
+      "loss": 0.1012,
+      "step": 5099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4155071226398916e-05,
+      "loss": 0.6466,
+      "step": 5100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4152940897846684e-05,
+      "loss": 0.0359,
+      "step": 5101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4150810275119083e-05,
+      "loss": 0.4604,
+      "step": 5102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4148679358284593e-05,
+      "loss": 0.144,
+      "step": 5103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.41465481474117e-05,
+      "loss": 0.5944,
+      "step": 5104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4144416642568903e-05,
+      "loss": 1.6689,
+      "step": 5105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.414228484382471e-05,
+      "loss": 0.1757,
+      "step": 5106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4140152751247638e-05,
+      "loss": 0.3674,
+      "step": 5107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.41380203649062e-05,
+      "loss": 0.0989,
+      "step": 5108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4135887684868945e-05,
+      "loss": 1.4478,
+      "step": 5109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.413375471120441e-05,
+      "loss": 0.5224,
+      "step": 5110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4131621443981154e-05,
+      "loss": 1.419,
+      "step": 5111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.412948788326774e-05,
+      "loss": 1.7663,
+      "step": 5112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4127354029132732e-05,
+      "loss": 0.6073,
+      "step": 5113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.412521988164472e-05,
+      "loss": 0.7562,
+      "step": 5114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4123085440872296e-05,
+      "loss": 0.7201,
+      "step": 5115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4120950706884053e-05,
+      "loss": 1.3994,
+      "step": 5116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4118815679748614e-05,
+      "loss": 0.6257,
+      "step": 5117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4116680359534587e-05,
+      "loss": 0.6845,
+      "step": 5118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4114544746310604e-05,
+      "loss": 0.1516,
+      "step": 5119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4112408840145307e-05,
+      "loss": 1.336,
+      "step": 5120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.411027264110734e-05,
+      "loss": 0.5241,
+      "step": 5121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4108136149265365e-05,
+      "loss": 0.7318,
+      "step": 5122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4105999364688046e-05,
+      "loss": 0.1431,
+      "step": 5123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4103862287444057e-05,
+      "loss": 1.2863,
+      "step": 5124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.410172491760209e-05,
+      "loss": 0.6304,
+      "step": 5125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4099587255230834e-05,
+      "loss": 0.1169,
+      "step": 5126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4097449300399e-05,
+      "loss": 1.1038,
+      "step": 5127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.409531105317529e-05,
+      "loss": 0.1985,
+      "step": 5128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.409317251362844e-05,
+      "loss": 1.2733,
+      "step": 5129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4091033681827178e-05,
+      "loss": 0.8637,
+      "step": 5130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4088894557840242e-05,
+      "loss": 1.1184,
+      "step": 5131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4086755141736386e-05,
+      "loss": 1.7151,
+      "step": 5132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4084615433584372e-05,
+      "loss": 0.9037,
+      "step": 5133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4082475433452973e-05,
+      "loss": 0.4834,
+      "step": 5134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4080335141410964e-05,
+      "loss": 1.0361,
+      "step": 5135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.407819455752713e-05,
+      "loss": 0.9063,
+      "step": 5136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4076053681870284e-05,
+      "loss": 0.95,
+      "step": 5137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4073912514509214e-05,
+      "loss": 0.7261,
+      "step": 5138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4071771055512757e-05,
+      "loss": 0.7341,
+      "step": 5139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4069629304949727e-05,
+      "loss": 0.7514,
+      "step": 5140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4067487262888958e-05,
+      "loss": 0.3275,
+      "step": 5141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4065344929399304e-05,
+      "loss": 1.1654,
+      "step": 5142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.406320230454961e-05,
+      "loss": 1.0101,
+      "step": 5143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4061059388408754e-05,
+      "loss": 0.2049,
+      "step": 5144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4058916181045593e-05,
+      "loss": 0.3242,
+      "step": 5145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.405677268252902e-05,
+      "loss": 0.9177,
+      "step": 5146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4054628892927925e-05,
+      "loss": 0.579,
+      "step": 5147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.405248481231121e-05,
+      "loss": 0.4475,
+      "step": 5148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4050340440747782e-05,
+      "loss": 0.5309,
+      "step": 5149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.404819577830656e-05,
+      "loss": 0.5804,
+      "step": 5150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4046050825056488e-05,
+      "loss": 0.0578,
+      "step": 5151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4043905581066485e-05,
+      "loss": 0.3494,
+      "step": 5152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.404176004640551e-05,
+      "loss": 0.0623,
+      "step": 5153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.403961422114252e-05,
+      "loss": 0.2133,
+      "step": 5154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.403746810534647e-05,
+      "loss": 0.7045,
+      "step": 5155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4035321699086355e-05,
+      "loss": 0.1853,
+      "step": 5156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.403317500243115e-05,
+      "loss": 1.6004,
+      "step": 5157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.403102801544985e-05,
+      "loss": 0.6235,
+      "step": 5158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.402888073821146e-05,
+      "loss": 1.4645,
+      "step": 5159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4026733170784995e-05,
+      "loss": 1.5808,
+      "step": 5160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.402458531323948e-05,
+      "loss": 0.3434,
+      "step": 5161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4022437165643937e-05,
+      "loss": 0.2478,
+      "step": 5162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4020288728067422e-05,
+      "loss": 0.757,
+      "step": 5163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4018140000578973e-05,
+      "loss": 0.4537,
+      "step": 5164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4015990983247654e-05,
+      "loss": 0.1939,
+      "step": 5165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4013841676142538e-05,
+      "loss": 0.941,
+      "step": 5166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4011692079332703e-05,
+      "loss": 1.2668,
+      "step": 5167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4009542192887233e-05,
+      "loss": 0.703,
+      "step": 5168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4007392016875225e-05,
+      "loss": 0.3322,
+      "step": 5169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.400524155136579e-05,
+      "loss": 2.0124,
+      "step": 5170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4003090796428045e-05,
+      "loss": 1.241,
+      "step": 5171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4000939752131107e-05,
+      "loss": 0.214,
+      "step": 5172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.399878841854412e-05,
+      "loss": 0.9963,
+      "step": 5173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3996636795736217e-05,
+      "loss": 0.2627,
+      "step": 5174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3994484883776567e-05,
+      "loss": 1.0599,
+      "step": 5175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3992332682734317e-05,
+      "loss": 0.4923,
+      "step": 5176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3990180192678645e-05,
+      "loss": 1.3573,
+      "step": 5177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.398802741367873e-05,
+      "loss": 0.5352,
+      "step": 5178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3985874345803768e-05,
+      "loss": 1.0762,
+      "step": 5179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3983720989122955e-05,
+      "loss": 0.7467,
+      "step": 5180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3981567343705496e-05,
+      "loss": 0.1895,
+      "step": 5181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3979413409620615e-05,
+      "loss": 0.2905,
+      "step": 5182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.397725918693754e-05,
+      "loss": 0.5528,
+      "step": 5183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.39751046757255e-05,
+      "loss": 0.6821,
+      "step": 5184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3972949876053746e-05,
+      "loss": 0.2944,
+      "step": 5185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3970794787991533e-05,
+      "loss": 0.3365,
+      "step": 5186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3968639411608125e-05,
+      "loss": 0.2306,
+      "step": 5187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3966483746972798e-05,
+      "loss": 1.2187,
+      "step": 5188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3964327794154833e-05,
+      "loss": 0.191,
+      "step": 5189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.396217155322352e-05,
+      "loss": 0.0717,
+      "step": 5190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3960015024248167e-05,
+      "loss": 0.9873,
+      "step": 5191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3957858207298075e-05,
+      "loss": 1.136,
+      "step": 5192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3955701102442575e-05,
+      "loss": 0.4996,
+      "step": 5193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3953543709750988e-05,
+      "loss": 0.3974,
+      "step": 5194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.395138602929266e-05,
+      "loss": 0.7905,
+      "step": 5195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3949228061136928e-05,
+      "loss": 1.1606,
+      "step": 5196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3947069805353158e-05,
+      "loss": 0.3246,
+      "step": 5197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3944911262010718e-05,
+      "loss": 0.1748,
+      "step": 5198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3942752431178977e-05,
+      "loss": 0.5412,
+      "step": 5199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3940593312927323e-05,
+      "loss": 0.534,
+      "step": 5200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3938433907325147e-05,
+      "loss": 1.0527,
+      "step": 5201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3936274214441856e-05,
+      "loss": 1.0521,
+      "step": 5202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3934114234346857e-05,
+      "loss": 0.0455,
+      "step": 5203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3931953967109583e-05,
+      "loss": 1.3251,
+      "step": 5204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.392979341279945e-05,
+      "loss": 1.6649,
+      "step": 5205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.392763257148591e-05,
+      "loss": 0.138,
+      "step": 5206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.392547144323841e-05,
+      "loss": 0.1525,
+      "step": 5207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.392331002812641e-05,
+      "loss": 1.4779,
+      "step": 5208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3921148326219363e-05,
+      "loss": 0.2426,
+      "step": 5209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3918986337586768e-05,
+      "loss": 0.4585,
+      "step": 5210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3916824062298098e-05,
+      "loss": 0.5941,
+      "step": 5211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.391466150042285e-05,
+      "loss": 0.2751,
+      "step": 5212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3912498652030533e-05,
+      "loss": 0.5533,
+      "step": 5213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3910335517190657e-05,
+      "loss": 0.5422,
+      "step": 5214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3908172095972746e-05,
+      "loss": 0.3672,
+      "step": 5215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.390600838844633e-05,
+      "loss": 0.5991,
+      "step": 5216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3903844394680955e-05,
+      "loss": 0.8232,
+      "step": 5217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3901680114746175e-05,
+      "loss": 0.7851,
+      "step": 5218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3899515548711543e-05,
+      "loss": 0.3012,
+      "step": 5219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3897350696646624e-05,
+      "loss": 0.5194,
+      "step": 5220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3895185558621007e-05,
+      "loss": 1.4226,
+      "step": 5221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.389302013470427e-05,
+      "loss": 0.2053,
+      "step": 5222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.389085442496602e-05,
+      "loss": 0.2089,
+      "step": 5223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3888688429475853e-05,
+      "loss": 0.1308,
+      "step": 5224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3886522148303393e-05,
+      "loss": 0.6882,
+      "step": 5225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3884355581518257e-05,
+      "loss": 1.6565,
+      "step": 5226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3882188729190082e-05,
+      "loss": 0.8728,
+      "step": 5227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3880021591388506e-05,
+      "loss": 1.5921,
+      "step": 5228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3877854168183187e-05,
+      "loss": 0.4749,
+      "step": 5229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3875686459643785e-05,
+      "loss": 0.7569,
+      "step": 5230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.387351846583997e-05,
+      "loss": 0.8047,
+      "step": 5231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3871350186841414e-05,
+      "loss": 0.5285,
+      "step": 5232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3869181622717813e-05,
+      "loss": 0.3587,
+      "step": 5233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3867012773538863e-05,
+      "loss": 0.5946,
+      "step": 5234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.386484363937427e-05,
+      "loss": 0.5012,
+      "step": 5235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.386267422029375e-05,
+      "loss": 0.451,
+      "step": 5236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3860504516367033e-05,
+      "loss": 0.5109,
+      "step": 5237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3858334527663848e-05,
+      "loss": 1.1394,
+      "step": 5238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3856164254253936e-05,
+      "loss": 1.0595,
+      "step": 5239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3853993696207055e-05,
+      "loss": 0.4205,
+      "step": 5240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.385182285359296e-05,
+      "loss": 1.7712,
+      "step": 5241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3849651726481435e-05,
+      "loss": 0.3351,
+      "step": 5242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3847480314942246e-05,
+      "loss": 0.7184,
+      "step": 5243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3845308619045185e-05,
+      "loss": 0.7239,
+      "step": 5244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3843136638860056e-05,
+      "loss": 1.139,
+      "step": 5245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3840964374456663e-05,
+      "loss": 1.0189,
+      "step": 5246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.383879182590482e-05,
+      "loss": 0.6961,
+      "step": 5247
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.383661899327436e-05,
+      "loss": 0.9963,
+      "step": 5248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.383444587663511e-05,
+      "loss": 0.7715,
+      "step": 5249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3832272476056917e-05,
+      "loss": 0.5795,
+      "step": 5250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3830098791609632e-05,
+      "loss": 0.6505,
+      "step": 5251
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3827924823363125e-05,
+      "loss": 1.3014,
+      "step": 5252
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3825750571387262e-05,
+      "loss": 0.5502,
+      "step": 5253
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.382357603575192e-05,
+      "loss": 0.2473,
+      "step": 5254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.382140121652699e-05,
+      "loss": 0.8277,
+      "step": 5255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3819226113782375e-05,
+      "loss": 0.4866,
+      "step": 5256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.381705072758798e-05,
+      "loss": 0.0427,
+      "step": 5257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.381487505801372e-05,
+      "loss": 1.4891,
+      "step": 5258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3812699105129522e-05,
+      "loss": 0.0663,
+      "step": 5259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3810522869005323e-05,
+      "loss": 0.3545,
+      "step": 5260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3808346349711067e-05,
+      "loss": 0.2261,
+      "step": 5261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3806169547316706e-05,
+      "loss": 0.5997,
+      "step": 5262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3803992461892205e-05,
+      "loss": 0.4058,
+      "step": 5263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3801815093507532e-05,
+      "loss": 1.1673,
+      "step": 5264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.379963744223267e-05,
+      "loss": 0.7881,
+      "step": 5265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3797459508137604e-05,
+      "loss": 0.3544,
+      "step": 5266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3795281291292337e-05,
+      "loss": 0.1633,
+      "step": 5267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3793102791766876e-05,
+      "loss": 0.8082,
+      "step": 5268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.379092400963124e-05,
+      "loss": 1.3246,
+      "step": 5269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3788744944955453e-05,
+      "loss": 1.7713,
+      "step": 5270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3786565597809553e-05,
+      "loss": 0.5484,
+      "step": 5271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3784385968263577e-05,
+      "loss": 1.4544,
+      "step": 5272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3782206056387587e-05,
+      "loss": 0.288,
+      "step": 5273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3780025862251633e-05,
+      "loss": 0.5603,
+      "step": 5274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3777845385925803e-05,
+      "loss": 0.9202,
+      "step": 5275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3775664627480166e-05,
+      "loss": 0.1703,
+      "step": 5276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3773483586984812e-05,
+      "loss": 0.2237,
+      "step": 5277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3771302264509842e-05,
+      "loss": 1.4088,
+      "step": 5278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.376912066012537e-05,
+      "loss": 0.2419,
+      "step": 5279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3766938773901503e-05,
+      "loss": 1.4942,
+      "step": 5280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3764756605908366e-05,
+      "loss": 0.5918,
+      "step": 5281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3762574156216102e-05,
+      "loss": 1.5876,
+      "step": 5282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3760391424894853e-05,
+      "loss": 0.9614,
+      "step": 5283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3758208412014767e-05,
+      "loss": 0.7645,
+      "step": 5284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3756025117646008e-05,
+      "loss": 0.3054,
+      "step": 5285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3753841541858752e-05,
+      "loss": 1.5423,
+      "step": 5286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.375165768472317e-05,
+      "loss": 0.727,
+      "step": 5287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.374947354630946e-05,
+      "loss": 0.4573,
+      "step": 5288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3747289126687813e-05,
+      "loss": 0.1992,
+      "step": 5289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3745104425928442e-05,
+      "loss": 0.2307,
+      "step": 5290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.374291944410156e-05,
+      "loss": 0.3131,
+      "step": 5291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3740734181277398e-05,
+      "loss": 0.1619,
+      "step": 5292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.373854863752618e-05,
+      "loss": 0.6265,
+      "step": 5293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3736362812918155e-05,
+      "loss": 0.9437,
+      "step": 5294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.373417670752358e-05,
+      "loss": 0.994,
+      "step": 5295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3731990321412705e-05,
+      "loss": 0.7874,
+      "step": 5296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.372980365465581e-05,
+      "loss": 0.7444,
+      "step": 5297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3727616707323167e-05,
+      "loss": 0.8167,
+      "step": 5298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3725429479485074e-05,
+      "loss": 0.6108,
+      "step": 5299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.372324197121182e-05,
+      "loss": 0.3692,
+      "step": 5300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3721054182573716e-05,
+      "loss": 0.0423,
+      "step": 5301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.371886611364108e-05,
+      "loss": 1.2438,
+      "step": 5302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.371667776448423e-05,
+      "loss": 1.0452,
+      "step": 5303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.37144891351735e-05,
+      "loss": 1.2183,
+      "step": 5304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3712300225779238e-05,
+      "loss": 1.8765,
+      "step": 5305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.371011103637179e-05,
+      "loss": 1.7343,
+      "step": 5306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.370792156702152e-05,
+      "loss": 0.9976,
+      "step": 5307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3705731817798795e-05,
+      "loss": 0.9877,
+      "step": 5308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3703541788773997e-05,
+      "loss": 0.2019,
+      "step": 5309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3701351480017507e-05,
+      "loss": 0.2745,
+      "step": 5310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.369916089159973e-05,
+      "loss": 0.604,
+      "step": 5311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.369697002359107e-05,
+      "loss": 0.3161,
+      "step": 5312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3694778876061927e-05,
+      "loss": 0.2004,
+      "step": 5313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.369258744908275e-05,
+      "loss": 0.3078,
+      "step": 5314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.369039574272395e-05,
+      "loss": 0.856,
+      "step": 5315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3688203757055977e-05,
+      "loss": 1.1255,
+      "step": 5316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.368601149214928e-05,
+      "loss": 0.4612,
+      "step": 5317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.368381894807432e-05,
+      "loss": 0.1535,
+      "step": 5318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3681626124901562e-05,
+      "loss": 0.4535,
+      "step": 5319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3679433022701485e-05,
+      "loss": 1.9077,
+      "step": 5320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3677239641544578e-05,
+      "loss": 0.0894,
+      "step": 5321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3675045981501338e-05,
+      "loss": 0.9926,
+      "step": 5322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3672852042642258e-05,
+      "loss": 0.9786,
+      "step": 5323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.367065782503786e-05,
+      "loss": 1.6198,
+      "step": 5324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3668463328758665e-05,
+      "loss": 0.7514,
+      "step": 5325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3666268553875203e-05,
+      "loss": 0.5665,
+      "step": 5326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3664073500458018e-05,
+      "loss": 0.8464,
+      "step": 5327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.366187816857765e-05,
+      "loss": 0.8647,
+      "step": 5328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3659682558304664e-05,
+      "loss": 0.074,
+      "step": 5329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3657486669709625e-05,
+      "loss": 0.3616,
+      "step": 5330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.365529050286311e-05,
+      "loss": 0.7676,
+      "step": 5331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3653094057835704e-05,
+      "loss": 0.6578,
+      "step": 5332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3650897334698e-05,
+      "loss": 1.2011,
+      "step": 5333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3648700333520598e-05,
+      "loss": 1.0563,
+      "step": 5334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.364650305437411e-05,
+      "loss": 0.575,
+      "step": 5335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.364430549732916e-05,
+      "loss": 0.1522,
+      "step": 5336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3642107662456378e-05,
+      "loss": 0.9716,
+      "step": 5337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3639909549826398e-05,
+      "loss": 0.806,
+      "step": 5338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.363771115950987e-05,
+      "loss": 0.8392,
+      "step": 5339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3635512491577452e-05,
+      "loss": 0.7239,
+      "step": 5340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.36333135460998e-05,
+      "loss": 0.8529,
+      "step": 5341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3631114323147602e-05,
+      "loss": 0.9401,
+      "step": 5342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.362891482279153e-05,
+      "loss": 0.4731,
+      "step": 5343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3626715045102278e-05,
+      "loss": 0.6722,
+      "step": 5344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.362451499015055e-05,
+      "loss": 1.2688,
+      "step": 5345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3622314658007056e-05,
+      "loss": 0.8768,
+      "step": 5346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3620114048742514e-05,
+      "loss": 0.2198,
+      "step": 5347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3617913162427643e-05,
+      "loss": 0.3607,
+      "step": 5348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3615711999133195e-05,
+      "loss": 0.3234,
+      "step": 5349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3613510558929904e-05,
+      "loss": 0.2688,
+      "step": 5350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3611308841888527e-05,
+      "loss": 1.0935,
+      "step": 5351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3609106848079833e-05,
+      "loss": 0.0382,
+      "step": 5352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3606904577574582e-05,
+      "loss": 0.1769,
+      "step": 5353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3604702030443567e-05,
+      "loss": 0.4564,
+      "step": 5354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3602499206757568e-05,
+      "loss": 0.6483,
+      "step": 5355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3600296106587388e-05,
+      "loss": 0.2482,
+      "step": 5356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3598092730003838e-05,
+      "loss": 0.0722,
+      "step": 5357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3595889077077734e-05,
+      "loss": 0.6437,
+      "step": 5358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.359368514787989e-05,
+      "loss": 1.5016,
+      "step": 5359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3591480942481154e-05,
+      "loss": 0.9921,
+      "step": 5360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3589276460952364e-05,
+      "loss": 0.8827,
+      "step": 5361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.358707170336437e-05,
+      "loss": 0.3842,
+      "step": 5362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3584866669788036e-05,
+      "loss": 0.5584,
+      "step": 5363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.358266136029423e-05,
+      "loss": 0.5809,
+      "step": 5364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3580455774953838e-05,
+      "loss": 0.2716,
+      "step": 5365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3578249913837732e-05,
+      "loss": 0.2084,
+      "step": 5366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.357604377701682e-05,
+      "loss": 0.4809,
+      "step": 5367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3573837364562004e-05,
+      "loss": 0.2512,
+      "step": 5368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3571630676544197e-05,
+      "loss": 0.0868,
+      "step": 5369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.356942371303433e-05,
+      "loss": 0.2293,
+      "step": 5370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3567216474103318e-05,
+      "loss": 1.3339,
+      "step": 5371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.356500895982212e-05,
+      "loss": 1.3676,
+      "step": 5372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.356280117026167e-05,
+      "loss": 0.9379,
+      "step": 5373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.356059310549294e-05,
+      "loss": 1.5094,
+      "step": 5374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3558384765586885e-05,
+      "loss": 0.3374,
+      "step": 5375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3556176150614494e-05,
+      "loss": 0.527,
+      "step": 5376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3553967260646737e-05,
+      "loss": 0.8782,
+      "step": 5377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3551758095754617e-05,
+      "loss": 0.4666,
+      "step": 5378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3549548656009135e-05,
+      "loss": 0.6526,
+      "step": 5379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.35473389414813e-05,
+      "loss": 0.3528,
+      "step": 5380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.354512895224214e-05,
+      "loss": 1.0572,
+      "step": 5381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3542918688362674e-05,
+      "loss": 1.1455,
+      "step": 5382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3540708149913944e-05,
+      "loss": 0.7517,
+      "step": 5383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3538497336966996e-05,
+      "loss": 0.3305,
+      "step": 5384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.353628624959289e-05,
+      "loss": 0.6548,
+      "step": 5385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3534074887862684e-05,
+      "loss": 0.8315,
+      "step": 5386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3531863251847452e-05,
+      "loss": 1.0616,
+      "step": 5387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.352965134161828e-05,
+      "loss": 0.7619,
+      "step": 5388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3527439157246257e-05,
+      "loss": 1.1153,
+      "step": 5389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.352522669880248e-05,
+      "loss": 1.0876,
+      "step": 5390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3523013966358057e-05,
+      "loss": 0.4037,
+      "step": 5391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3520800959984117e-05,
+      "loss": 0.9982,
+      "step": 5392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3518587679751763e-05,
+      "loss": 0.6241,
+      "step": 5393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3516374125732154e-05,
+      "loss": 0.6116,
+      "step": 5394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.351416029799642e-05,
+      "loss": 0.4508,
+      "step": 5395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3511946196615714e-05,
+      "loss": 1.0853,
+      "step": 5396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.35097318216612e-05,
+      "loss": 0.5338,
+      "step": 5397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.350751717320405e-05,
+      "loss": 0.5983,
+      "step": 5398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.350530225131544e-05,
+      "loss": 0.5109,
+      "step": 5399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3503087056066554e-05,
+      "loss": 0.4584,
+      "step": 5400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3500871587528593e-05,
+      "loss": 1.4905,
+      "step": 5401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.349865584577276e-05,
+      "loss": 1.5982,
+      "step": 5402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3496439830870275e-05,
+      "loss": 0.3331,
+      "step": 5403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3494223542892353e-05,
+      "loss": 0.4754,
+      "step": 5404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3492006981910226e-05,
+      "loss": 0.3078,
+      "step": 5405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.348979014799514e-05,
+      "loss": 0.513,
+      "step": 5406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3487573041218342e-05,
+      "loss": 1.412,
+      "step": 5407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3485355661651087e-05,
+      "loss": 0.6488,
+      "step": 5408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.348313800936464e-05,
+      "loss": 0.5322,
+      "step": 5409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3480920084430278e-05,
+      "loss": 1.078,
+      "step": 5410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.347870188691929e-05,
+      "loss": 0.8886,
+      "step": 5411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.347648341690296e-05,
+      "loss": 0.0938,
+      "step": 5412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.34742646744526e-05,
+      "loss": 0.1088,
+      "step": 5413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3472045659639513e-05,
+      "loss": 0.1987,
+      "step": 5414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3469826372535022e-05,
+      "loss": 0.5023,
+      "step": 5415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3467606813210446e-05,
+      "loss": 0.3093,
+      "step": 5416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3465386981737135e-05,
+      "loss": 0.1973,
+      "step": 5417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3463166878186426e-05,
+      "loss": 1.864,
+      "step": 5418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3460946502629676e-05,
+      "loss": 0.1968,
+      "step": 5419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3458725855138244e-05,
+      "loss": 0.716,
+      "step": 5420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3456504935783507e-05,
+      "loss": 1.1556,
+      "step": 5421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3454283744636842e-05,
+      "loss": 0.1745,
+      "step": 5422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3452062281769638e-05,
+      "loss": 0.4565,
+      "step": 5423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3449840547253296e-05,
+      "loss": 0.7245,
+      "step": 5424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3447618541159214e-05,
+      "loss": 0.4378,
+      "step": 5425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3445396263558818e-05,
+      "loss": 1.475,
+      "step": 5426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.344317371452353e-05,
+      "loss": 0.8163,
+      "step": 5427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3440950894124772e-05,
+      "loss": 0.3618,
+      "step": 5428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3438727802433996e-05,
+      "loss": 1.2281,
+      "step": 5429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.343650443952265e-05,
+      "loss": 0.9287,
+      "step": 5430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3434280805462195e-05,
+      "loss": 0.6291,
+      "step": 5431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.343205690032409e-05,
+      "loss": 0.3806,
+      "step": 5432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3429832724179822e-05,
+      "loss": 0.2423,
+      "step": 5433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3427608277100873e-05,
+      "loss": 0.6752,
+      "step": 5434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3425383559158725e-05,
+      "loss": 1.0039,
+      "step": 5435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3423158570424904e-05,
+      "loss": 0.2207,
+      "step": 5436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3420933310970896e-05,
+      "loss": 1.101,
+      "step": 5437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3418707780868238e-05,
+      "loss": 0.7859,
+      "step": 5438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3416481980188448e-05,
+      "loss": 1.084,
+      "step": 5439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.341425590900307e-05,
+      "loss": 0.6645,
+      "step": 5440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.341202956738365e-05,
+      "loss": 0.3506,
+      "step": 5441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3409802955401736e-05,
+      "loss": 0.4276,
+      "step": 5442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3407576073128898e-05,
+      "loss": 0.0915,
+      "step": 5443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3405348920636704e-05,
+      "loss": 1.1945,
+      "step": 5444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.340312149799674e-05,
+      "loss": 0.115,
+      "step": 5445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3400893805280585e-05,
+      "loss": 0.91,
+      "step": 5446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.339866584255985e-05,
+      "loss": 1.0113,
+      "step": 5447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3396437609906128e-05,
+      "loss": 0.4294,
+      "step": 5448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3394209107391047e-05,
+      "loss": 0.317,
+      "step": 5449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3391980335086225e-05,
+      "loss": 0.5726,
+      "step": 5450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.338975129306329e-05,
+      "loss": 0.0398,
+      "step": 5451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3387521981393894e-05,
+      "loss": 1.0409,
+      "step": 5452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3385292400149683e-05,
+      "loss": 1.5934,
+      "step": 5453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.338306254940231e-05,
+      "loss": 0.6204,
+      "step": 5454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.338083242922345e-05,
+      "loss": 1.9729,
+      "step": 5455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3378602039684773e-05,
+      "loss": 0.2379,
+      "step": 5456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3376371380857966e-05,
+      "loss": 0.0683,
+      "step": 5457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3374140452814726e-05,
+      "loss": 1.2259,
+      "step": 5458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3371909255626748e-05,
+      "loss": 0.9357,
+      "step": 5459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.336967778936575e-05,
+      "loss": 0.3938,
+      "step": 5460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3367446054103445e-05,
+      "loss": 0.6374,
+      "step": 5461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.336521404991156e-05,
+      "loss": 0.8313,
+      "step": 5462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3362981776861842e-05,
+      "loss": 0.6479,
+      "step": 5463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3360749235026027e-05,
+      "loss": 0.5707,
+      "step": 5464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3358516424475874e-05,
+      "loss": 0.2279,
+      "step": 5465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3356283345283137e-05,
+      "loss": 0.4155,
+      "step": 5466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3354049997519597e-05,
+      "loss": 0.322,
+      "step": 5467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.335181638125703e-05,
+      "loss": 0.8921,
+      "step": 5468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3349582496567222e-05,
+      "loss": 0.1319,
+      "step": 5469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.334734834352197e-05,
+      "loss": 0.9537,
+      "step": 5470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3345113922193088e-05,
+      "loss": 1.3913,
+      "step": 5471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.334287923265238e-05,
+      "loss": 0.6972,
+      "step": 5472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.334064427497167e-05,
+      "loss": 0.9368,
+      "step": 5473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3338409049222797e-05,
+      "loss": 0.5252,
+      "step": 5474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.333617355547759e-05,
+      "loss": 0.6981,
+      "step": 5475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3333937793807912e-05,
+      "loss": 0.1138,
+      "step": 5476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3331701764285613e-05,
+      "loss": 1.0439,
+      "step": 5477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.332946546698255e-05,
+      "loss": 0.1952,
+      "step": 5478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3327228901970616e-05,
+      "loss": 0.706,
+      "step": 5479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3324992069321677e-05,
+      "loss": 0.3578,
+      "step": 5480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3322754969107634e-05,
+      "loss": 0.9288,
+      "step": 5481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3320517601400384e-05,
+      "loss": 0.554,
+      "step": 5482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3318279966271834e-05,
+      "loss": 0.7039,
+      "step": 5483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3316042063793914e-05,
+      "loss": 0.7637,
+      "step": 5484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3313803894038534e-05,
+      "loss": 0.5946,
+      "step": 5485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.331156545707764e-05,
+      "loss": 0.2076,
+      "step": 5486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3309326752983167e-05,
+      "loss": 0.2117,
+      "step": 5487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3307087781827074e-05,
+      "loss": 0.4269,
+      "step": 5488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.330484854368132e-05,
+      "loss": 0.6749,
+      "step": 5489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.330260903861787e-05,
+      "loss": 0.289,
+      "step": 5490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3300369266708703e-05,
+      "loss": 0.7885,
+      "step": 5491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3298129228025805e-05,
+      "loss": 0.2787,
+      "step": 5492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3295888922641174e-05,
+      "loss": 0.3126,
+      "step": 5493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3293648350626817e-05,
+      "loss": 0.995,
+      "step": 5494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3291407512054735e-05,
+      "loss": 0.8711,
+      "step": 5495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3289166406996957e-05,
+      "loss": 1.0422,
+      "step": 5496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3286925035525504e-05,
+      "loss": 0.7669,
+      "step": 5497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3284683397712422e-05,
+      "loss": 0.5077,
+      "step": 5498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3282441493629756e-05,
+      "loss": 1.0531,
+      "step": 5499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3280199323349554e-05,
+      "loss": 0.6763,
+      "step": 5500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.327795688694389e-05,
+      "loss": 1.4234,
+      "step": 5501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3275714184484823e-05,
+      "loss": 1.3359,
+      "step": 5502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3273471216044448e-05,
+      "loss": 0.0493,
+      "step": 5503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3271227981694837e-05,
+      "loss": 0.5347,
+      "step": 5504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3268984481508103e-05,
+      "loss": 0.3347,
+      "step": 5505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3266740715556346e-05,
+      "loss": 1.6569,
+      "step": 5506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.326449668391168e-05,
+      "loss": 0.9994,
+      "step": 5507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.326225238664623e-05,
+      "loss": 1.1301,
+      "step": 5508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3260007823832122e-05,
+      "loss": 1.1851,
+      "step": 5509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3257762995541505e-05,
+      "loss": 0.6099,
+      "step": 5510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.325551790184652e-05,
+      "loss": 0.7892,
+      "step": 5511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3253272542819326e-05,
+      "loss": 0.3311,
+      "step": 5512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3251026918532095e-05,
+      "loss": 0.518,
+      "step": 5513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3248781029056992e-05,
+      "loss": 0.8425,
+      "step": 5514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.324653487446621e-05,
+      "loss": 0.8546,
+      "step": 5515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3244288454831933e-05,
+      "loss": 1.228,
+      "step": 5516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3242041770226362e-05,
+      "loss": 0.9102,
+      "step": 5517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3239794820721706e-05,
+      "loss": 0.7051,
+      "step": 5518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3237547606390183e-05,
+      "loss": 0.0884,
+      "step": 5519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3235300127304015e-05,
+      "loss": 1.421,
+      "step": 5520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3233052383535443e-05,
+      "loss": 0.3527,
+      "step": 5521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.32308043751567e-05,
+      "loss": 0.8051,
+      "step": 5522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3228556102240044e-05,
+      "loss": 0.2705,
+      "step": 5523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3226307564857728e-05,
+      "loss": 0.1066,
+      "step": 5524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.322405876308203e-05,
+      "loss": 1.8005,
+      "step": 5525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.322180969698522e-05,
+      "loss": 0.6233,
+      "step": 5526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3219560366639584e-05,
+      "loss": 1.1117,
+      "step": 5527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3217310772117416e-05,
+      "loss": 1.0302,
+      "step": 5528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3215060913491015e-05,
+      "loss": 0.2521,
+      "step": 5529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.321281079083269e-05,
+      "loss": 0.2381,
+      "step": 5530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.321056040421477e-05,
+      "loss": 0.1616,
+      "step": 5531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.320830975370957e-05,
+      "loss": 0.1034,
+      "step": 5532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3206058839389437e-05,
+      "loss": 1.1414,
+      "step": 5533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3203807661326705e-05,
+      "loss": 0.6879,
+      "step": 5534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3201556219593733e-05,
+      "loss": 1.9959,
+      "step": 5535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3199304514262882e-05,
+      "loss": 0.4825,
+      "step": 5536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3197052545406524e-05,
+      "loss": 0.2018,
+      "step": 5537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3194800313097032e-05,
+      "loss": 0.7697,
+      "step": 5538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3192547817406788e-05,
+      "loss": 0.7463,
+      "step": 5539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3190295058408204e-05,
+      "loss": 0.936,
+      "step": 5540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3188042036173667e-05,
+      "loss": 0.236,
+      "step": 5541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.31857887507756e-05,
+      "loss": 0.8906,
+      "step": 5542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3183535202286414e-05,
+      "loss": 0.9927,
+      "step": 5543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.318128139077855e-05,
+      "loss": 1.0616,
+      "step": 5544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3179027316324434e-05,
+      "loss": 0.6615,
+      "step": 5545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.317677297899652e-05,
+      "loss": 1.1381,
+      "step": 5546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.317451837886726e-05,
+      "loss": 0.5904,
+      "step": 5547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3172263516009108e-05,
+      "loss": 0.1857,
+      "step": 5548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.317000839049455e-05,
+      "loss": 0.0956,
+      "step": 5549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3167753002396056e-05,
+      "loss": 0.4759,
+      "step": 5550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3165497351786117e-05,
+      "loss": 0.0355,
+      "step": 5551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.316324143873723e-05,
+      "loss": 0.1118,
+      "step": 5552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3160985263321902e-05,
+      "loss": 1.216,
+      "step": 5553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3158728825612642e-05,
+      "loss": 0.1691,
+      "step": 5554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.315647212568197e-05,
+      "loss": 1.454,
+      "step": 5555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3154215163602425e-05,
+      "loss": 1.4972,
+      "step": 5556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3151957939446543e-05,
+      "loss": 0.305,
+      "step": 5557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3149700453286866e-05,
+      "loss": 0.4133,
+      "step": 5558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.314744270519595e-05,
+      "loss": 1.4265,
+      "step": 5559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3145184695246367e-05,
+      "loss": 0.1837,
+      "step": 5560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.314292642351068e-05,
+      "loss": 1.4638,
+      "step": 5561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3140667890061472e-05,
+      "loss": 0.6437,
+      "step": 5562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3138409094971336e-05,
+      "loss": 0.7475,
+      "step": 5563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3136150038312873e-05,
+      "loss": 0.3252,
+      "step": 5564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3133890720158675e-05,
+      "loss": 0.4431,
+      "step": 5565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.313163114058137e-05,
+      "loss": 0.6085,
+      "step": 5566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3129371299653575e-05,
+      "loss": 0.4297,
+      "step": 5567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.312711119744792e-05,
+      "loss": 0.2096,
+      "step": 5568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.312485083403705e-05,
+      "loss": 0.0594,
+      "step": 5569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.312259020949361e-05,
+      "loss": 0.9762,
+      "step": 5570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3120329323890256e-05,
+      "loss": 0.0452,
+      "step": 5571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3118068177299653e-05,
+      "loss": 0.6435,
+      "step": 5572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3115806769794472e-05,
+      "loss": 0.2944,
+      "step": 5573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.31135451014474e-05,
+      "loss": 1.263,
+      "step": 5574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.311128317233112e-05,
+      "loss": 1.7734,
+      "step": 5575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3109020982518337e-05,
+      "loss": 1.4208,
+      "step": 5576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3106758532081756e-05,
+      "loss": 0.6502,
+      "step": 5577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.310449582109409e-05,
+      "loss": 0.3321,
+      "step": 5578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3102232849628068e-05,
+      "loss": 0.8179,
+      "step": 5579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.309996961775641e-05,
+      "loss": 0.7357,
+      "step": 5580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.309770612555187e-05,
+      "loss": 0.7406,
+      "step": 5581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3095442373087182e-05,
+      "loss": 0.8642,
+      "step": 5582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.309317836043512e-05,
+      "loss": 0.8956,
+      "step": 5583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3090914087668436e-05,
+      "loss": 1.7657,
+      "step": 5584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.308864955485991e-05,
+      "loss": 0.6815,
+      "step": 5585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3086384762082325e-05,
+      "loss": 0.5088,
+      "step": 5586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3084119709408464e-05,
+      "loss": 1.5067,
+      "step": 5587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3081854396911136e-05,
+      "loss": 0.8573,
+      "step": 5588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3079588824663138e-05,
+      "loss": 1.6582,
+      "step": 5589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.307732299273729e-05,
+      "loss": 0.1751,
+      "step": 5590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3075056901206422e-05,
+      "loss": 1.4,
+      "step": 5591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3072790550143352e-05,
+      "loss": 0.1556,
+      "step": 5592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3070523939620935e-05,
+      "loss": 0.8142,
+      "step": 5593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3068257069712007e-05,
+      "loss": 1.5882,
+      "step": 5594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3065989940489435e-05,
+      "loss": 0.8614,
+      "step": 5595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3063722552026084e-05,
+      "loss": 1.0679,
+      "step": 5596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3061454904394818e-05,
+      "loss": 0.5465,
+      "step": 5597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.305918699766853e-05,
+      "loss": 0.6908,
+      "step": 5598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3056918831920113e-05,
+      "loss": 0.4218,
+      "step": 5599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3054650407222454e-05,
+      "loss": 0.3202,
+      "step": 5600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3052381723648458e-05,
+      "loss": 0.2419,
+      "step": 5601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3050112781271055e-05,
+      "loss": 0.1452,
+      "step": 5602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3047843580163163e-05,
+      "loss": 0.2304,
+      "step": 5603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.304557412039771e-05,
+      "loss": 0.1763,
+      "step": 5604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3043304402047642e-05,
+      "loss": 0.8237,
+      "step": 5605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3041034425185903e-05,
+      "loss": 0.0982,
+      "step": 5606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.303876418988545e-05,
+      "loss": 0.0465,
+      "step": 5607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3036493696219252e-05,
+      "loss": 0.135,
+      "step": 5608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3034222944260282e-05,
+      "loss": 1.5245,
+      "step": 5609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3031951934081515e-05,
+      "loss": 0.4486,
+      "step": 5610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3029680665755953e-05,
+      "loss": 0.303,
+      "step": 5611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3027409139356585e-05,
+      "loss": 0.718,
+      "step": 5612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.302513735495643e-05,
+      "loss": 0.5057,
+      "step": 5613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3022865312628487e-05,
+      "loss": 1.545,
+      "step": 5614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3020593012445786e-05,
+      "loss": 1.3691,
+      "step": 5615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.301832045448136e-05,
+      "loss": 0.6591,
+      "step": 5616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3016047638808247e-05,
+      "loss": 0.2703,
+      "step": 5617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3013774565499503e-05,
+      "loss": 0.4314,
+      "step": 5618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3011501234628175e-05,
+      "loss": 0.5772,
+      "step": 5619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3009227646267328e-05,
+      "loss": 0.0748,
+      "step": 5620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3006953800490042e-05,
+      "loss": 0.357,
+      "step": 5621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3004679697369398e-05,
+      "loss": 0.1913,
+      "step": 5622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3002405336978477e-05,
+      "loss": 1.4417,
+      "step": 5623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3000130719390377e-05,
+      "loss": 1.1032,
+      "step": 5624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.299785584467822e-05,
+      "loss": 1.3048,
+      "step": 5625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2995580712915108e-05,
+      "loss": 1.3973,
+      "step": 5626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.299330532417416e-05,
+      "loss": 0.5398,
+      "step": 5627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2991029678528516e-05,
+      "loss": 1.7695,
+      "step": 5628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.298875377605131e-05,
+      "loss": 1.8102,
+      "step": 5629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2986477616815692e-05,
+      "loss": 0.1686,
+      "step": 5630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2984201200894818e-05,
+      "loss": 2.1763,
+      "step": 5631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.298192452836185e-05,
+      "loss": 0.8532,
+      "step": 5632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.297964759928996e-05,
+      "loss": 2.029,
+      "step": 5633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2977370413752328e-05,
+      "loss": 0.2941,
+      "step": 5634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.297509297182214e-05,
+      "loss": 0.6907,
+      "step": 5635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2972815273572606e-05,
+      "loss": 0.7386,
+      "step": 5636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.297053731907692e-05,
+      "loss": 1.6373,
+      "step": 5637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2968259108408297e-05,
+      "loss": 0.8874,
+      "step": 5638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2965980641639962e-05,
+      "loss": 1.4562,
+      "step": 5639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2963701918845137e-05,
+      "loss": 0.9355,
+      "step": 5640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2961422940097067e-05,
+      "loss": 0.4785,
+      "step": 5641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2959143705469e-05,
+      "loss": 0.7404,
+      "step": 5642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2956864215034186e-05,
+      "loss": 0.7538,
+      "step": 5643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2954584468865884e-05,
+      "loss": 0.7901,
+      "step": 5644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.295230446703738e-05,
+      "loss": 0.1255,
+      "step": 5645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2950024209621936e-05,
+      "loss": 0.1011,
+      "step": 5646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2947743696692847e-05,
+      "loss": 1.0908,
+      "step": 5647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.294546292832341e-05,
+      "loss": 0.4273,
+      "step": 5648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2943181904586924e-05,
+      "loss": 0.7911,
+      "step": 5649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2940900625556707e-05,
+      "loss": 0.4664,
+      "step": 5650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.293861909130608e-05,
+      "loss": 1.2838,
+      "step": 5651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.293633730190836e-05,
+      "loss": 0.8965,
+      "step": 5652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2934055257436896e-05,
+      "loss": 0.0348,
+      "step": 5653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2931772957965034e-05,
+      "loss": 0.1768,
+      "step": 5654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.292949040356611e-05,
+      "loss": 0.2145,
+      "step": 5655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2927207594313507e-05,
+      "loss": 0.478,
+      "step": 5656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.292492453028058e-05,
+      "loss": 1.2318,
+      "step": 5657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2922641211540707e-05,
+      "loss": 1.2988,
+      "step": 5658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2920357638167285e-05,
+      "loss": 0.3959,
+      "step": 5659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.29180738102337e-05,
+      "loss": 0.0768,
+      "step": 5660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2915789727813353e-05,
+      "loss": 0.3444,
+      "step": 5661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2913505390979655e-05,
+      "loss": 0.6036,
+      "step": 5662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.291122079980603e-05,
+      "loss": 1.4863,
+      "step": 5663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2908935954365892e-05,
+      "loss": 0.5909,
+      "step": 5664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2906650854732692e-05,
+      "loss": 0.0737,
+      "step": 5665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2904365500979864e-05,
+      "loss": 0.1481,
+      "step": 5666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2902079893180862e-05,
+      "loss": 0.772,
+      "step": 5667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.289979403140914e-05,
+      "loss": 0.211,
+      "step": 5668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2897507915738176e-05,
+      "loss": 1.1885,
+      "step": 5669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2895221546241436e-05,
+      "loss": 1.6708,
+      "step": 5670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2892934922992408e-05,
+      "loss": 0.1519,
+      "step": 5671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.289064804606458e-05,
+      "loss": 0.3832,
+      "step": 5672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2888360915531458e-05,
+      "loss": 0.488,
+      "step": 5673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.288607353146655e-05,
+      "loss": 0.8286,
+      "step": 5674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2883785893943368e-05,
+      "loss": 0.6519,
+      "step": 5675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2881498003035443e-05,
+      "loss": 0.711,
+      "step": 5676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2879209858816305e-05,
+      "loss": 0.2266,
+      "step": 5677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2876921461359482e-05,
+      "loss": 0.6655,
+      "step": 5678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2874632810738545e-05,
+      "loss": 0.4173,
+      "step": 5679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.287234390702704e-05,
+      "loss": 0.6936,
+      "step": 5680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.287005475029853e-05,
+      "loss": 0.8761,
+      "step": 5681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2867765340626594e-05,
+      "loss": 0.8227,
+      "step": 5682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2865475678084807e-05,
+      "loss": 1.1248,
+      "step": 5683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2863185762746773e-05,
+      "loss": 0.6599,
+      "step": 5684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2860895594686074e-05,
+      "loss": 0.3898,
+      "step": 5685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.285860517397632e-05,
+      "loss": 0.8773,
+      "step": 5686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2856314500691124e-05,
+      "loss": 0.6627,
+      "step": 5687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2854023574904122e-05,
+      "loss": 1.1337,
+      "step": 5688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2851732396688925e-05,
+      "loss": 1.2073,
+      "step": 5689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.284944096611918e-05,
+      "loss": 0.6814,
+      "step": 5690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.284714928326854e-05,
+      "loss": 0.643,
+      "step": 5691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2844857348210648e-05,
+      "loss": 0.6889,
+      "step": 5692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2842565161019172e-05,
+      "loss": 0.3567,
+      "step": 5693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2840272721767785e-05,
+      "loss": 0.9535,
+      "step": 5694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.283798003053016e-05,
+      "loss": 0.5819,
+      "step": 5695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.283568708737999e-05,
+      "loss": 0.1831,
+      "step": 5696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2833393892390967e-05,
+      "loss": 0.3271,
+      "step": 5697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.28311004456368e-05,
+      "loss": 1.0488,
+      "step": 5698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2828806747191192e-05,
+      "loss": 0.2074,
+      "step": 5699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2826512797127865e-05,
+      "loss": 0.0742,
+      "step": 5700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2824218595520544e-05,
+      "loss": 1.8581,
+      "step": 5701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2821924142442974e-05,
+      "loss": 1.3246,
+      "step": 5702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2819629437968892e-05,
+      "loss": 0.0382,
+      "step": 5703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2817334482172047e-05,
+      "loss": 0.0723,
+      "step": 5704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2815039275126202e-05,
+      "loss": 0.8344,
+      "step": 5705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2812743816905127e-05,
+      "loss": 1.3896,
+      "step": 5706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.28104481075826e-05,
+      "loss": 0.708,
+      "step": 5707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.28081521472324e-05,
+      "loss": 0.2668,
+      "step": 5708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2805855935928315e-05,
+      "loss": 1.1265,
+      "step": 5709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.280355947374416e-05,
+      "loss": 0.0835,
+      "step": 5710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2801262760753725e-05,
+      "loss": 0.142,
+      "step": 5711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2798965797030838e-05,
+      "loss": 0.2301,
+      "step": 5712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2796668582649323e-05,
+      "loss": 0.0784,
+      "step": 5713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2794371117683008e-05,
+      "loss": 1.3505,
+      "step": 5714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2792073402205732e-05,
+      "loss": 0.5048,
+      "step": 5715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.278977543629135e-05,
+      "loss": 0.3488,
+      "step": 5716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2787477220013713e-05,
+      "loss": 1.119,
+      "step": 5717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2785178753446692e-05,
+      "loss": 0.9362,
+      "step": 5718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2782880036664157e-05,
+      "loss": 0.8621,
+      "step": 5719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2780581069739985e-05,
+      "loss": 0.2685,
+      "step": 5720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2778281852748068e-05,
+      "loss": 1.4164,
+      "step": 5721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2775982385762302e-05,
+      "loss": 1.8771,
+      "step": 5722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2773682668856595e-05,
+      "loss": 0.4315,
+      "step": 5723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2771382702104852e-05,
+      "loss": 0.4091,
+      "step": 5724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2769082485581003e-05,
+      "loss": 0.6228,
+      "step": 5725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2766782019358966e-05,
+      "loss": 0.2581,
+      "step": 5726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.276448130351269e-05,
+      "loss": 1.2458,
+      "step": 5727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.276218033811611e-05,
+      "loss": 0.5735,
+      "step": 5728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.275987912324319e-05,
+      "loss": 1.1487,
+      "step": 5729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2757577658967886e-05,
+      "loss": 0.1379,
+      "step": 5730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2755275945364162e-05,
+      "loss": 0.3009,
+      "step": 5731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2752973982505994e-05,
+      "loss": 1.4945,
+      "step": 5732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.275067177046738e-05,
+      "loss": 0.256,
+      "step": 5733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2748369309322296e-05,
+      "loss": 0.9828,
+      "step": 5734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2746066599144757e-05,
+      "loss": 0.6021,
+      "step": 5735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2743763640008767e-05,
+      "loss": 1.45,
+      "step": 5736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2741460431988344e-05,
+      "loss": 1.485,
+      "step": 5737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2739156975157506e-05,
+      "loss": 1.0071,
+      "step": 5738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.27368532695903e-05,
+      "loss": 0.8519,
+      "step": 5739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2734549315360752e-05,
+      "loss": 0.7991,
+      "step": 5740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2732245112542925e-05,
+      "loss": 0.1687,
+      "step": 5741
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2729940661210866e-05,
+      "loss": 0.1864,
+      "step": 5742
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.272763596143864e-05,
+      "loss": 1.01,
+      "step": 5743
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2725331013300334e-05,
+      "loss": 0.4335,
+      "step": 5744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2723025816870003e-05,
+      "loss": 0.405,
+      "step": 5745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2720720372221763e-05,
+      "loss": 0.4128,
+      "step": 5746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2718414679429695e-05,
+      "loss": 1.3608,
+      "step": 5747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2716108738567908e-05,
+      "loss": 0.2232,
+      "step": 5748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2713802549710517e-05,
+      "loss": 0.6529,
+      "step": 5749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2711496112931637e-05,
+      "loss": 1.0046,
+      "step": 5750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2709189428305404e-05,
+      "loss": 0.1343,
+      "step": 5751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.270688249590595e-05,
+      "loss": 0.0403,
+      "step": 5752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2704575315807426e-05,
+      "loss": 0.3037,
+      "step": 5753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2702267888083973e-05,
+      "loss": 1.3678,
+      "step": 5754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2699960212809757e-05,
+      "loss": 0.3972,
+      "step": 5755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.269765229005896e-05,
+      "loss": 0.388,
+      "step": 5756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2695344119905736e-05,
+      "loss": 1.1861,
+      "step": 5757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2693035702424284e-05,
+      "loss": 0.2446,
+      "step": 5758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2690727037688792e-05,
+      "loss": 1.5198,
+      "step": 5759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2688418125773463e-05,
+      "loss": 0.6197,
+      "step": 5760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2686108966752508e-05,
+      "loss": 0.7119,
+      "step": 5761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2683799560700133e-05,
+      "loss": 0.1349,
+      "step": 5762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.268148990769057e-05,
+      "loss": 0.1974,
+      "step": 5763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.267918000779805e-05,
+      "loss": 0.1447,
+      "step": 5764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2676869861096813e-05,
+      "loss": 0.3344,
+      "step": 5765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2674559467661106e-05,
+      "loss": 0.3822,
+      "step": 5766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2672248827565187e-05,
+      "loss": 0.0656,
+      "step": 5767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2669937940883323e-05,
+      "loss": 0.2716,
+      "step": 5768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2667626807689775e-05,
+      "loss": 1.5513,
+      "step": 5769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2665315428058832e-05,
+      "loss": 0.4616,
+      "step": 5770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.266300380206478e-05,
+      "loss": 0.0658,
+      "step": 5771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.266069192978191e-05,
+      "loss": 0.468,
+      "step": 5772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2658379811284534e-05,
+      "loss": 1.0313,
+      "step": 5773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2656067446646956e-05,
+      "loss": 1.253,
+      "step": 5774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2653754835943498e-05,
+      "loss": 0.1687,
+      "step": 5775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2651441979248492e-05,
+      "loss": 0.5637,
+      "step": 5776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2649128876636264e-05,
+      "loss": 0.3078,
+      "step": 5777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2646815528181156e-05,
+      "loss": 0.1871,
+      "step": 5778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2644501933957534e-05,
+      "loss": 0.4415,
+      "step": 5779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2642188094039748e-05,
+      "loss": 0.1072,
+      "step": 5780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.263987400850216e-05,
+      "loss": 0.9418,
+      "step": 5781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.263755967741915e-05,
+      "loss": 0.6965,
+      "step": 5782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.26352451008651e-05,
+      "loss": 0.8798,
+      "step": 5783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.26329302789144e-05,
+      "loss": 0.6886,
+      "step": 5784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.263061521164145e-05,
+      "loss": 0.8291,
+      "step": 5785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.262829989912065e-05,
+      "loss": 0.9806,
+      "step": 5786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.262598434142642e-05,
+      "loss": 1.1625,
+      "step": 5787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2623668538633185e-05,
+      "loss": 1.3984,
+      "step": 5788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2621352490815364e-05,
+      "loss": 0.1551,
+      "step": 5789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2619036198047406e-05,
+      "loss": 0.6524,
+      "step": 5790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.261671966040375e-05,
+      "loss": 0.6364,
+      "step": 5791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2614402877958856e-05,
+      "loss": 1.3834,
+      "step": 5792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2612085850787176e-05,
+      "loss": 1.3192,
+      "step": 5793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.260976857896318e-05,
+      "loss": 0.0898,
+      "step": 5794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2607451062561358e-05,
+      "loss": 0.2719,
+      "step": 5795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2605133301656183e-05,
+      "loss": 0.8004,
+      "step": 5796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2602815296322145e-05,
+      "loss": 0.5568,
+      "step": 5797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2600497046633755e-05,
+      "loss": 0.5474,
+      "step": 5798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2598178552665515e-05,
+      "loss": 0.661,
+      "step": 5799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2595859814491947e-05,
+      "loss": 0.6784,
+      "step": 5800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2593540832187573e-05,
+      "loss": 1.1592,
+      "step": 5801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2591221605826914e-05,
+      "loss": 1.2909,
+      "step": 5802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2588902135484525e-05,
+      "loss": 0.5739,
+      "step": 5803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.258658242123495e-05,
+      "loss": 0.0384,
+      "step": 5804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.258426246315274e-05,
+      "loss": 0.1854,
+      "step": 5805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.258194226131246e-05,
+      "loss": 0.543,
+      "step": 5806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2579621815788684e-05,
+      "loss": 0.2667,
+      "step": 5807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.257730112665599e-05,
+      "loss": 0.704,
+      "step": 5808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2574980193988965e-05,
+      "loss": 0.0611,
+      "step": 5809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2572659017862203e-05,
+      "loss": 0.1177,
+      "step": 5810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.257033759835031e-05,
+      "loss": 0.4007,
+      "step": 5811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2568015935527882e-05,
+      "loss": 1.1307,
+      "step": 5812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2565694029469557e-05,
+      "loss": 0.1594,
+      "step": 5813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2563371880249946e-05,
+      "loss": 0.0492,
+      "step": 5814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2561049487943697e-05,
+      "loss": 0.0457,
+      "step": 5815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.255872685262544e-05,
+      "loss": 0.9373,
+      "step": 5816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.255640397436983e-05,
+      "loss": 0.335,
+      "step": 5817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2554080853251518e-05,
+      "loss": 1.4868,
+      "step": 5818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.255175748934518e-05,
+      "loss": 0.1165,
+      "step": 5819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2549433882725477e-05,
+      "loss": 0.5818,
+      "step": 5820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.25471100334671e-05,
+      "loss": 0.9024,
+      "step": 5821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.254478594164473e-05,
+      "loss": 0.1645,
+      "step": 5822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2542461607333062e-05,
+      "loss": 0.441,
+      "step": 5823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2540137030606806e-05,
+      "loss": 0.9154,
+      "step": 5824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2537812211540674e-05,
+      "loss": 0.7321,
+      "step": 5825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2535487150209383e-05,
+      "loss": 0.7237,
+      "step": 5826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2533161846687662e-05,
+      "loss": 1.568,
+      "step": 5827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.253083630105024e-05,
+      "loss": 1.1786,
+      "step": 5828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2528510513371867e-05,
+      "loss": 2.0859,
+      "step": 5829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.25261844837273e-05,
+      "loss": 1.5205,
+      "step": 5830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.252385821219128e-05,
+      "loss": 0.3529,
+      "step": 5831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2521531698838583e-05,
+      "loss": 0.186,
+      "step": 5832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2519204943743988e-05,
+      "loss": 0.7081,
+      "step": 5833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.251687794698227e-05,
+      "loss": 0.9999,
+      "step": 5834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2514550708628217e-05,
+      "loss": 0.3823,
+      "step": 5835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2512223228756634e-05,
+      "loss": 1.4327,
+      "step": 5836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.250989550744232e-05,
+      "loss": 0.7815,
+      "step": 5837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.250756754476009e-05,
+      "loss": 0.8112,
+      "step": 5838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2505239340784766e-05,
+      "loss": 0.4041,
+      "step": 5839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.250291089559117e-05,
+      "loss": 1.3245,
+      "step": 5840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2500582209254147e-05,
+      "loss": 1.3125,
+      "step": 5841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.249825328184854e-05,
+      "loss": 0.1928,
+      "step": 5842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2495924113449192e-05,
+      "loss": 0.4134,
+      "step": 5843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2493594704130972e-05,
+      "loss": 0.1935,
+      "step": 5844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2491265053968742e-05,
+      "loss": 0.1617,
+      "step": 5845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2488935163037376e-05,
+      "loss": 1.1087,
+      "step": 5846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.248660503141176e-05,
+      "loss": 1.2714,
+      "step": 5847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2484274659166786e-05,
+      "loss": 0.5338,
+      "step": 5848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2481944046377347e-05,
+      "loss": 0.4255,
+      "step": 5849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.247961319311835e-05,
+      "loss": 0.116,
+      "step": 5850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2477282099464714e-05,
+      "loss": 1.6002,
+      "step": 5851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2474950765491353e-05,
+      "loss": 0.0724,
+      "step": 5852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2472619191273203e-05,
+      "loss": 0.1933,
+      "step": 5853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2470287376885194e-05,
+      "loss": 0.0364,
+      "step": 5854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2467955322402277e-05,
+      "loss": 0.0674,
+      "step": 5855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2465623027899395e-05,
+      "loss": 0.0916,
+      "step": 5856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2463290493451517e-05,
+      "loss": 1.5615,
+      "step": 5857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2460957719133608e-05,
+      "loss": 0.5167,
+      "step": 5858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2458624705020638e-05,
+      "loss": 0.7272,
+      "step": 5859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.24562914511876e-05,
+      "loss": 0.3158,
+      "step": 5860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2453957957709475e-05,
+      "loss": 0.4033,
+      "step": 5861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.245162422466127e-05,
+      "loss": 1.4698,
+      "step": 5862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2449290252117983e-05,
+      "loss": 0.6224,
+      "step": 5863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2446956040154637e-05,
+      "loss": 0.3308,
+      "step": 5864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2444621588846245e-05,
+      "loss": 0.5578,
+      "step": 5865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2442286898267838e-05,
+      "loss": 0.3811,
+      "step": 5866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2439951968494454e-05,
+      "loss": 0.7564,
+      "step": 5867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2437616799601142e-05,
+      "loss": 0.9248,
+      "step": 5868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2435281391662946e-05,
+      "loss": 0.9227,
+      "step": 5869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.243294574475493e-05,
+      "loss": 0.8565,
+      "step": 5870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2430609858952168e-05,
+      "loss": 1.4871,
+      "step": 5871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.242827373432972e-05,
+      "loss": 0.8474,
+      "step": 5872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.242593737096269e-05,
+      "loss": 0.59,
+      "step": 5873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2423600768926145e-05,
+      "loss": 1.4474,
+      "step": 5874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.24212639282952e-05,
+      "loss": 0.817,
+      "step": 5875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2418926849144964e-05,
+      "loss": 1.555,
+      "step": 5876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2416589531550533e-05,
+      "loss": 0.727,
+      "step": 5877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2414251975587042e-05,
+      "loss": 0.9652,
+      "step": 5878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2411914181329615e-05,
+      "loss": 0.7406,
+      "step": 5879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.240957614885339e-05,
+      "loss": 2.4825,
+      "step": 5880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2407237878233513e-05,
+      "loss": 0.9719,
+      "step": 5881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2404899369545134e-05,
+      "loss": 1.357,
+      "step": 5882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.240256062286341e-05,
+      "loss": 0.9028,
+      "step": 5883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2400221638263514e-05,
+      "loss": 0.3737,
+      "step": 5884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.239788241582062e-05,
+      "loss": 1.9045,
+      "step": 5885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2395542955609904e-05,
+      "loss": 0.4531,
+      "step": 5886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2393203257706565e-05,
+      "loss": 1.5346,
+      "step": 5887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2390863322185795e-05,
+      "loss": 0.2724,
+      "step": 5888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2388523149122797e-05,
+      "loss": 1.8939,
+      "step": 5889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2386182738592798e-05,
+      "loss": 0.891,
+      "step": 5890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2383842090671003e-05,
+      "loss": 0.7181,
+      "step": 5891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2381501205432647e-05,
+      "loss": 0.8113,
+      "step": 5892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2379160082952967e-05,
+      "loss": 0.41,
+      "step": 5893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2376818723307205e-05,
+      "loss": 0.7849,
+      "step": 5894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.237447712657061e-05,
+      "loss": 0.7891,
+      "step": 5895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.237213529281845e-05,
+      "loss": 0.6942,
+      "step": 5896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2369793222125976e-05,
+      "loss": 0.1154,
+      "step": 5897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2367450914568475e-05,
+      "loss": 0.9542,
+      "step": 5898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2365108370221224e-05,
+      "loss": 0.4364,
+      "step": 5899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2362765589159512e-05,
+      "loss": 0.9768,
+      "step": 5900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2360422571458642e-05,
+      "loss": 0.0409,
+      "step": 5901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2358079317193908e-05,
+      "loss": 0.0372,
+      "step": 5902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.235573582644063e-05,
+      "loss": 0.3795,
+      "step": 5903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2353392099274125e-05,
+      "loss": 0.5773,
+      "step": 5904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.235104813576972e-05,
+      "loss": 0.2811,
+      "step": 5905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2348703936002754e-05,
+      "loss": 1.6283,
+      "step": 5906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2346359500048565e-05,
+      "loss": 0.5707,
+      "step": 5907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.23440148279825e-05,
+      "loss": 0.3288,
+      "step": 5908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.234166991987992e-05,
+      "loss": 0.7489,
+      "step": 5909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2339324775816203e-05,
+      "loss": 0.4046,
+      "step": 5910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.23369793958667e-05,
+      "loss": 1.5359,
+      "step": 5911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2334633780106805e-05,
+      "loss": 0.3254,
+      "step": 5912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.23322879286119e-05,
+      "loss": 0.6215,
+      "step": 5913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2329941841457387e-05,
+      "loss": 0.1702,
+      "step": 5914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2327595518718663e-05,
+      "loss": 0.514,
+      "step": 5915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2325248960471146e-05,
+      "loss": 0.3947,
+      "step": 5916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2322902166790245e-05,
+      "loss": 0.1981,
+      "step": 5917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2320555137751394e-05,
+      "loss": 0.2365,
+      "step": 5918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2318207873430023e-05,
+      "loss": 0.6345,
+      "step": 5919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.231586037390157e-05,
+      "loss": 0.3995,
+      "step": 5920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.231351263924149e-05,
+      "loss": 0.9182,
+      "step": 5921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2311164669525235e-05,
+      "loss": 1.4211,
+      "step": 5922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.230881646482827e-05,
+      "loss": 0.2027,
+      "step": 5923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2306468025226064e-05,
+      "loss": 0.3459,
+      "step": 5924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2304119350794098e-05,
+      "loss": 0.2556,
+      "step": 5925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.230177044160786e-05,
+      "loss": 0.7987,
+      "step": 5926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.229942129774284e-05,
+      "loss": 0.7412,
+      "step": 5927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2297071919274542e-05,
+      "loss": 1.817,
+      "step": 5928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2294722306278473e-05,
+      "loss": 0.7999,
+      "step": 5929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2292372458830155e-05,
+      "loss": 0.8142,
+      "step": 5930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2290022377005104e-05,
+      "loss": 0.8356,
+      "step": 5931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.228767206087886e-05,
+      "loss": 0.4899,
+      "step": 5932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2285321510526953e-05,
+      "loss": 0.2424,
+      "step": 5933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2282970726024937e-05,
+      "loss": 0.5694,
+      "step": 5934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2280619707448362e-05,
+      "loss": 0.3411,
+      "step": 5935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2278268454872792e-05,
+      "loss": 1.3579,
+      "step": 5936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2275916968373793e-05,
+      "loss": 0.1623,
+      "step": 5937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.227356524802695e-05,
+      "loss": 0.653,
+      "step": 5938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2271213293907836e-05,
+      "loss": 0.438,
+      "step": 5939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2268861106092052e-05,
+      "loss": 1.7785,
+      "step": 5940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2266508684655193e-05,
+      "loss": 1.9304,
+      "step": 5941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.226415602967286e-05,
+      "loss": 1.1527,
+      "step": 5942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.226180314122068e-05,
+      "loss": 0.5421,
+      "step": 5943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.225945001937427e-05,
+      "loss": 0.5145,
+      "step": 5944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.225709666420925e-05,
+      "loss": 0.9561,
+      "step": 5945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2254743075801273e-05,
+      "loss": 0.7014,
+      "step": 5946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2252389254225965e-05,
+      "loss": 0.5565,
+      "step": 5947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2250035199558995e-05,
+      "loss": 0.4651,
+      "step": 5948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2247680911876012e-05,
+      "loss": 0.4021,
+      "step": 5949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.224532639125269e-05,
+      "loss": 0.4161,
+      "step": 5950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2242971637764694e-05,
+      "loss": 1.0058,
+      "step": 5951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.224061665148771e-05,
+      "loss": 1.6328,
+      "step": 5952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2238261432497432e-05,
+      "loss": 0.0709,
+      "step": 5953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.223590598086955e-05,
+      "loss": 0.7714,
+      "step": 5954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.223355029667977e-05,
+      "loss": 0.4237,
+      "step": 5955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2231194380003803e-05,
+      "loss": 0.4751,
+      "step": 5956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2228838230917374e-05,
+      "loss": 0.0381,
+      "step": 5957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.22264818494962e-05,
+      "loss": 0.3472,
+      "step": 5958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.222412523581602e-05,
+      "loss": 1.0078,
+      "step": 5959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.222176838995258e-05,
+      "loss": 0.0594,
+      "step": 5960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.221941131198162e-05,
+      "loss": 0.5222,
+      "step": 5961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2217054001978908e-05,
+      "loss": 1.0142,
+      "step": 5962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2214696460020194e-05,
+      "loss": 0.2075,
+      "step": 5963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.221233868618126e-05,
+      "loss": 0.3403,
+      "step": 5964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2209980680537882e-05,
+      "loss": 0.6737,
+      "step": 5965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2207622443165843e-05,
+      "loss": 0.5115,
+      "step": 5966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.220526397414094e-05,
+      "loss": 0.0522,
+      "step": 5967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.220290527353897e-05,
+      "loss": 0.3035,
+      "step": 5968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2200546341435747e-05,
+      "loss": 0.3247,
+      "step": 5969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2198187177907085e-05,
+      "loss": 1.2606,
+      "step": 5970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2195827783028806e-05,
+      "loss": 0.3933,
+      "step": 5971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2193468156876745e-05,
+      "loss": 1.2308,
+      "step": 5972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2191108299526737e-05,
+      "loss": 0.2691,
+      "step": 5973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2188748211054628e-05,
+      "loss": 0.1749,
+      "step": 5974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2186387891536274e-05,
+      "loss": 1.2781,
+      "step": 5975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2184027341047527e-05,
+      "loss": 1.0509,
+      "step": 5976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2181666559664265e-05,
+      "loss": 0.4367,
+      "step": 5977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.217930554746236e-05,
+      "loss": 0.9683,
+      "step": 5978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2176944304517694e-05,
+      "loss": 1.0998,
+      "step": 5979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.217458283090616e-05,
+      "loss": 0.9147,
+      "step": 5980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2172221126703652e-05,
+      "loss": 0.7189,
+      "step": 5981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2169859191986076e-05,
+      "loss": 0.8877,
+      "step": 5982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2167497026829344e-05,
+      "loss": 0.3776,
+      "step": 5983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2165134631309382e-05,
+      "loss": 0.6325,
+      "step": 5984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2162772005502103e-05,
+      "loss": 0.1882,
+      "step": 5985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.216040914948346e-05,
+      "loss": 0.7445,
+      "step": 5986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.215804606332939e-05,
+      "loss": 0.7336,
+      "step": 5987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2155682747115832e-05,
+      "loss": 0.7335,
+      "step": 5988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2153319200918748e-05,
+      "loss": 0.6479,
+      "step": 5989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.215095542481411e-05,
+      "loss": 1.4804,
+      "step": 5990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2148591418877883e-05,
+      "loss": 0.3785,
+      "step": 5991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2146227183186046e-05,
+      "loss": 0.3664,
+      "step": 5992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2143862717814587e-05,
+      "loss": 0.0808,
+      "step": 5993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2141498022839497e-05,
+      "loss": 0.5077,
+      "step": 5994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2139133098336784e-05,
+      "loss": 0.1495,
+      "step": 5995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.213676794438245e-05,
+      "loss": 0.5502,
+      "step": 5996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2134402561052515e-05,
+      "loss": 0.4577,
+      "step": 5997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2132036948423004e-05,
+      "loss": 1.7277,
+      "step": 5998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2129671106569938e-05,
+      "loss": 0.55,
+      "step": 5999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2127305035569364e-05,
+      "loss": 0.5728,
+      "step": 6000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.212493873549733e-05,
+      "loss": 1.3718,
+      "step": 6001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.212257220642988e-05,
+      "loss": 0.0348,
+      "step": 6002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2120205448443073e-05,
+      "loss": 0.0469,
+      "step": 6003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.211783846161299e-05,
+      "loss": 1.4554,
+      "step": 6004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2115471246015694e-05,
+      "loss": 0.0369,
+      "step": 6005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.211310380172727e-05,
+      "loss": 0.521,
+      "step": 6006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2110736128823815e-05,
+      "loss": 0.1994,
+      "step": 6007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.210836822738141e-05,
+      "loss": 0.0332,
+      "step": 6008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2106000097476177e-05,
+      "loss": 0.0436,
+      "step": 6009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2103631739184217e-05,
+      "loss": 1.0391,
+      "step": 6010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.210126315258165e-05,
+      "loss": 0.1774,
+      "step": 6011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2098894337744606e-05,
+      "loss": 0.2061,
+      "step": 6012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2096525294749213e-05,
+      "loss": 0.6505,
+      "step": 6013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.209415602367162e-05,
+      "loss": 1.8112,
+      "step": 6014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2091786524587968e-05,
+      "loss": 0.3482,
+      "step": 6015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2089416797574415e-05,
+      "loss": 1.6949,
+      "step": 6016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2087046842707126e-05,
+      "loss": 0.7254,
+      "step": 6017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2084676660062272e-05,
+      "loss": 0.1078,
+      "step": 6018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2082306249716026e-05,
+      "loss": 0.45,
+      "step": 6019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2079935611744573e-05,
+      "loss": 0.7491,
+      "step": 6020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2077564746224112e-05,
+      "loss": 1.0384,
+      "step": 6021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.207519365323084e-05,
+      "loss": 0.0763,
+      "step": 6022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2072822332840956e-05,
+      "loss": 0.1373,
+      "step": 6023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2070450785130683e-05,
+      "loss": 0.8247,
+      "step": 6024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.206807901017624e-05,
+      "loss": 0.3266,
+      "step": 6025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2065707008053857e-05,
+      "loss": 0.8866,
+      "step": 6026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.206333477883977e-05,
+      "loss": 0.591,
+      "step": 6027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2060962322610215e-05,
+      "loss": 0.4593,
+      "step": 6028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.205858963944145e-05,
+      "loss": 1.036,
+      "step": 6029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.205621672940974e-05,
+      "loss": 0.3553,
+      "step": 6030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.205384359259133e-05,
+      "loss": 0.4814,
+      "step": 6031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.205147022906252e-05,
+      "loss": 1.8472,
+      "step": 6032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2049096638899562e-05,
+      "loss": 1.2592,
+      "step": 6033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.204672282217876e-05,
+      "loss": 1.0418,
+      "step": 6034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2044348778976404e-05,
+      "loss": 0.3896,
+      "step": 6035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2041974509368795e-05,
+      "loss": 0.5799,
+      "step": 6036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2039600013432243e-05,
+      "loss": 0.8349,
+      "step": 6037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2037225291243065e-05,
+      "loss": 1.0892,
+      "step": 6038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2034850342877583e-05,
+      "loss": 0.0781,
+      "step": 6039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.203247516841213e-05,
+      "loss": 1.4487,
+      "step": 6040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2030099767923046e-05,
+      "loss": 0.8124,
+      "step": 6041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2027724141486667e-05,
+      "loss": 0.6867,
+      "step": 6042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2025348289179354e-05,
+      "loss": 0.1915,
+      "step": 6043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2022972211077462e-05,
+      "loss": 0.1991,
+      "step": 6044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2020595907257363e-05,
+      "loss": 0.4979,
+      "step": 6045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2018219377795428e-05,
+      "loss": 0.577,
+      "step": 6046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2015842622768037e-05,
+      "loss": 0.8217,
+      "step": 6047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2013465642251585e-05,
+      "loss": 0.5935,
+      "step": 6048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2011088436322464e-05,
+      "loss": 0.5015,
+      "step": 6049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.200871100505708e-05,
+      "loss": 0.2861,
+      "step": 6050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.200633334853183e-05,
+      "loss": 0.0677,
+      "step": 6051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2003955466823156e-05,
+      "loss": 1.5118,
+      "step": 6052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2001577360007464e-05,
+      "loss": 1.5773,
+      "step": 6053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1999199028161194e-05,
+      "loss": 0.2428,
+      "step": 6054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1996820471360785e-05,
+      "loss": 1.6813,
+      "step": 6055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1994441689682682e-05,
+      "loss": 0.8969,
+      "step": 6056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1992062683203336e-05,
+      "loss": 0.468,
+      "step": 6057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1989683451999218e-05,
+      "loss": 1.3829,
+      "step": 6058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.198730399614679e-05,
+      "loss": 0.1926,
+      "step": 6059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1984924315722525e-05,
+      "loss": 0.318,
+      "step": 6060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1982544410802918e-05,
+      "loss": 0.1083,
+      "step": 6061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1980164281464435e-05,
+      "loss": 0.2942,
+      "step": 6062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.19777839277836e-05,
+      "loss": 1.2525,
+      "step": 6063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1975403349836904e-05,
+      "loss": 1.6392,
+      "step": 6064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1973022547700863e-05,
+      "loss": 0.3814,
+      "step": 6065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.197064152145199e-05,
+      "loss": 0.4628,
+      "step": 6066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1968260271166815e-05,
+      "loss": 0.4168,
+      "step": 6067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.196587879692187e-05,
+      "loss": 0.1958,
+      "step": 6068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.19634970987937e-05,
+      "loss": 0.2078,
+      "step": 6069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.196111517685885e-05,
+      "loss": 0.8106,
+      "step": 6070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1958733031193873e-05,
+      "loss": 0.3038,
+      "step": 6071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.195635066187533e-05,
+      "loss": 0.1583,
+      "step": 6072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.19539680689798e-05,
+      "loss": 1.358,
+      "step": 6073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1951585252583843e-05,
+      "loss": 0.104,
+      "step": 6074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.194920221276406e-05,
+      "loss": 0.4674,
+      "step": 6075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1946818949597033e-05,
+      "loss": 0.6941,
+      "step": 6076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1944435463159356e-05,
+      "loss": 1.6806,
+      "step": 6077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1942051753527642e-05,
+      "loss": 0.8866,
+      "step": 6078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.19396678207785e-05,
+      "loss": 0.6768,
+      "step": 6079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1937283664988553e-05,
+      "loss": 0.5881,
+      "step": 6080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.193489928623442e-05,
+      "loss": 0.8567,
+      "step": 6081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1932514684592742e-05,
+      "loss": 0.9435,
+      "step": 6082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1930129860140157e-05,
+      "loss": 0.7042,
+      "step": 6083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.192774481295332e-05,
+      "loss": 1.0516,
+      "step": 6084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1925359543108872e-05,
+      "loss": 1.1212,
+      "step": 6085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1922974050683493e-05,
+      "loss": 0.6758,
+      "step": 6086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1920588335753837e-05,
+      "loss": 0.9229,
+      "step": 6087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.191820239839659e-05,
+      "loss": 0.7893,
+      "step": 6088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.191581623868843e-05,
+      "loss": 0.7386,
+      "step": 6089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1913429856706055e-05,
+      "loss": 1.8227,
+      "step": 6090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1911043252526158e-05,
+      "loss": 1.0104,
+      "step": 6091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.190865642622545e-05,
+      "loss": 0.2408,
+      "step": 6092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1906269377880635e-05,
+      "loss": 1.129,
+      "step": 6093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.190388210756844e-05,
+      "loss": 0.7303,
+      "step": 6094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.190149461536559e-05,
+      "loss": 1.7281,
+      "step": 6095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1899106901348818e-05,
+      "loss": 0.8737,
+      "step": 6096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.189671896559486e-05,
+      "loss": 0.6433,
+      "step": 6097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1894330808180474e-05,
+      "loss": 0.6615,
+      "step": 6098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1891942429182414e-05,
+      "loss": 0.4501,
+      "step": 6099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.188955382867744e-05,
+      "loss": 0.609,
+      "step": 6100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1887165006742315e-05,
+      "loss": 0.9806,
+      "step": 6101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1884775963453827e-05,
+      "loss": 0.3208,
+      "step": 6102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.188238669888875e-05,
+      "loss": 0.6243,
+      "step": 6103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1879997213123883e-05,
+      "loss": 0.0341,
+      "step": 6104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1877607506236015e-05,
+      "loss": 0.0403,
+      "step": 6105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.187521757830196e-05,
+      "loss": 0.0361,
+      "step": 6106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1872827429398534e-05,
+      "loss": 1.552,
+      "step": 6107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1870437059602538e-05,
+      "loss": 0.033,
+      "step": 6108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.186804646899082e-05,
+      "loss": 1.3925,
+      "step": 6109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1865655657640195e-05,
+      "loss": 1.4349,
+      "step": 6110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1863264625627516e-05,
+      "loss": 0.0295,
+      "step": 6111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1860873373029623e-05,
+      "loss": 0.7327,
+      "step": 6112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1858481899923377e-05,
+      "loss": 0.0493,
+      "step": 6113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1856090206385637e-05,
+      "loss": 0.225,
+      "step": 6114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1853698292493275e-05,
+      "loss": 0.5231,
+      "step": 6115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.185130615832316e-05,
+      "loss": 1.7483,
+      "step": 6116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.184891380395218e-05,
+      "loss": 1.1726,
+      "step": 6117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1846521229457226e-05,
+      "loss": 0.4974,
+      "step": 6118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1844128434915192e-05,
+      "loss": 1.518,
+      "step": 6119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1841735420402988e-05,
+      "loss": 0.6283,
+      "step": 6120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.183934218599752e-05,
+      "loss": 0.1944,
+      "step": 6121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.183694873177571e-05,
+      "loss": 0.5603,
+      "step": 6122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1834555057814477e-05,
+      "loss": 0.4717,
+      "step": 6123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.183216116419076e-05,
+      "loss": 0.2446,
+      "step": 6124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.18297670509815e-05,
+      "loss": 0.4486,
+      "step": 6125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.182737271826364e-05,
+      "loss": 0.7793,
+      "step": 6126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.182497816611414e-05,
+      "loss": 1.1751,
+      "step": 6127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1822583394609946e-05,
+      "loss": 0.493,
+      "step": 6128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1820188403828045e-05,
+      "loss": 0.7664,
+      "step": 6129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1817793193845398e-05,
+      "loss": 0.6922,
+      "step": 6130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.181539776473899e-05,
+      "loss": 1.5225,
+      "step": 6131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1813002116585816e-05,
+      "loss": 0.7265,
+      "step": 6132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1810606249462866e-05,
+      "loss": 1.7291,
+      "step": 6133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1808210163447143e-05,
+      "loss": 0.2278,
+      "step": 6134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1805813858615658e-05,
+      "loss": 0.5604,
+      "step": 6135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1803417335045433e-05,
+      "loss": 0.1803,
+      "step": 6136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1801020592813486e-05,
+      "loss": 0.3713,
+      "step": 6137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1798623631996853e-05,
+      "loss": 0.3495,
+      "step": 6138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1796226452672565e-05,
+      "loss": 1.1636,
+      "step": 6139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.179382905491768e-05,
+      "loss": 0.4392,
+      "step": 6140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1791431438809238e-05,
+      "loss": 0.7506,
+      "step": 6141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1789033604424305e-05,
+      "loss": 0.9481,
+      "step": 6142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.178663555183994e-05,
+      "loss": 0.4594,
+      "step": 6143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.178423728113323e-05,
+      "loss": 0.284,
+      "step": 6144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1781838792381242e-05,
+      "loss": 0.9622,
+      "step": 6145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.177944008566107e-05,
+      "loss": 0.9333,
+      "step": 6146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1777041161049807e-05,
+      "loss": 1.0623,
+      "step": 6147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1774642018624552e-05,
+      "loss": 0.4111,
+      "step": 6148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1772242658462417e-05,
+      "loss": 0.8495,
+      "step": 6149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.176984308064052e-05,
+      "loss": 0.4123,
+      "step": 6150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1767443285235972e-05,
+      "loss": 0.1458,
+      "step": 6151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1765043272325917e-05,
+      "loss": 0.2957,
+      "step": 6152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.176264304198748e-05,
+      "loss": 1.2653,
+      "step": 6153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1760242594297808e-05,
+      "loss": 0.4844,
+      "step": 6154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1757841929334052e-05,
+      "loss": 0.5223,
+      "step": 6155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1755441047173373e-05,
+      "loss": 1.1162,
+      "step": 6156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.175303994789293e-05,
+      "loss": 0.406,
+      "step": 6157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1750638631569892e-05,
+      "loss": 0.7922,
+      "step": 6158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1748237098281444e-05,
+      "loss": 0.1486,
+      "step": 6159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.174583534810477e-05,
+      "loss": 0.5965,
+      "step": 6160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1743433381117063e-05,
+      "loss": 1.5071,
+      "step": 6161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.174103119739551e-05,
+      "loss": 0.2867,
+      "step": 6162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1738628797017338e-05,
+      "loss": 1.5018,
+      "step": 6163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.173622618005974e-05,
+      "loss": 1.1637,
+      "step": 6164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1733823346599954e-05,
+      "loss": 0.595,
+      "step": 6165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.173142029671519e-05,
+      "loss": 0.344,
+      "step": 6166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1729017030482698e-05,
+      "loss": 1.4297,
+      "step": 6167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.17266135479797e-05,
+      "loss": 0.924,
+      "step": 6168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1724209849283462e-05,
+      "loss": 0.1373,
+      "step": 6169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1721805934471233e-05,
+      "loss": 0.5846,
+      "step": 6170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1719401803620273e-05,
+      "loss": 0.8817,
+      "step": 6171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.171699745680785e-05,
+      "loss": 0.8818,
+      "step": 6172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.171459289411124e-05,
+      "loss": 0.1239,
+      "step": 6173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1712188115607725e-05,
+      "loss": 0.3421,
+      "step": 6174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1709783121374604e-05,
+      "loss": 0.3817,
+      "step": 6175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1707377911489158e-05,
+      "loss": 1.0544,
+      "step": 6176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.17049724860287e-05,
+      "loss": 0.7018,
+      "step": 6177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1702566845070536e-05,
+      "loss": 1.3773,
+      "step": 6178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.170016098869199e-05,
+      "loss": 0.5964,
+      "step": 6179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.169775491697038e-05,
+      "loss": 0.674,
+      "step": 6180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1695348629983035e-05,
+      "loss": 0.3715,
+      "step": 6181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1692942127807302e-05,
+      "loss": 1.4265,
+      "step": 6182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.169053541052052e-05,
+      "loss": 1.386,
+      "step": 6183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1688128478200043e-05,
+      "loss": 0.2313,
+      "step": 6184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.168572133092322e-05,
+      "loss": 0.6057,
+      "step": 6185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1683313968767437e-05,
+      "loss": 0.4144,
+      "step": 6186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1680906391810046e-05,
+      "loss": 0.2063,
+      "step": 6187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.167849860012844e-05,
+      "loss": 0.6754,
+      "step": 6188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1676090593799996e-05,
+      "loss": 0.1725,
+      "step": 6189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1673682372902114e-05,
+      "loss": 0.4316,
+      "step": 6190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1671273937512193e-05,
+      "loss": 0.0568,
+      "step": 6191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1668865287707642e-05,
+      "loss": 0.9233,
+      "step": 6192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1666456423565865e-05,
+      "loss": 0.4035,
+      "step": 6193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1664047345164296e-05,
+      "loss": 0.5397,
+      "step": 6194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1661638052580353e-05,
+      "loss": 0.9986,
+      "step": 6195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1659228545891475e-05,
+      "loss": 0.5589,
+      "step": 6196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1656818825175104e-05,
+      "loss": 0.0772,
+      "step": 6197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1654408890508684e-05,
+      "loss": 0.314,
+      "step": 6198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1651998741969677e-05,
+      "loss": 0.5132,
+      "step": 6199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.164958837963554e-05,
+      "loss": 0.6193,
+      "step": 6200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1647177803583744e-05,
+      "loss": 0.2191,
+      "step": 6201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1644767013891767e-05,
+      "loss": 1.9598,
+      "step": 6202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.164235601063708e-05,
+      "loss": 0.2886,
+      "step": 6203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.163994479389719e-05,
+      "loss": 1.4105,
+      "step": 6204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1637533363749586e-05,
+      "loss": 0.0564,
+      "step": 6205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.163512172027177e-05,
+      "loss": 1.4159,
+      "step": 6206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1632709863541252e-05,
+      "loss": 0.6034,
+      "step": 6207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1630297793635543e-05,
+      "loss": 1.3045,
+      "step": 6208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.162788551063218e-05,
+      "loss": 0.4381,
+      "step": 6209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.162547301460869e-05,
+      "loss": 0.1676,
+      "step": 6210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1623060305642603e-05,
+      "loss": 0.9077,
+      "step": 6211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.162064738381147e-05,
+      "loss": 0.5469,
+      "step": 6212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.161823424919284e-05,
+      "loss": 0.3456,
+      "step": 6213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.161582090186427e-05,
+      "loss": 0.302,
+      "step": 6214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1613407341903333e-05,
+      "loss": 1.2163,
+      "step": 6215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.161099356938759e-05,
+      "loss": 0.0867,
+      "step": 6216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1608579584394625e-05,
+      "loss": 0.3797,
+      "step": 6217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1606165387002023e-05,
+      "loss": 0.1523,
+      "step": 6218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1603750977287375e-05,
+      "loss": 0.279,
+      "step": 6219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.160133635532828e-05,
+      "loss": 0.8401,
+      "step": 6220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1598921521202346e-05,
+      "loss": 0.1624,
+      "step": 6221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.159650647498718e-05,
+      "loss": 0.7827,
+      "step": 6222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1594091216760413e-05,
+      "loss": 0.4793,
+      "step": 6223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.159167574659966e-05,
+      "loss": 0.3093,
+      "step": 6224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1589260064582563e-05,
+      "loss": 1.6062,
+      "step": 6225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1586844170786753e-05,
+      "loss": 0.5896,
+      "step": 6226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1584428065289885e-05,
+      "loss": 1.1982,
+      "step": 6227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1582011748169604e-05,
+      "loss": 1.1431,
+      "step": 6228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1579595219503583e-05,
+      "loss": 0.3216,
+      "step": 6229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1577178479369474e-05,
+      "loss": 0.3781,
+      "step": 6230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1574761527844965e-05,
+      "loss": 0.6925,
+      "step": 6231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1572344365007724e-05,
+      "loss": 1.4518,
+      "step": 6232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.156992699093545e-05,
+      "loss": 0.9322,
+      "step": 6233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.156750940570583e-05,
+      "loss": 0.2892,
+      "step": 6234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1565091609396566e-05,
+      "loss": 0.6649,
+      "step": 6235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.156267360208537e-05,
+      "loss": 1.2939,
+      "step": 6236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1560255383849952e-05,
+      "loss": 0.6799,
+      "step": 6237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1557836954768037e-05,
+      "loss": 0.1499,
+      "step": 6238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.155541831491735e-05,
+      "loss": 0.7957,
+      "step": 6239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1552999464375634e-05,
+      "loss": 1.0276,
+      "step": 6240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1550580403220617e-05,
+      "loss": 1.3581,
+      "step": 6241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.154816113153006e-05,
+      "loss": 0.5024,
+      "step": 6242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.154574164938171e-05,
+      "loss": 0.5762,
+      "step": 6243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1543321956853335e-05,
+      "loss": 1.1144,
+      "step": 6244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1540902054022702e-05,
+      "loss": 1.0642,
+      "step": 6245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1538481940967585e-05,
+      "loss": 0.066,
+      "step": 6246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1536061617765766e-05,
+      "loss": 0.7158,
+      "step": 6247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.153364108449504e-05,
+      "loss": 0.3377,
+      "step": 6248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1531220341233196e-05,
+      "loss": 0.5463,
+      "step": 6249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.152879938805804e-05,
+      "loss": 0.7689,
+      "step": 6250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1526378225047375e-05,
+      "loss": 0.1535,
+      "step": 6251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.152395685227903e-05,
+      "loss": 0.1972,
+      "step": 6252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1521535269830822e-05,
+      "loss": 0.2697,
+      "step": 6253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1519113477780576e-05,
+      "loss": 0.7141,
+      "step": 6254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1516691476206133e-05,
+      "loss": 0.4256,
+      "step": 6255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1514269265185335e-05,
+      "loss": 0.2093,
+      "step": 6256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.151184684479603e-05,
+      "loss": 0.4087,
+      "step": 6257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1509424215116075e-05,
+      "loss": 0.0586,
+      "step": 6258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1507001376223336e-05,
+      "loss": 0.6337,
+      "step": 6259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1504578328195682e-05,
+      "loss": 0.0448,
+      "step": 6260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.150215507111099e-05,
+      "loss": 1.6669,
+      "step": 6261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.149973160504714e-05,
+      "loss": 0.0423,
+      "step": 6262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1497307930082028e-05,
+      "loss": 1.8164,
+      "step": 6263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1494884046293545e-05,
+      "loss": 1.1632,
+      "step": 6264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.14924599537596e-05,
+      "loss": 0.0914,
+      "step": 6265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1490035652558098e-05,
+      "loss": 0.419,
+      "step": 6266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.148761114276696e-05,
+      "loss": 0.3288,
+      "step": 6267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1485186424464104e-05,
+      "loss": 0.0461,
+      "step": 6268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.148276149772747e-05,
+      "loss": 0.3243,
+      "step": 6269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1480336362634985e-05,
+      "loss": 1.0025,
+      "step": 6270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.14779110192646e-05,
+      "loss": 0.5754,
+      "step": 6271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1475485467694267e-05,
+      "loss": 0.5588,
+      "step": 6272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1473059708001936e-05,
+      "loss": 1.4344,
+      "step": 6273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1470633740265574e-05,
+      "loss": 0.1916,
+      "step": 6274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.146820756456315e-05,
+      "loss": 0.0904,
+      "step": 6275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1465781180972643e-05,
+      "loss": 0.8443,
+      "step": 6276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.146335458957204e-05,
+      "loss": 1.5533,
+      "step": 6277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.146092779043933e-05,
+      "loss": 1.7669,
+      "step": 6278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1458500783652502e-05,
+      "loss": 0.8103,
+      "step": 6279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.145607356928957e-05,
+      "loss": 0.2217,
+      "step": 6280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1453646147428537e-05,
+      "loss": 1.0892,
+      "step": 6281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1451218518147433e-05,
+      "loss": 0.8344,
+      "step": 6282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.144879068152427e-05,
+      "loss": 0.9798,
+      "step": 6283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1446362637637083e-05,
+      "loss": 0.6822,
+      "step": 6284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1443934386563905e-05,
+      "loss": 0.4899,
+      "step": 6285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.144150592838279e-05,
+      "loss": 1.3054,
+      "step": 6286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1439077263171775e-05,
+      "loss": 0.6179,
+      "step": 6287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.143664839100893e-05,
+      "loss": 1.1602,
+      "step": 6288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1434219311972308e-05,
+      "loss": 1.024,
+      "step": 6289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1431790026139985e-05,
+      "loss": 1.4543,
+      "step": 6290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.142936053359004e-05,
+      "loss": 0.598,
+      "step": 6291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1426930834400557e-05,
+      "loss": 1.194,
+      "step": 6292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1424500928649618e-05,
+      "loss": 0.3772,
+      "step": 6293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1422070816415333e-05,
+      "loss": 0.6029,
+      "step": 6294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.14196404977758e-05,
+      "loss": 0.2346,
+      "step": 6295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1417209972809127e-05,
+      "loss": 0.6991,
+      "step": 6296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1414779241593427e-05,
+      "loss": 0.0881,
+      "step": 6297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1412348304206838e-05,
+      "loss": 0.0941,
+      "step": 6298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1409917160727473e-05,
+      "loss": 0.5728,
+      "step": 6299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1407485811233483e-05,
+      "loss": 0.4695,
+      "step": 6300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1405054255803005e-05,
+      "loss": 1.4373,
+      "step": 6301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.140262249451419e-05,
+      "loss": 0.1908,
+      "step": 6302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1400190527445195e-05,
+      "loss": 0.9512,
+      "step": 6303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.139775835467418e-05,
+      "loss": 0.3384,
+      "step": 6304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1395325976279326e-05,
+      "loss": 0.6842,
+      "step": 6305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1392893392338797e-05,
+      "loss": 1.47,
+      "step": 6306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1390460602930783e-05,
+      "loss": 0.2673,
+      "step": 6307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.138802760813347e-05,
+      "loss": 0.2447,
+      "step": 6308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.138559440802506e-05,
+      "loss": 0.3204,
+      "step": 6309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1383161002683745e-05,
+      "loss": 0.091,
+      "step": 6310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1380727392187748e-05,
+      "loss": 0.064,
+      "step": 6311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1378293576615276e-05,
+      "loss": 0.3682,
+      "step": 6312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1375859556044557e-05,
+      "loss": 0.2347,
+      "step": 6313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1373425330553818e-05,
+      "loss": 0.1904,
+      "step": 6314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.13709909002213e-05,
+      "loss": 0.0906,
+      "step": 6315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.136855626512523e-05,
+      "loss": 0.2288,
+      "step": 6316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.136612142534388e-05,
+      "loss": 0.5396,
+      "step": 6317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1363686380955486e-05,
+      "loss": 0.4755,
+      "step": 6318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.136125113203832e-05,
+      "loss": 1.6804,
+      "step": 6319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.135881567867065e-05,
+      "loss": 0.2895,
+      "step": 6320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1356380020930746e-05,
+      "loss": 1.5184,
+      "step": 6321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.13539441588969e-05,
+      "loss": 0.0515,
+      "step": 6322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.135150809264739e-05,
+      "loss": 1.4929,
+      "step": 6323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.134907182226052e-05,
+      "loss": 1.2688,
+      "step": 6324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1346635347814583e-05,
+      "loss": 0.6518,
+      "step": 6325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1344198669387898e-05,
+      "loss": 1.4045,
+      "step": 6326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1341761787058772e-05,
+      "loss": 0.0704,
+      "step": 6327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1339324700905523e-05,
+      "loss": 0.1336,
+      "step": 6328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.133688741100649e-05,
+      "loss": 1.2879,
+      "step": 6329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.133444991744e-05,
+      "loss": 0.7118,
+      "step": 6330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1332012220284394e-05,
+      "loss": 0.1151,
+      "step": 6331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1329574319618022e-05,
+      "loss": 1.0141,
+      "step": 6332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1327136215519236e-05,
+      "loss": 1.084,
+      "step": 6333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.13246979080664e-05,
+      "loss": 1.0846,
+      "step": 6334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1322259397337875e-05,
+      "loss": 0.7713,
+      "step": 6335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.131982068341204e-05,
+      "loss": 0.1706,
+      "step": 6336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1317381766367277e-05,
+      "loss": 0.5874,
+      "step": 6337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.131494264628197e-05,
+      "loss": 0.8968,
+      "step": 6338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1312503323234506e-05,
+      "loss": 0.8943,
+      "step": 6339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1310063797303294e-05,
+      "loss": 0.5459,
+      "step": 6340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1307624068566736e-05,
+      "loss": 0.5776,
+      "step": 6341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1305184137103247e-05,
+      "loss": 0.9348,
+      "step": 6342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1302744002991245e-05,
+      "loss": 1.4445,
+      "step": 6343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.130030366630916e-05,
+      "loss": 0.4819,
+      "step": 6344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.129786312713541e-05,
+      "loss": 0.4202,
+      "step": 6345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1295422385548455e-05,
+      "loss": 1.0282,
+      "step": 6346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.129298144162672e-05,
+      "loss": 0.7067,
+      "step": 6347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1290540295448675e-05,
+      "loss": 0.6677,
+      "step": 6348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1288098947092768e-05,
+      "loss": 0.6946,
+      "step": 6349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1285657396637466e-05,
+      "loss": 0.3307,
+      "step": 6350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1283215644161235e-05,
+      "loss": 0.6662,
+      "step": 6351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1280773689742565e-05,
+      "loss": 0.0363,
+      "step": 6352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1278331533459928e-05,
+      "loss": 0.1022,
+      "step": 6353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1275889175391823e-05,
+      "loss": 1.7168,
+      "step": 6354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1273446615616744e-05,
+      "loss": 0.3412,
+      "step": 6355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1271003854213196e-05,
+      "loss": 0.7047,
+      "step": 6356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1268560891259687e-05,
+      "loss": 1.6637,
+      "step": 6357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1266117726834736e-05,
+      "loss": 1.3918,
+      "step": 6358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.126367436101687e-05,
+      "loss": 0.1739,
+      "step": 6359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1261230793884612e-05,
+      "loss": 1.6032,
+      "step": 6360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1258787025516507e-05,
+      "loss": 0.3591,
+      "step": 6361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.125634305599108e-05,
+      "loss": 1.4736,
+      "step": 6362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.12538988853869e-05,
+      "loss": 0.3231,
+      "step": 6363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1251454513782514e-05,
+      "loss": 0.0352,
+      "step": 6364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.124900994125648e-05,
+      "loss": 0.4906,
+      "step": 6365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1246565167887373e-05,
+      "loss": 0.846,
+      "step": 6366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1244120193753762e-05,
+      "loss": 0.2822,
+      "step": 6367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1241675018934238e-05,
+      "loss": 0.5111,
+      "step": 6368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1239229643507384e-05,
+      "loss": 1.2606,
+      "step": 6369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1236784067551788e-05,
+      "loss": 0.09,
+      "step": 6370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1234338291146056e-05,
+      "loss": 0.0809,
+      "step": 6371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.12318923143688e-05,
+      "loss": 0.0853,
+      "step": 6372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1229446137298633e-05,
+      "loss": 0.1428,
+      "step": 6373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1226999760014162e-05,
+      "loss": 0.0822,
+      "step": 6374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1224553182594027e-05,
+      "loss": 0.6603,
+      "step": 6375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1222106405116856e-05,
+      "loss": 0.8703,
+      "step": 6376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1219659427661288e-05,
+      "loss": 1.4806,
+      "step": 6377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1217212250305968e-05,
+      "loss": 1.0139,
+      "step": 6378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1214764873129555e-05,
+      "loss": 0.2201,
+      "step": 6379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1212317296210705e-05,
+      "loss": 0.8317,
+      "step": 6380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1209869519628072e-05,
+      "loss": 0.9421,
+      "step": 6381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.120742154346034e-05,
+      "loss": 1.8467,
+      "step": 6382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.120497336778619e-05,
+      "loss": 0.835,
+      "step": 6383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1202524992684292e-05,
+      "loss": 1.3866,
+      "step": 6384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1200076418233344e-05,
+      "loss": 1.3411,
+      "step": 6385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1197627644512048e-05,
+      "loss": 0.6652,
+      "step": 6386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.11951786715991e-05,
+      "loss": 0.5003,
+      "step": 6387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1192729499573213e-05,
+      "loss": 0.2332,
+      "step": 6388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1190280128513104e-05,
+      "loss": 0.6653,
+      "step": 6389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1187830558497495e-05,
+      "loss": 0.8401,
+      "step": 6390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1185380789605117e-05,
+      "loss": 1.2275,
+      "step": 6391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1182930821914702e-05,
+      "loss": 0.8709,
+      "step": 6392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.118048065550499e-05,
+      "loss": 0.2222,
+      "step": 6393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.117803029045474e-05,
+      "loss": 1.7764,
+      "step": 6394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.11755797268427e-05,
+      "loss": 0.8489,
+      "step": 6395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1173128964747625e-05,
+      "loss": 1.4155,
+      "step": 6396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.117067800424829e-05,
+      "loss": 0.8341,
+      "step": 6397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1168226845423466e-05,
+      "loss": 0.8605,
+      "step": 6398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1165775488351934e-05,
+      "loss": 0.2706,
+      "step": 6399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1163323933112485e-05,
+      "loss": 0.5402,
+      "step": 6400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1160872179783905e-05,
+      "loss": 0.0333,
+      "step": 6401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1158420228444998e-05,
+      "loss": 0.0386,
+      "step": 6402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1155968079174566e-05,
+      "loss": 0.059,
+      "step": 6403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.115351573205142e-05,
+      "loss": 0.0381,
+      "step": 6404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1151063187154383e-05,
+      "loss": 0.5284,
+      "step": 6405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.114861044456228e-05,
+      "loss": 1.4664,
+      "step": 6406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.114615750435394e-05,
+      "loss": 1.475,
+      "step": 6407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1143704366608196e-05,
+      "loss": 0.0265,
+      "step": 6408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1141251031403902e-05,
+      "loss": 0.5349,
+      "step": 6409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1138797498819898e-05,
+      "loss": 1.7258,
+      "step": 6410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.113634376893505e-05,
+      "loss": 0.0532,
+      "step": 6411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.113388984182821e-05,
+      "loss": 0.6195,
+      "step": 6412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1131435717578257e-05,
+      "loss": 1.301,
+      "step": 6413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1128981396264065e-05,
+      "loss": 0.8165,
+      "step": 6414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1126526877964506e-05,
+      "loss": 0.2195,
+      "step": 6415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1124072162758475e-05,
+      "loss": 1.1329,
+      "step": 6416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1121617250724874e-05,
+      "loss": 0.1349,
+      "step": 6417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1119162141942592e-05,
+      "loss": 0.1952,
+      "step": 6418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1116706836490545e-05,
+      "loss": 1.6967,
+      "step": 6419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1114251334447637e-05,
+      "loss": 0.2104,
+      "step": 6420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1111795635892797e-05,
+      "loss": 0.98,
+      "step": 6421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1109339740904945e-05,
+      "loss": 0.0445,
+      "step": 6422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.110688364956302e-05,
+      "loss": 0.0862,
+      "step": 6423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.110442736194595e-05,
+      "loss": 0.6872,
+      "step": 6424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1101970878132694e-05,
+      "loss": 0.4075,
+      "step": 6425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.109951419820219e-05,
+      "loss": 0.4389,
+      "step": 6426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1097057322233404e-05,
+      "loss": 1.3682,
+      "step": 6427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.10946002503053e-05,
+      "loss": 0.8041,
+      "step": 6428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1092142982496845e-05,
+      "loss": 0.9973,
+      "step": 6429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.108968551888701e-05,
+      "loss": 0.4532,
+      "step": 6430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1087227859554792e-05,
+      "loss": 0.2914,
+      "step": 6431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1084770004579168e-05,
+      "loss": 1.0727,
+      "step": 6432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1082311954039138e-05,
+      "loss": 1.0068,
+      "step": 6433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.10798537080137e-05,
+      "loss": 0.3926,
+      "step": 6434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1077395266581874e-05,
+      "loss": 0.756,
+      "step": 6435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.107493662982266e-05,
+      "loss": 0.9083,
+      "step": 6436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1072477797815087e-05,
+      "loss": 0.8712,
+      "step": 6437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1070018770638174e-05,
+      "loss": 0.7223,
+      "step": 6438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1067559548370962e-05,
+      "loss": 0.2306,
+      "step": 6439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.106510013109249e-05,
+      "loss": 0.6672,
+      "step": 6440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.10626405188818e-05,
+      "loss": 0.5824,
+      "step": 6441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.106018071181794e-05,
+      "loss": 0.93,
+      "step": 6442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.105772070997998e-05,
+      "loss": 0.5726,
+      "step": 6443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1055260513446973e-05,
+      "loss": 0.4747,
+      "step": 6444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1052800122297998e-05,
+      "loss": 0.8754,
+      "step": 6445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1050339536612127e-05,
+      "loss": 0.2151,
+      "step": 6446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.104787875646844e-05,
+      "loss": 1.0521,
+      "step": 6447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.104541778194603e-05,
+      "loss": 0.5296,
+      "step": 6448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1042956613124004e-05,
+      "loss": 0.2192,
+      "step": 6449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1040495250081446e-05,
+      "loss": 0.4415,
+      "step": 6450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1038033692897475e-05,
+      "loss": 0.027,
+      "step": 6451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1035571941651196e-05,
+      "loss": 1.0374,
+      "step": 6452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.103310999642174e-05,
+      "loss": 0.8319,
+      "step": 6453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1030647857288226e-05,
+      "loss": 0.4237,
+      "step": 6454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1028185524329795e-05,
+      "loss": 0.3591,
+      "step": 6455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1025722997625577e-05,
+      "loss": 0.0316,
+      "step": 6456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1023260277254734e-05,
+      "loss": 1.665,
+      "step": 6457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.102079736329639e-05,
+      "loss": 0.2783,
+      "step": 6458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.101833425582973e-05,
+      "loss": 0.4165,
+      "step": 6459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1015870954933904e-05,
+      "loss": 1.174,
+      "step": 6460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1013407460688088e-05,
+      "loss": 1.7126,
+      "step": 6461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1010943773171454e-05,
+      "loss": 0.2783,
+      "step": 6462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.100847989246319e-05,
+      "loss": 1.1836,
+      "step": 6463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1006015818642483e-05,
+      "loss": 0.1927,
+      "step": 6464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1003551551788523e-05,
+      "loss": 0.5567,
+      "step": 6465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.100108709198052e-05,
+      "loss": 0.9927,
+      "step": 6466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0998622439297676e-05,
+      "loss": 0.1793,
+      "step": 6467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.099615759381921e-05,
+      "loss": 0.5501,
+      "step": 6468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0993692555624343e-05,
+      "loss": 0.293,
+      "step": 6469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0991227324792287e-05,
+      "loss": 0.6893,
+      "step": 6470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0988761901402296e-05,
+      "loss": 1.4919,
+      "step": 6471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.098629628553359e-05,
+      "loss": 0.3461,
+      "step": 6472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0983830477265423e-05,
+      "loss": 0.8836,
+      "step": 6473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0981364476677044e-05,
+      "loss": 0.9674,
+      "step": 6474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.097889828384771e-05,
+      "loss": 1.4394,
+      "step": 6475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0976431898856686e-05,
+      "loss": 0.7913,
+      "step": 6476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.097396532178324e-05,
+      "loss": 0.5601,
+      "step": 6477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0971498552706652e-05,
+      "loss": 0.3665,
+      "step": 6478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0969031591706195e-05,
+      "loss": 0.5237,
+      "step": 6479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0966564438861165e-05,
+      "loss": 0.4687,
+      "step": 6480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0964097094250853e-05,
+      "loss": 1.1513,
+      "step": 6481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.096162955795456e-05,
+      "loss": 0.797,
+      "step": 6482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0959161830051598e-05,
+      "loss": 0.9531,
+      "step": 6483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.095669391062127e-05,
+      "loss": 0.7604,
+      "step": 6484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0954225799742893e-05,
+      "loss": 0.4818,
+      "step": 6485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0951757497495807e-05,
+      "loss": 1.0969,
+      "step": 6486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.094928900395933e-05,
+      "loss": 0.393,
+      "step": 6487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.09468203192128e-05,
+      "loss": 0.5011,
+      "step": 6488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0944351443335572e-05,
+      "loss": 0.8544,
+      "step": 6489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0941882376406986e-05,
+      "loss": 1.247,
+      "step": 6490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0939413118506405e-05,
+      "loss": 0.5453,
+      "step": 6491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0936943669713176e-05,
+      "loss": 0.5794,
+      "step": 6492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0934474030106678e-05,
+      "loss": 0.6058,
+      "step": 6493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.093200419976629e-05,
+      "loss": 0.484,
+      "step": 6494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.092953417877138e-05,
+      "loss": 0.7511,
+      "step": 6495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0927063967201342e-05,
+      "loss": 0.6148,
+      "step": 6496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0924593565135568e-05,
+      "loss": 0.533,
+      "step": 6497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0922122972653454e-05,
+      "loss": 0.7399,
+      "step": 6498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0919652189834406e-05,
+      "loss": 0.1782,
+      "step": 6499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.091718121675784e-05,
+      "loss": 0.2218,
+      "step": 6500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0914710053503164e-05,
+      "loss": 1.6453,
+      "step": 6501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0912238700149808e-05,
+      "loss": 0.0902,
+      "step": 6502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.09097671567772e-05,
+      "loss": 0.3753,
+      "step": 6503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.090729542346477e-05,
+      "loss": 0.0671,
+      "step": 6504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.090482350029197e-05,
+      "loss": 0.6176,
+      "step": 6505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0902351387338237e-05,
+      "loss": 0.6203,
+      "step": 6506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0899879084683032e-05,
+      "loss": 0.6725,
+      "step": 6507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0897406592405808e-05,
+      "loss": 0.2997,
+      "step": 6508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0894933910586035e-05,
+      "loss": 0.0309,
+      "step": 6509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0892461039303186e-05,
+      "loss": 0.2903,
+      "step": 6510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.088998797863674e-05,
+      "loss": 0.2753,
+      "step": 6511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0887514728666176e-05,
+      "loss": 1.3043,
+      "step": 6512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0885041289470985e-05,
+      "loss": 1.3555,
+      "step": 6513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.088256766113067e-05,
+      "loss": 0.4229,
+      "step": 6514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0880093843724726e-05,
+      "loss": 0.0808,
+      "step": 6515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0877619837332658e-05,
+      "loss": 0.3216,
+      "step": 6516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0875145642034e-05,
+      "loss": 0.0423,
+      "step": 6517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0872671257908253e-05,
+      "loss": 0.5966,
+      "step": 6518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0870196685034946e-05,
+      "loss": 1.8068,
+      "step": 6519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.086772192349362e-05,
+      "loss": 0.2194,
+      "step": 6520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.086524697336381e-05,
+      "loss": 0.3187,
+      "step": 6521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0862771834725054e-05,
+      "loss": 1.754,
+      "step": 6522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0860296507656915e-05,
+      "loss": 0.9019,
+      "step": 6523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.085782099223894e-05,
+      "loss": 0.4152,
+      "step": 6524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.08553452885507e-05,
+      "loss": 0.9369,
+      "step": 6525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.085286939667176e-05,
+      "loss": 0.1823,
+      "step": 6526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0850393316681692e-05,
+      "loss": 0.7629,
+      "step": 6527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0847917048660085e-05,
+      "loss": 0.8461,
+      "step": 6528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.084544059268652e-05,
+      "loss": 0.7189,
+      "step": 6529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0842963948840587e-05,
+      "loss": 0.6153,
+      "step": 6530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0840487117201898e-05,
+      "loss": 0.2364,
+      "step": 6531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0838010097850046e-05,
+      "loss": 1.079,
+      "step": 6532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.083553289086465e-05,
+      "loss": 0.7598,
+      "step": 6533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0833055496325324e-05,
+      "loss": 0.6786,
+      "step": 6534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0830577914311687e-05,
+      "loss": 0.3773,
+      "step": 6535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0828100144903374e-05,
+      "loss": 0.3873,
+      "step": 6536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.082562218818003e-05,
+      "loss": 1.7903,
+      "step": 6537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0823144044221277e-05,
+      "loss": 0.6817,
+      "step": 6538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0820665713106772e-05,
+      "loss": 0.1643,
+      "step": 6539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0818187194916167e-05,
+      "loss": 1.5405,
+      "step": 6540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0815708489729127e-05,
+      "loss": 0.1947,
+      "step": 6541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.081322959762531e-05,
+      "loss": 0.8983,
+      "step": 6542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0810750518684393e-05,
+      "loss": 1.0968,
+      "step": 6543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.080827125298605e-05,
+      "loss": 1.1577,
+      "step": 6544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0805791800609967e-05,
+      "loss": 0.2544,
+      "step": 6545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0803312161635835e-05,
+      "loss": 0.8988,
+      "step": 6546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0800832336143338e-05,
+      "loss": 0.6204,
+      "step": 6547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0798352324212197e-05,
+      "loss": 0.2047,
+      "step": 6548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0795872125922105e-05,
+      "loss": 0.3349,
+      "step": 6549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.079339174135278e-05,
+      "loss": 0.2837,
+      "step": 6550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.079091117058394e-05,
+      "loss": 0.9035,
+      "step": 6551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0788430413695314e-05,
+      "loss": 0.5224,
+      "step": 6552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.078594947076663e-05,
+      "loss": 1.4125,
+      "step": 6553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0783468341877623e-05,
+      "loss": 0.0782,
+      "step": 6554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0780987027108044e-05,
+      "loss": 0.9583,
+      "step": 6555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0778505526537637e-05,
+      "loss": 0.1007,
+      "step": 6556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0776023840246165e-05,
+      "loss": 1.3266,
+      "step": 6557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0773541968313373e-05,
+      "loss": 0.1765,
+      "step": 6558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0771059910819044e-05,
+      "loss": 1.17,
+      "step": 6559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0768577667842948e-05,
+      "loss": 0.484,
+      "step": 6560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0766095239464858e-05,
+      "loss": 0.0327,
+      "step": 6561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0763612625764565e-05,
+      "loss": 0.4316,
+      "step": 6562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.076112982682186e-05,
+      "loss": 0.2769,
+      "step": 6563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0758646842716536e-05,
+      "loss": 0.1211,
+      "step": 6564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.07561636735284e-05,
+      "loss": 0.3252,
+      "step": 6565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.075368031933726e-05,
+      "loss": 0.1622,
+      "step": 6566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.075119678022293e-05,
+      "loss": 1.4173,
+      "step": 6567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0748713056265233e-05,
+      "loss": 0.1917,
+      "step": 6568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0746229147543992e-05,
+      "loss": 0.418,
+      "step": 6569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0743745054139038e-05,
+      "loss": 0.9498,
+      "step": 6570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0741260776130223e-05,
+      "loss": 0.0983,
+      "step": 6571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.073877631359738e-05,
+      "loss": 0.2069,
+      "step": 6572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.073629166662036e-05,
+      "loss": 0.7406,
+      "step": 6573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0733806835279024e-05,
+      "loss": 1.8779,
+      "step": 6574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0731321819653227e-05,
+      "loss": 0.9669,
+      "step": 6575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.072883661982285e-05,
+      "loss": 0.172,
+      "step": 6576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0726351235867754e-05,
+      "loss": 1.3086,
+      "step": 6577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0723865667867824e-05,
+      "loss": 1.2779,
+      "step": 6578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.072137991590295e-05,
+      "loss": 0.6911,
+      "step": 6579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.071889398005302e-05,
+      "loss": 0.4897,
+      "step": 6580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.071640786039793e-05,
+      "loss": 0.7643,
+      "step": 6581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0713921557017596e-05,
+      "loss": 0.6262,
+      "step": 6582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0711435069991913e-05,
+      "loss": 0.3416,
+      "step": 6583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0708948399400804e-05,
+      "loss": 0.608,
+      "step": 6584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0706461545324185e-05,
+      "loss": 0.753,
+      "step": 6585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.070397450784199e-05,
+      "loss": 1.2029,
+      "step": 6586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.070148728703415e-05,
+      "loss": 0.9931,
+      "step": 6587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0698999882980602e-05,
+      "loss": 0.3557,
+      "step": 6588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0696512295761296e-05,
+      "loss": 1.0592,
+      "step": 6589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0694024525456175e-05,
+      "loss": 0.5432,
+      "step": 6590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0691536572145208e-05,
+      "loss": 0.3086,
+      "step": 6591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0689048435908345e-05,
+      "loss": 0.3952,
+      "step": 6592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.068656011682556e-05,
+      "loss": 0.9888,
+      "step": 6593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.068407161497683e-05,
+      "loss": 0.4662,
+      "step": 6594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.068158293044213e-05,
+      "loss": 0.7242,
+      "step": 6595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0679094063301456e-05,
+      "loss": 0.2192,
+      "step": 6596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.067660501363479e-05,
+      "loss": 0.7448,
+      "step": 6597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0674115781522132e-05,
+      "loss": 0.42,
+      "step": 6598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0671626367043488e-05,
+      "loss": 0.3159,
+      "step": 6599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.066913677027887e-05,
+      "loss": 0.4526,
+      "step": 6600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0666646991308288e-05,
+      "loss": 1.265,
+      "step": 6601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.066415703021177e-05,
+      "loss": 0.0497,
+      "step": 6602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0661666887069338e-05,
+      "loss": 0.0368,
+      "step": 6603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0659176561961022e-05,
+      "loss": 0.0495,
+      "step": 6604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0656686054966875e-05,
+      "loss": 0.8974,
+      "step": 6605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0654195366166925e-05,
+      "loss": 0.8173,
+      "step": 6606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0651704495641236e-05,
+      "loss": 1.8707,
+      "step": 6607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0649213443469854e-05,
+      "loss": 0.0346,
+      "step": 6608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0646722209732846e-05,
+      "loss": 0.2295,
+      "step": 6609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.064423079451028e-05,
+      "loss": 0.6175,
+      "step": 6610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.064173919788224e-05,
+      "loss": 0.2581,
+      "step": 6611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0639247419928783e-05,
+      "loss": 0.7837,
+      "step": 6612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0636755460730015e-05,
+      "loss": 1.409,
+      "step": 6613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0634263320366023e-05,
+      "loss": 0.3316,
+      "step": 6614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.06317709989169e-05,
+      "loss": 1.0017,
+      "step": 6615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.062927849646275e-05,
+      "loss": 0.0469,
+      "step": 6616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0626785813083683e-05,
+      "loss": 0.1149,
+      "step": 6617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.062429294885982e-05,
+      "loss": 0.5221,
+      "step": 6618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0621799903871267e-05,
+      "loss": 0.4239,
+      "step": 6619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.061930667819816e-05,
+      "loss": 0.5691,
+      "step": 6620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0616813271920633e-05,
+      "loss": 0.0578,
+      "step": 6621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.061431968511882e-05,
+      "loss": 1.5148,
+      "step": 6622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.061182591787287e-05,
+      "loss": 0.4914,
+      "step": 6623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.060933197026292e-05,
+      "loss": 0.2297,
+      "step": 6624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0606837842369145e-05,
+      "loss": 1.1052,
+      "step": 6625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0604343534271693e-05,
+      "loss": 0.8046,
+      "step": 6626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0601849046050726e-05,
+      "loss": 0.927,
+      "step": 6627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.059935437778643e-05,
+      "loss": 0.4489,
+      "step": 6628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.059685952955898e-05,
+      "loss": 0.8954,
+      "step": 6629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0594364501448552e-05,
+      "loss": 2.0491,
+      "step": 6630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0591869293535345e-05,
+      "loss": 0.117,
+      "step": 6631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0589373905899554e-05,
+      "loss": 0.634,
+      "step": 6632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0586878338621374e-05,
+      "loss": 0.4004,
+      "step": 6633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0584382591781025e-05,
+      "loss": 0.6117,
+      "step": 6634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0581886665458708e-05,
+      "loss": 0.6057,
+      "step": 6635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0579390559734646e-05,
+      "loss": 0.4332,
+      "step": 6636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0576894274689068e-05,
+      "loss": 0.134,
+      "step": 6637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.05743978104022e-05,
+      "loss": 0.8401,
+      "step": 6638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0571901166954278e-05,
+      "loss": 0.0978,
+      "step": 6639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0569404344425546e-05,
+      "loss": 1.1136,
+      "step": 6640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.056690734289625e-05,
+      "loss": 0.2419,
+      "step": 6641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.056441016244665e-05,
+      "loss": 1.3433,
+      "step": 6642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0561912803156996e-05,
+      "loss": 1.0112,
+      "step": 6643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0559415265107556e-05,
+      "loss": 0.0901,
+      "step": 6644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0556917548378606e-05,
+      "loss": 0.4214,
+      "step": 6645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0554419653050418e-05,
+      "loss": 0.9352,
+      "step": 6646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0551921579203274e-05,
+      "loss": 0.8315,
+      "step": 6647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0549423326917465e-05,
+      "loss": 0.621,
+      "step": 6648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0546924896273284e-05,
+      "loss": 0.4242,
+      "step": 6649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.054442628735103e-05,
+      "loss": 0.0871,
+      "step": 6650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0541927500231002e-05,
+      "loss": 1.1464,
+      "step": 6651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0539428534993516e-05,
+      "loss": 0.0417,
+      "step": 6652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0536929391718897e-05,
+      "loss": 0.5489,
+      "step": 6653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0534430070487453e-05,
+      "loss": 0.0292,
+      "step": 6654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0531930571379523e-05,
+      "loss": 0.4061,
+      "step": 6655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0529430894475434e-05,
+      "loss": 0.4843,
+      "step": 6656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0526931039855536e-05,
+      "loss": 0.034,
+      "step": 6657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0524431007600157e-05,
+      "loss": 1.5825,
+      "step": 6658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0521930797789663e-05,
+      "loss": 0.1902,
+      "step": 6659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0519430410504402e-05,
+      "loss": 0.2091,
+      "step": 6660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0516929845824742e-05,
+      "loss": 1.2642,
+      "step": 6661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.051442910383105e-05,
+      "loss": 0.3747,
+      "step": 6662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0511928184603698e-05,
+      "loss": 0.3583,
+      "step": 6663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0509427088223067e-05,
+      "loss": 2.0026,
+      "step": 6664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0506925814769537e-05,
+      "loss": 0.6234,
+      "step": 6665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.050442436432351e-05,
+      "loss": 0.0683,
+      "step": 6666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.050192273696537e-05,
+      "loss": 0.6284,
+      "step": 6667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.049942093277553e-05,
+      "loss": 0.0831,
+      "step": 6668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.049691895183439e-05,
+      "loss": 0.1097,
+      "step": 6669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0494416794222363e-05,
+      "loss": 0.1641,
+      "step": 6670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0491914460019878e-05,
+      "loss": 0.3927,
+      "step": 6671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0489411949307355e-05,
+      "loss": 0.7011,
+      "step": 6672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0486909262165218e-05,
+      "loss": 0.4028,
+      "step": 6673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0484406398673912e-05,
+      "loss": 1.4511,
+      "step": 6674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0481903358913877e-05,
+      "loss": 1.151,
+      "step": 6675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0479400142965557e-05,
+      "loss": 0.2297,
+      "step": 6676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0476896750909408e-05,
+      "loss": 0.9056,
+      "step": 6677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.047439318282589e-05,
+      "loss": 0.5794,
+      "step": 6678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0471889438795474e-05,
+      "loss": 2.0297,
+      "step": 6679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.046938551889861e-05,
+      "loss": 0.4001,
+      "step": 6680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0466881423215794e-05,
+      "loss": 0.4263,
+      "step": 6681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.04643771518275e-05,
+      "loss": 0.3184,
+      "step": 6682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0461872704814213e-05,
+      "loss": 0.2987,
+      "step": 6683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.045936808225643e-05,
+      "loss": 0.2276,
+      "step": 6684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0456863284234648e-05,
+      "loss": 0.6411,
+      "step": 6685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0454358310829372e-05,
+      "loss": 1.1222,
+      "step": 6686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.045185316212111e-05,
+      "loss": 1.4635,
+      "step": 6687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.044934783819038e-05,
+      "loss": 0.5333,
+      "step": 6688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0446842339117698e-05,
+      "loss": 0.9942,
+      "step": 6689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0444336664983593e-05,
+      "loss": 0.2605,
+      "step": 6690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0441830815868606e-05,
+      "loss": 0.7695,
+      "step": 6691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0439324791853262e-05,
+      "loss": 1.3679,
+      "step": 6692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.043681859301811e-05,
+      "loss": 0.7345,
+      "step": 6693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0434312219443704e-05,
+      "loss": 0.2481,
+      "step": 6694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0431805671210588e-05,
+      "loss": 0.6355,
+      "step": 6695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.042929894839933e-05,
+      "loss": 0.7382,
+      "step": 6696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0426792051090497e-05,
+      "loss": 0.6298,
+      "step": 6697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0424284979364657e-05,
+      "loss": 0.452,
+      "step": 6698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0421777733302388e-05,
+      "loss": 0.3695,
+      "step": 6699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0419270312984272e-05,
+      "loss": 0.2709,
+      "step": 6700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.04167627184909e-05,
+      "loss": 0.5415,
+      "step": 6701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0414254949902865e-05,
+      "loss": 0.0403,
+      "step": 6702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.041174700730077e-05,
+      "loss": 0.844,
+      "step": 6703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0409238890765206e-05,
+      "loss": 0.0785,
+      "step": 6704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0406730600376805e-05,
+      "loss": 0.921,
+      "step": 6705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0404222136216166e-05,
+      "loss": 0.3997,
+      "step": 6706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0401713498363924e-05,
+      "loss": 0.6187,
+      "step": 6707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0399204686900695e-05,
+      "loss": 0.1287,
+      "step": 6708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.039669570190712e-05,
+      "loss": 1.2169,
+      "step": 6709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0394186543463834e-05,
+      "loss": 1.5737,
+      "step": 6710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0391677211651487e-05,
+      "loss": 0.0913,
+      "step": 6711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0389167706550718e-05,
+      "loss": 0.1948,
+      "step": 6712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0386658028242194e-05,
+      "loss": 0.2701,
+      "step": 6713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.038414817680657e-05,
+      "loss": 0.0359,
+      "step": 6714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0381638152324512e-05,
+      "loss": 0.0767,
+      "step": 6715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.037912795487669e-05,
+      "loss": 0.9348,
+      "step": 6716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.037661758454379e-05,
+      "loss": 0.292,
+      "step": 6717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.037410704140649e-05,
+      "loss": 0.3257,
+      "step": 6718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0371596325545477e-05,
+      "loss": 1.5042,
+      "step": 6719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.036908543704145e-05,
+      "loss": 0.5077,
+      "step": 6720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0366574375975108e-05,
+      "loss": 0.1125,
+      "step": 6721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0364063142427156e-05,
+      "loss": 0.8002,
+      "step": 6722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0361551736478303e-05,
+      "loss": 0.2006,
+      "step": 6723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0359040158209263e-05,
+      "loss": 0.7055,
+      "step": 6724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.035652840770077e-05,
+      "loss": 0.4806,
+      "step": 6725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0354016485033535e-05,
+      "loss": 1.2568,
+      "step": 6726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0351504390288304e-05,
+      "loss": 1.6195,
+      "step": 6727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.034899212354581e-05,
+      "loss": 1.1928,
+      "step": 6728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.03464796848868e-05,
+      "loss": 0.842,
+      "step": 6729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0343967074392017e-05,
+      "loss": 0.8116,
+      "step": 6730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0341454292142225e-05,
+      "loss": 0.5698,
+      "step": 6731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0338941338218184e-05,
+      "loss": 0.3776,
+      "step": 6732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0336428212700653e-05,
+      "loss": 0.5166,
+      "step": 6733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0333914915670416e-05,
+      "loss": 0.9916,
+      "step": 6734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0331401447208235e-05,
+      "loss": 1.2108,
+      "step": 6735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0328887807394904e-05,
+      "loss": 0.6641,
+      "step": 6736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0326373996311207e-05,
+      "loss": 0.3932,
+      "step": 6737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.032386001403794e-05,
+      "loss": 0.5925,
+      "step": 6738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0321345860655895e-05,
+      "loss": 0.6106,
+      "step": 6739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0318831536245887e-05,
+      "loss": 0.922,
+      "step": 6740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0316317040888724e-05,
+      "loss": 0.6907,
+      "step": 6741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0313802374665218e-05,
+      "loss": 0.5904,
+      "step": 6742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0311287537656194e-05,
+      "loss": 0.4111,
+      "step": 6743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.030877252994248e-05,
+      "loss": 0.8677,
+      "step": 6744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0306257351604903e-05,
+      "loss": 1.2014,
+      "step": 6745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0303742002724304e-05,
+      "loss": 0.6507,
+      "step": 6746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.030122648338152e-05,
+      "loss": 0.1684,
+      "step": 6747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.029871079365741e-05,
+      "loss": 0.3897,
+      "step": 6748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.029619493363282e-05,
+      "loss": 0.4826,
+      "step": 6749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.029367890338862e-05,
+      "loss": 0.2731,
+      "step": 6750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0291162703005666e-05,
+      "loss": 0.0834,
+      "step": 6751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0288646332564827e-05,
+      "loss": 0.4686,
+      "step": 6752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.028612979214699e-05,
+      "loss": 1.3858,
+      "step": 6753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.028361308183302e-05,
+      "loss": 0.132,
+      "step": 6754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.028109620170382e-05,
+      "loss": 0.881,
+      "step": 6755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.027857915184028e-05,
+      "loss": 0.0353,
+      "step": 6756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.027606193232329e-05,
+      "loss": 0.0486,
+      "step": 6757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0273544543233758e-05,
+      "loss": 1.6838,
+      "step": 6758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0271026984652593e-05,
+      "loss": 0.3167,
+      "step": 6759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.026850925666071e-05,
+      "loss": 1.4972,
+      "step": 6760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0265991359339024e-05,
+      "loss": 0.4762,
+      "step": 6761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0263473292768464e-05,
+      "loss": 0.0351,
+      "step": 6762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.026095505702996e-05,
+      "loss": 0.0283,
+      "step": 6763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.025843665220445e-05,
+      "loss": 0.617,
+      "step": 6764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.025591807837288e-05,
+      "loss": 0.8802,
+      "step": 6765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.025339933561618e-05,
+      "loss": 0.1469,
+      "step": 6766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0250880424015313e-05,
+      "loss": 0.7677,
+      "step": 6767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.024836134365125e-05,
+      "loss": 0.399,
+      "step": 6768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.024584209460493e-05,
+      "loss": 0.1585,
+      "step": 6769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0243322676957337e-05,
+      "loss": 0.2283,
+      "step": 6770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0240803090789436e-05,
+      "loss": 0.2041,
+      "step": 6771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0238283336182212e-05,
+      "loss": 0.1592,
+      "step": 6772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0235763413216654e-05,
+      "loss": 0.6165,
+      "step": 6773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0233243321973743e-05,
+      "loss": 1.4215,
+      "step": 6774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.023072306253448e-05,
+      "loss": 1.3472,
+      "step": 6775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0228202634979868e-05,
+      "loss": 1.9906,
+      "step": 6776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.022568203939091e-05,
+      "loss": 0.4675,
+      "step": 6777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0223161275848615e-05,
+      "loss": 0.4336,
+      "step": 6778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.022064034443401e-05,
+      "loss": 0.1612,
+      "step": 6779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0218119245228104e-05,
+      "loss": 1.2712,
+      "step": 6780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0215597978311933e-05,
+      "loss": 0.2599,
+      "step": 6781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.021307654376654e-05,
+      "loss": 0.6618,
+      "step": 6782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0210554941672945e-05,
+      "loss": 0.9607,
+      "step": 6783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0208033172112203e-05,
+      "loss": 0.7802,
+      "step": 6784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.020551123516536e-05,
+      "loss": 1.3263,
+      "step": 6785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0202989130913475e-05,
+      "loss": 0.3092,
+      "step": 6786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0200466859437607e-05,
+      "loss": 0.9222,
+      "step": 6787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0197944420818825e-05,
+      "loss": 1.6723,
+      "step": 6788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0195421815138186e-05,
+      "loss": 0.1543,
+      "step": 6789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0192899042476783e-05,
+      "loss": 0.3552,
+      "step": 6790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0190376102915692e-05,
+      "loss": 0.6395,
+      "step": 6791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0187852996535996e-05,
+      "loss": 0.6746,
+      "step": 6792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0185329723418793e-05,
+      "loss": 1.5666,
+      "step": 6793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0182806283645183e-05,
+      "loss": 0.566,
+      "step": 6794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.018028267729626e-05,
+      "loss": 1.4839,
+      "step": 6795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.017775890445314e-05,
+      "loss": 1.2393,
+      "step": 6796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0175234965196936e-05,
+      "loss": 0.2302,
+      "step": 6797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0172710859608766e-05,
+      "loss": 1.3131,
+      "step": 6798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.017018658776976e-05,
+      "loss": 0.4106,
+      "step": 6799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0167662149761035e-05,
+      "loss": 0.4562,
+      "step": 6800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.016513754566374e-05,
+      "loss": 0.3633,
+      "step": 6801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.016261277555901e-05,
+      "loss": 0.2926,
+      "step": 6802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0160087839527996e-05,
+      "loss": 0.32,
+      "step": 6803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.015756273765184e-05,
+      "loss": 0.0348,
+      "step": 6804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.01550374700117e-05,
+      "loss": 1.3111,
+      "step": 6805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0152512036688748e-05,
+      "loss": 0.4033,
+      "step": 6806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0149986437764144e-05,
+      "loss": 0.2974,
+      "step": 6807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.014746067331906e-05,
+      "loss": 0.1517,
+      "step": 6808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.014493474343468e-05,
+      "loss": 0.1498,
+      "step": 6809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.014240864819218e-05,
+      "loss": 0.5287,
+      "step": 6810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0139882387672754e-05,
+      "loss": 1.021,
+      "step": 6811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0137355961957586e-05,
+      "loss": 0.1432,
+      "step": 6812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0134829371127898e-05,
+      "loss": 1.1289,
+      "step": 6813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0132302615264867e-05,
+      "loss": 0.1802,
+      "step": 6814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0129775694449723e-05,
+      "loss": 0.7252,
+      "step": 6815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0127248608763667e-05,
+      "loss": 1.1954,
+      "step": 6816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.012472135828793e-05,
+      "loss": 0.6036,
+      "step": 6817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0122193943103735e-05,
+      "loss": 1.4779,
+      "step": 6818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0119666363292313e-05,
+      "loss": 0.1924,
+      "step": 6819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0117138618934898e-05,
+      "loss": 0.3076,
+      "step": 6820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.011461071011273e-05,
+      "loss": 0.7458,
+      "step": 6821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.011208263690707e-05,
+      "loss": 0.2264,
+      "step": 6822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.010955439939915e-05,
+      "loss": 0.5914,
+      "step": 6823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0107025997670236e-05,
+      "loss": 1.4275,
+      "step": 6824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0104497431801596e-05,
+      "loss": 0.7197,
+      "step": 6825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0101968701874493e-05,
+      "loss": 0.2593,
+      "step": 6826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0099439807970203e-05,
+      "loss": 0.6608,
+      "step": 6827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0096910750169998e-05,
+      "loss": 0.6946,
+      "step": 6828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.009438152855517e-05,
+      "loss": 0.8007,
+      "step": 6829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0091852143207005e-05,
+      "loss": 0.6493,
+      "step": 6830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0089322594206793e-05,
+      "loss": 0.2402,
+      "step": 6831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.008679288163584e-05,
+      "loss": 0.8833,
+      "step": 6832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0084263005575454e-05,
+      "loss": 0.7644,
+      "step": 6833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0081732966106937e-05,
+      "loss": 1.2102,
+      "step": 6834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0079202763311602e-05,
+      "loss": 0.6694,
+      "step": 6835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.007667239727078e-05,
+      "loss": 0.3428,
+      "step": 6836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.007414186806579e-05,
+      "loss": 0.6824,
+      "step": 6837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0071611175777967e-05,
+      "loss": 0.7765,
+      "step": 6838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0069080320488644e-05,
+      "loss": 0.7387,
+      "step": 6839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0066549302279163e-05,
+      "loss": 1.5338,
+      "step": 6840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0064018121230875e-05,
+      "loss": 0.6666,
+      "step": 6841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.006148677742513e-05,
+      "loss": 0.5506,
+      "step": 6842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.005895527094328e-05,
+      "loss": 0.8818,
+      "step": 6843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0056423601866694e-05,
+      "loss": 0.312,
+      "step": 6844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.005389177027674e-05,
+      "loss": 0.8853,
+      "step": 6845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.005135977625479e-05,
+      "loss": 1.0295,
+      "step": 6846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.004882761988222e-05,
+      "loss": 0.7933,
+      "step": 6847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.004629530124041e-05,
+      "loss": 0.2471,
+      "step": 6848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0043762820410755e-05,
+      "loss": 0.4003,
+      "step": 6849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.004123017747465e-05,
+      "loss": 0.4764,
+      "step": 6850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.003869737251349e-05,
+      "loss": 1.1837,
+      "step": 6851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.003616440560868e-05,
+      "loss": 0.0465,
+      "step": 6852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.003363127684163e-05,
+      "loss": 1.4224,
+      "step": 6853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0031097986293753e-05,
+      "loss": 0.2947,
+      "step": 6854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0028564534046475e-05,
+      "loss": 1.4627,
+      "step": 6855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0026030920181218e-05,
+      "loss": 1.339,
+      "step": 6856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.002349714477941e-05,
+      "loss": 1.6804,
+      "step": 6857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.002096320792248e-05,
+      "loss": 0.1263,
+      "step": 6858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0018429109691885e-05,
+      "loss": 1.1597,
+      "step": 6859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0015894850169063e-05,
+      "loss": 0.153,
+      "step": 6860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.001336042943546e-05,
+      "loss": 0.5536,
+      "step": 6861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0010825847572534e-05,
+      "loss": 0.0699,
+      "step": 6862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0008291104661754e-05,
+      "loss": 0.3984,
+      "step": 6863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.000575620078458e-05,
+      "loss": 0.7527,
+      "step": 6864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.000322113602249e-05,
+      "loss": 0.247,
+      "step": 6865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0000685910456943e-05,
+      "loss": 0.186,
+      "step": 6866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9998150524169447e-05,
+      "loss": 0.3135,
+      "step": 6867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9995614977241475e-05,
+      "loss": 0.9855,
+      "step": 6868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9993079269754518e-05,
+      "loss": 0.1655,
+      "step": 6869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9990543401790077e-05,
+      "loss": 0.1281,
+      "step": 6870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9988007373429655e-05,
+      "loss": 1.2826,
+      "step": 6871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9985471184754756e-05,
+      "loss": 0.1471,
+      "step": 6872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9982934835846902e-05,
+      "loss": 0.2127,
+      "step": 6873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9980398326787602e-05,
+      "loss": 1.3404,
+      "step": 6874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9977861657658383e-05,
+      "loss": 1.6483,
+      "step": 6875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.997532482854078e-05,
+      "loss": 0.5348,
+      "step": 6876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.997278783951631e-05,
+      "loss": 0.0894,
+      "step": 6877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9970250690666526e-05,
+      "loss": 0.0929,
+      "step": 6878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9967713382072972e-05,
+      "loss": 0.4098,
+      "step": 6879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9965175913817187e-05,
+      "loss": 0.8221,
+      "step": 6880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9962638285980734e-05,
+      "loss": 0.7859,
+      "step": 6881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9960100498645168e-05,
+      "loss": 1.0591,
+      "step": 6882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9957562551892058e-05,
+      "loss": 0.4894,
+      "step": 6883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9955024445802968e-05,
+      "loss": 0.9783,
+      "step": 6884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9952486180459472e-05,
+      "loss": 0.2706,
+      "step": 6885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9949947755943155e-05,
+      "loss": 0.3877,
+      "step": 6886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.99474091723356e-05,
+      "loss": 0.4277,
+      "step": 6887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9944870429718402e-05,
+      "loss": 0.9767,
+      "step": 6888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9942331528173143e-05,
+      "loss": 0.4781,
+      "step": 6889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9939792467781436e-05,
+      "loss": 0.8613,
+      "step": 6890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.993725324862488e-05,
+      "loss": 1.2783,
+      "step": 6891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.993471387078509e-05,
+      "loss": 1.3897,
+      "step": 6892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9932174334343675e-05,
+      "loss": 0.293,
+      "step": 6893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9929634639382257e-05,
+      "loss": 1.0528,
+      "step": 6894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9927094785982466e-05,
+      "loss": 0.5109,
+      "step": 6895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9924554774225935e-05,
+      "loss": 0.4053,
+      "step": 6896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.992201460419429e-05,
+      "loss": 0.1574,
+      "step": 6897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9919474275969182e-05,
+      "loss": 0.3517,
+      "step": 6898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9916933789632258e-05,
+      "loss": 0.3875,
+      "step": 6899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9914393145265152e-05,
+      "loss": 0.2395,
+      "step": 6900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9911852342949543e-05,
+      "loss": 0.0744,
+      "step": 6901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9909311382767077e-05,
+      "loss": 1.1125,
+      "step": 6902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9906770264799425e-05,
+      "loss": 1.6609,
+      "step": 6903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9904228989128255e-05,
+      "loss": 1.5157,
+      "step": 6904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9901687555835252e-05,
+      "loss": 1.2733,
+      "step": 6905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9899145965002093e-05,
+      "loss": 1.3234,
+      "step": 6906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9896604216710462e-05,
+      "loss": 1.3339,
+      "step": 6907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9894062311042054e-05,
+      "loss": 0.0804,
+      "step": 6908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9891520248078565e-05,
+      "loss": 0.9877,
+      "step": 6909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.98889780279017e-05,
+      "loss": 1.459,
+      "step": 6910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.988643565059316e-05,
+      "loss": 0.5565,
+      "step": 6911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9883893116234657e-05,
+      "loss": 0.1119,
+      "step": 6912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.988135042490792e-05,
+      "loss": 1.1448,
+      "step": 6913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9878807576694653e-05,
+      "loss": 0.2605,
+      "step": 6914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9876264571676596e-05,
+      "loss": 0.6809,
+      "step": 6915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9873721409935476e-05,
+      "loss": 0.1769,
+      "step": 6916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.987117809155303e-05,
+      "loss": 0.6044,
+      "step": 6917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9868634616611005e-05,
+      "loss": 1.0918,
+      "step": 6918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9866090985191145e-05,
+      "loss": 0.6184,
+      "step": 6919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9863547197375196e-05,
+      "loss": 1.3147,
+      "step": 6920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9861003253244925e-05,
+      "loss": 1.4523,
+      "step": 6921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9858459152882092e-05,
+      "loss": 0.5671,
+      "step": 6922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.985591489636846e-05,
+      "loss": 0.2624,
+      "step": 6923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.985337048378581e-05,
+      "loss": 1.3403,
+      "step": 6924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.985082591521591e-05,
+      "loss": 0.3905,
+      "step": 6925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9848281190740546e-05,
+      "loss": 0.2507,
+      "step": 6926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9845736310441503e-05,
+      "loss": 1.3034,
+      "step": 6927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9843191274400577e-05,
+      "loss": 0.7112,
+      "step": 6928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9840646082699567e-05,
+      "loss": 0.2435,
+      "step": 6929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983810073542027e-05,
+      "loss": 0.7408,
+      "step": 6930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9835555232644494e-05,
+      "loss": 0.1083,
+      "step": 6931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983300957445406e-05,
+      "loss": 0.8477,
+      "step": 6932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983046376093078e-05,
+      "loss": 0.4062,
+      "step": 6933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9827917792156468e-05,
+      "loss": 0.3616,
+      "step": 6934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.982537166821296e-05,
+      "loss": 0.6855,
+      "step": 6935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9822825389182096e-05,
+      "loss": 0.5757,
+      "step": 6936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9820278955145695e-05,
+      "loss": 0.302,
+      "step": 6937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9817732366185614e-05,
+      "loss": 0.3578,
+      "step": 6938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.981518562238369e-05,
+      "loss": 0.5703,
+      "step": 6939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9812638723821784e-05,
+      "loss": 1.1231,
+      "step": 6940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9810091670581748e-05,
+      "loss": 0.7547,
+      "step": 6941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9807544462745453e-05,
+      "loss": 1.1753,
+      "step": 6942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.980499710039475e-05,
+      "loss": 0.2593,
+      "step": 6943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9802449583611528e-05,
+      "loss": 0.9378,
+      "step": 6944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.979990191247765e-05,
+      "loss": 0.5953,
+      "step": 6945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.979735408707501e-05,
+      "loss": 0.7482,
+      "step": 6946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9794806107485485e-05,
+      "loss": 0.4402,
+      "step": 6947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9792257973790976e-05,
+      "loss": 0.7609,
+      "step": 6948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.978970968607337e-05,
+      "loss": 0.2237,
+      "step": 6949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.978716124441458e-05,
+      "loss": 1.0332,
+      "step": 6950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9784612648896508e-05,
+      "loss": 1.3797,
+      "step": 6951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.978206389960106e-05,
+      "loss": 0.0361,
+      "step": 6952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9779514996610164e-05,
+      "loss": 0.0329,
+      "step": 6953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.977696594000573e-05,
+      "loss": 1.4531,
+      "step": 6954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9774416729869695e-05,
+      "loss": 0.0935,
+      "step": 6955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9771867366283988e-05,
+      "loss": 0.7149,
+      "step": 6956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.976931784933054e-05,
+      "loss": 0.2522,
+      "step": 6957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9766768179091298e-05,
+      "loss": 0.8876,
+      "step": 6958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9764218355648205e-05,
+      "loss": 0.0282,
+      "step": 6959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9761668379083212e-05,
+      "loss": 0.2348,
+      "step": 6960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975911824947828e-05,
+      "loss": 0.3108,
+      "step": 6961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9756567966915363e-05,
+      "loss": 0.2704,
+      "step": 6962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9754017531476434e-05,
+      "loss": 1.8201,
+      "step": 6963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975146694324346e-05,
+      "loss": 0.6555,
+      "step": 6964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974891620229842e-05,
+      "loss": 0.4736,
+      "step": 6965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746365308723288e-05,
+      "loss": 0.0372,
+      "step": 6966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743814262600062e-05,
+      "loss": 0.8739,
+      "step": 6967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974126306401072e-05,
+      "loss": 1.5276,
+      "step": 6968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973871171303726e-05,
+      "loss": 0.5768,
+      "step": 6969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9736160209761685e-05,
+      "loss": 0.3952,
+      "step": 6970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9733608554266003e-05,
+      "loss": 0.5025,
+      "step": 6971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973105674663222e-05,
+      "loss": 0.5625,
+      "step": 6972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9728504786942353e-05,
+      "loss": 0.8555,
+      "step": 6973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.972595267527842e-05,
+      "loss": 0.2373,
+      "step": 6974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9723400411722446e-05,
+      "loss": 0.2125,
+      "step": 6975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9720847996356465e-05,
+      "loss": 1.9007,
+      "step": 6976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.97182954292625e-05,
+      "loss": 0.4026,
+      "step": 6977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.971574271052261e-05,
+      "loss": 0.2382,
+      "step": 6978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9713189840218824e-05,
+      "loss": 0.7126,
+      "step": 6979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9710636818433194e-05,
+      "loss": 0.1545,
+      "step": 6980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9708083645247773e-05,
+      "loss": 0.8661,
+      "step": 6981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9705530320744626e-05,
+      "loss": 0.3901,
+      "step": 6982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.970297684500581e-05,
+      "loss": 1.1123,
+      "step": 6983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9700423218113398e-05,
+      "loss": 0.9419,
+      "step": 6984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9697869440149465e-05,
+      "loss": 1.638,
+      "step": 6985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.969531551119608e-05,
+      "loss": 0.5732,
+      "step": 6986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9692761431335342e-05,
+      "loss": 0.3109,
+      "step": 6987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9690207200649328e-05,
+      "loss": 0.113,
+      "step": 6988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9687652819220126e-05,
+      "loss": 0.0665,
+      "step": 6989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9685098287129844e-05,
+      "loss": 0.7167,
+      "step": 6990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9682543604460585e-05,
+      "loss": 0.1106,
+      "step": 6991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9679988771294448e-05,
+      "loss": 0.2201,
+      "step": 6992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9677433787713555e-05,
+      "loss": 0.8573,
+      "step": 6993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9674878653800012e-05,
+      "loss": 0.3941,
+      "step": 6994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9672323369635953e-05,
+      "loss": 0.4818,
+      "step": 6995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.96697679353035e-05,
+      "loss": 0.6043,
+      "step": 6996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9667212350884777e-05,
+      "loss": 1.5077,
+      "step": 6997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.966465661646193e-05,
+      "loss": 0.5576,
+      "step": 6998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9662100732117098e-05,
+      "loss": 0.8336,
+      "step": 6999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9659544697932422e-05,
+      "loss": 0.3236,
+      "step": 7000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9656988513990064e-05,
+      "loss": 0.7218,
+      "step": 7001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.965443218037217e-05,
+      "loss": 0.195,
+      "step": 7002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.96518756971609e-05,
+      "loss": 0.7448,
+      "step": 7003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9649319064438428e-05,
+      "loss": 0.4645,
+      "step": 7004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.964676228228692e-05,
+      "loss": 0.2018,
+      "step": 7005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9644205350788542e-05,
+      "loss": 0.583,
+      "step": 7006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.964164827002549e-05,
+      "loss": 0.1334,
+      "step": 7007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9639091040079934e-05,
+      "loss": 1.324,
+      "step": 7008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9636533661034074e-05,
+      "loss": 1.4286,
+      "step": 7009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9633976132970098e-05,
+      "loss": 0.5029,
+      "step": 7010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9631418455970205e-05,
+      "loss": 1.4267,
+      "step": 7011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9628860630116603e-05,
+      "loss": 0.1913,
+      "step": 7012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9626302655491497e-05,
+      "loss": 0.161,
+      "step": 7013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.96237445321771e-05,
+      "loss": 0.2183,
+      "step": 7014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.962118626025563e-05,
+      "loss": 0.5558,
+      "step": 7015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9618627839809314e-05,
+      "loss": 0.1033,
+      "step": 7016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9616069270920373e-05,
+      "loss": 0.0869,
+      "step": 7017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.961351055367104e-05,
+      "loss": 0.056,
+      "step": 7018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9610951688143558e-05,
+      "loss": 0.2957,
+      "step": 7019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9608392674420167e-05,
+      "loss": 0.7116,
+      "step": 7020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9605833512583114e-05,
+      "loss": 1.3672,
+      "step": 7021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9603274202714647e-05,
+      "loss": 0.1481,
+      "step": 7022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9600714744897022e-05,
+      "loss": 0.5048,
+      "step": 7023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9598155139212503e-05,
+      "loss": 0.8668,
+      "step": 7024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9595595385743353e-05,
+      "loss": 1.5601,
+      "step": 7025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9593035484571848e-05,
+      "loss": 1.2815,
+      "step": 7026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.959047543578026e-05,
+      "loss": 0.5796,
+      "step": 7027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9587915239450865e-05,
+      "loss": 0.4197,
+      "step": 7028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.958535489566595e-05,
+      "loss": 0.0972,
+      "step": 7029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9582794404507813e-05,
+      "loss": 0.8735,
+      "step": 7030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.958023376605873e-05,
+      "loss": 0.7553,
+      "step": 7031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9577672980401016e-05,
+      "loss": 0.8821,
+      "step": 7032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9575112047616965e-05,
+      "loss": 1.039,
+      "step": 7033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9572550967788888e-05,
+      "loss": 0.6604,
+      "step": 7034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9569989740999107e-05,
+      "loss": 1.0505,
+      "step": 7035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9567428367329927e-05,
+      "loss": 0.4376,
+      "step": 7036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9564866846863676e-05,
+      "loss": 0.5392,
+      "step": 7037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9562305179682682e-05,
+      "loss": 0.3394,
+      "step": 7038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9559743365869277e-05,
+      "loss": 0.5462,
+      "step": 7039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9557181405505796e-05,
+      "loss": 0.1102,
+      "step": 7040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9554619298674574e-05,
+      "loss": 1.0208,
+      "step": 7041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.955205704545797e-05,
+      "loss": 0.8399,
+      "step": 7042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9549494645938327e-05,
+      "loss": 0.9196,
+      "step": 7043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9546932100198006e-05,
+      "loss": 0.3555,
+      "step": 7044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.954436940831936e-05,
+      "loss": 0.9849,
+      "step": 7045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9541806570384756e-05,
+      "loss": 0.7734,
+      "step": 7046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9539243586476565e-05,
+      "loss": 0.4169,
+      "step": 7047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9536680456677157e-05,
+      "loss": 0.2731,
+      "step": 7048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.953411718106892e-05,
+      "loss": 0.3414,
+      "step": 7049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9531553759734234e-05,
+      "loss": 0.2789,
+      "step": 7050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9528990192755484e-05,
+      "loss": 1.9077,
+      "step": 7051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9526426480215064e-05,
+      "loss": 1.5062,
+      "step": 7052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9523862622195377e-05,
+      "loss": 0.0699,
+      "step": 7053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9521298618778815e-05,
+      "loss": 1.3964,
+      "step": 7054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9518734470047798e-05,
+      "loss": 1.7889,
+      "step": 7055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.951617017608472e-05,
+      "loss": 0.3718,
+      "step": 7056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9513605736972013e-05,
+      "loss": 0.0503,
+      "step": 7057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9511041152792098e-05,
+      "loss": 0.4365,
+      "step": 7058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.950847642362739e-05,
+      "loss": 0.4072,
+      "step": 7059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9505911549560325e-05,
+      "loss": 0.5698,
+      "step": 7060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.950334653067334e-05,
+      "loss": 0.4634,
+      "step": 7061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9500781367048876e-05,
+      "loss": 0.067,
+      "step": 7062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.949821605876937e-05,
+      "loss": 1.1827,
+      "step": 7063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9495650605917274e-05,
+      "loss": 0.9879,
+      "step": 7064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.949308500857504e-05,
+      "loss": 0.0753,
+      "step": 7065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9490519266825134e-05,
+      "loss": 1.5133,
+      "step": 7066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9487953380750016e-05,
+      "loss": 0.161,
+      "step": 7067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9485387350432146e-05,
+      "loss": 0.4071,
+      "step": 7068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9482821175954e-05,
+      "loss": 0.1377,
+      "step": 7069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9480254857398064e-05,
+      "loss": 0.3379,
+      "step": 7070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.947768839484681e-05,
+      "loss": 0.175,
+      "step": 7071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.947512178838272e-05,
+      "loss": 0.1621,
+      "step": 7072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.947255503808829e-05,
+      "loss": 1.6903,
+      "step": 7073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9469988144046022e-05,
+      "loss": 1.4999,
+      "step": 7074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9467421106338407e-05,
+      "loss": 0.4346,
+      "step": 7075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9464853925047953e-05,
+      "loss": 0.2779,
+      "step": 7076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.946228660025717e-05,
+      "loss": 0.2716,
+      "step": 7077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9459719132048565e-05,
+      "loss": 0.597,
+      "step": 7078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9457151520504663e-05,
+      "loss": 1.3509,
+      "step": 7079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9454583765707988e-05,
+      "loss": 0.2286,
+      "step": 7080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9452015867741064e-05,
+      "loss": 0.2002,
+      "step": 7081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9449447826686425e-05,
+      "loss": 0.8512,
+      "step": 7082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9446879642626605e-05,
+      "loss": 0.5892,
+      "step": 7083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9444311315644154e-05,
+      "loss": 0.4612,
+      "step": 7084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9441742845821608e-05,
+      "loss": 0.2834,
+      "step": 7085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.943917423324152e-05,
+      "loss": 0.4432,
+      "step": 7086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9436605477986438e-05,
+      "loss": 0.3275,
+      "step": 7087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9434036580138943e-05,
+      "loss": 0.4141,
+      "step": 7088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9431467539781583e-05,
+      "loss": 0.2436,
+      "step": 7089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.942889835699693e-05,
+      "loss": 0.7608,
+      "step": 7090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9426329031867558e-05,
+      "loss": 0.9043,
+      "step": 7091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9423759564476044e-05,
+      "loss": 0.6044,
+      "step": 7092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.942118995490497e-05,
+      "loss": 0.2187,
+      "step": 7093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9418620203236928e-05,
+      "loss": 0.1421,
+      "step": 7094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9416050309554498e-05,
+      "loss": 0.3713,
+      "step": 7095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9413480273940293e-05,
+      "loss": 1.005,
+      "step": 7096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9410910096476905e-05,
+      "loss": 0.662,
+      "step": 7097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9408339777246938e-05,
+      "loss": 0.4622,
+      "step": 7098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9405769316333007e-05,
+      "loss": 0.34,
+      "step": 7099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9403198713817723e-05,
+      "loss": 0.663,
+      "step": 7100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9400627969783704e-05,
+      "loss": 0.0377,
+      "step": 7101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.939805708431357e-05,
+      "loss": 1.4318,
+      "step": 7102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9395486057489965e-05,
+      "loss": 1.4512,
+      "step": 7103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.939291488939551e-05,
+      "loss": 0.1419,
+      "step": 7104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9390343580112842e-05,
+      "loss": 0.1018,
+      "step": 7105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9387772129724605e-05,
+      "loss": 0.4331,
+      "step": 7106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9385200538313445e-05,
+      "loss": 1.2663,
+      "step": 7107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9382628805962013e-05,
+      "loss": 0.4439,
+      "step": 7108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9380056932752963e-05,
+      "loss": 0.1971,
+      "step": 7109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9377484918768955e-05,
+      "loss": 0.2173,
+      "step": 7110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9374912764092653e-05,
+      "loss": 1.7837,
+      "step": 7111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9372340468806736e-05,
+      "loss": 0.259,
+      "step": 7112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9369768032993866e-05,
+      "loss": 1.0323,
+      "step": 7113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9367195456736722e-05,
+      "loss": 0.9396,
+      "step": 7114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9364622740117993e-05,
+      "loss": 0.2563,
+      "step": 7115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.936204988322036e-05,
+      "loss": 1.8441,
+      "step": 7116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9359476886126514e-05,
+      "loss": 0.7201,
+      "step": 7117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9356903748919155e-05,
+      "loss": 0.7588,
+      "step": 7118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.935433047168098e-05,
+      "loss": 0.0653,
+      "step": 7119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.93517570544947e-05,
+      "loss": 0.956,
+      "step": 7120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9349183497443024e-05,
+      "loss": 0.786,
+      "step": 7121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.934660980060866e-05,
+      "loss": 0.0538,
+      "step": 7122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.934403596407433e-05,
+      "loss": 1.5198,
+      "step": 7123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9341461987922754e-05,
+      "loss": 0.9286,
+      "step": 7124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.933888787223666e-05,
+      "loss": 0.4706,
+      "step": 7125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9336313617098788e-05,
+      "loss": 0.4636,
+      "step": 7126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9333739222591867e-05,
+      "loss": 1.1719,
+      "step": 7127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.933116468879864e-05,
+      "loss": 1.0499,
+      "step": 7128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9328590015801853e-05,
+      "loss": 0.2613,
+      "step": 7129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9326015203684256e-05,
+      "loss": 1.2915,
+      "step": 7130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9323440252528604e-05,
+      "loss": 1.8237,
+      "step": 7131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9320865162417647e-05,
+      "loss": 0.5906,
+      "step": 7132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9318289933434158e-05,
+      "loss": 0.6198,
+      "step": 7133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9315714565660908e-05,
+      "loss": 0.9763,
+      "step": 7134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9313139059180664e-05,
+      "loss": 0.6515,
+      "step": 7135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9310563414076196e-05,
+      "loss": 0.5701,
+      "step": 7136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9307987630430295e-05,
+      "loss": 1.2146,
+      "step": 7137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9305411708325746e-05,
+      "loss": 0.4476,
+      "step": 7138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9302835647845333e-05,
+      "loss": 0.781,
+      "step": 7139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9300259449071857e-05,
+      "loss": 0.1478,
+      "step": 7140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.929768311208811e-05,
+      "loss": 1.075,
+      "step": 7141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9295106636976902e-05,
+      "loss": 0.6491,
+      "step": 7142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.929253002382104e-05,
+      "loss": 0.5305,
+      "step": 7143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9289953272703336e-05,
+      "loss": 1.079,
+      "step": 7144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.92873763837066e-05,
+      "loss": 1.2322,
+      "step": 7145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9284799356913662e-05,
+      "loss": 0.9803,
+      "step": 7146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9282222192407346e-05,
+      "loss": 0.7345,
+      "step": 7147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9279644890270474e-05,
+      "loss": 0.5804,
+      "step": 7148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9277067450585893e-05,
+      "loss": 0.7621,
+      "step": 7149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.927448987343643e-05,
+      "loss": 0.5517,
+      "step": 7150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.927191215890494e-05,
+      "loss": 1.1105,
+      "step": 7151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.926933430707426e-05,
+      "loss": 1.1308,
+      "step": 7152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9266756318027254e-05,
+      "loss": 0.8606,
+      "step": 7153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9264178191846763e-05,
+      "loss": 0.3054,
+      "step": 7154
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.926159992861566e-05,
+      "loss": 0.3638,
+      "step": 7155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9259021528416802e-05,
+      "loss": 0.512,
+      "step": 7156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.925644299133307e-05,
+      "loss": 0.3627,
+      "step": 7157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9253864317447325e-05,
+      "loss": 1.7415,
+      "step": 7158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9251285506842458e-05,
+      "loss": 1.4844,
+      "step": 7159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.924870655960134e-05,
+      "loss": 0.7769,
+      "step": 7160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.924612747580687e-05,
+      "loss": 1.4498,
+      "step": 7161
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9243548255541933e-05,
+      "loss": 0.4299,
+      "step": 7162
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.924096889888942e-05,
+      "loss": 0.0307,
+      "step": 7163
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.923838940593224e-05,
+      "loss": 1.5158,
+      "step": 7164
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9235809776753293e-05,
+      "loss": 1.6401,
+      "step": 7165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.92332300114355e-05,
+      "loss": 0.2549,
+      "step": 7166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.923065011006176e-05,
+      "loss": 0.9045,
+      "step": 7167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.922807007271499e-05,
+      "loss": 0.5329,
+      "step": 7168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9225489899478125e-05,
+      "loss": 1.2467,
+      "step": 7169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9222909590434087e-05,
+      "loss": 1.2984,
+      "step": 7170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9220329145665796e-05,
+      "loss": 0.2585,
+      "step": 7171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9217748565256204e-05,
+      "loss": 0.1861,
+      "step": 7172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9215167849288245e-05,
+      "loss": 1.1483,
+      "step": 7173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9212586997844854e-05,
+      "loss": 0.2411,
+      "step": 7174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9210006011008996e-05,
+      "loss": 0.543,
+      "step": 7175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9207424888863613e-05,
+      "loss": 0.2309,
+      "step": 7176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9204843631491662e-05,
+      "loss": 1.0315,
+      "step": 7177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.920226223897611e-05,
+      "loss": 1.1151,
+      "step": 7178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9199680711399913e-05,
+      "loss": 1.3388,
+      "step": 7179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9197099048846056e-05,
+      "loss": 1.5402,
+      "step": 7180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.91945172513975e-05,
+      "loss": 0.3629,
+      "step": 7181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.919193531913723e-05,
+      "loss": 1.0058,
+      "step": 7182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.918935325214823e-05,
+      "loss": 0.4699,
+      "step": 7183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9186771050513492e-05,
+      "loss": 0.5136,
+      "step": 7184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9184188714316e-05,
+      "loss": 0.374,
+      "step": 7185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.918160624363875e-05,
+      "loss": 0.8574,
+      "step": 7186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.917902363856474e-05,
+      "loss": 0.6477,
+      "step": 7187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9176440899176984e-05,
+      "loss": 0.4677,
+      "step": 7188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9173858025558495e-05,
+      "loss": 0.0932,
+      "step": 7189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.917127501779227e-05,
+      "loss": 0.4662,
+      "step": 7190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.916869187596134e-05,
+      "loss": 0.233,
+      "step": 7191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.916610860014872e-05,
+      "loss": 0.5441,
+      "step": 7192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.916352519043744e-05,
+      "loss": 0.1269,
+      "step": 7193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.916094164691053e-05,
+      "loss": 0.0956,
+      "step": 7194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9158357969651024e-05,
+      "loss": 0.1475,
+      "step": 7195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.915577415874196e-05,
+      "loss": 0.4481,
+      "step": 7196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9153190214266387e-05,
+      "loss": 0.2976,
+      "step": 7197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9150606136307356e-05,
+      "loss": 0.7349,
+      "step": 7198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9148021924947908e-05,
+      "loss": 0.6214,
+      "step": 7199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.91454375802711e-05,
+      "loss": 1.0942,
+      "step": 7200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9142853102360002e-05,
+      "loss": 0.0464,
+      "step": 7201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9140268491297672e-05,
+      "loss": 0.0353,
+      "step": 7202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9137683747167185e-05,
+      "loss": 0.274,
+      "step": 7203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.913509887005161e-05,
+      "loss": 0.607,
+      "step": 7204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9132513860034025e-05,
+      "loss": 0.2477,
+      "step": 7205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.912992871719752e-05,
+      "loss": 0.5506,
+      "step": 7206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9127343441625167e-05,
+      "loss": 0.4079,
+      "step": 7207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.912475803340007e-05,
+      "loss": 0.3388,
+      "step": 7208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9122172492605317e-05,
+      "loss": 0.1779,
+      "step": 7209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9119586819324003e-05,
+      "loss": 0.4001,
+      "step": 7210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9117001013639242e-05,
+      "loss": 0.0653,
+      "step": 7211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9114415075634146e-05,
+      "loss": 0.9562,
+      "step": 7212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9111829005391807e-05,
+      "loss": 0.3552,
+      "step": 7213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9109242802995353e-05,
+      "loss": 0.1692,
+      "step": 7214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.910665646852791e-05,
+      "loss": 0.3821,
+      "step": 7215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9104070002072595e-05,
+      "loss": 0.0448,
+      "step": 7216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9101483403712535e-05,
+      "loss": 0.7451,
+      "step": 7217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.909889667353087e-05,
+      "loss": 0.5234,
+      "step": 7218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9096309811610727e-05,
+      "loss": 0.7561,
+      "step": 7219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9093722818035262e-05,
+      "loss": 1.3035,
+      "step": 7220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9091135692887615e-05,
+      "loss": 0.5729,
+      "step": 7221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9088548436250936e-05,
+      "loss": 1.9159,
+      "step": 7222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9085961048208375e-05,
+      "loss": 0.1446,
+      "step": 7223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9083373528843097e-05,
+      "loss": 1.3841,
+      "step": 7224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9080785878238255e-05,
+      "loss": 0.2999,
+      "step": 7225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9078198096477033e-05,
+      "loss": 0.83,
+      "step": 7226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9075610183642586e-05,
+      "loss": 0.1113,
+      "step": 7227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9073022139818098e-05,
+      "loss": 1.3403,
+      "step": 7228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9070433965086744e-05,
+      "loss": 0.5914,
+      "step": 7229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9067845659531715e-05,
+      "loss": 0.7624,
+      "step": 7230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.906525722323619e-05,
+      "loss": 0.2497,
+      "step": 7231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.906266865628337e-05,
+      "loss": 0.0602,
+      "step": 7232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9060079958756444e-05,
+      "loss": 0.6052,
+      "step": 7233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9057491130738614e-05,
+      "loss": 0.0518,
+      "step": 7234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.905490217231309e-05,
+      "loss": 1.1707,
+      "step": 7235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.905231308356308e-05,
+      "loss": 1.5845,
+      "step": 7236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9049723864571792e-05,
+      "loss": 1.5565,
+      "step": 7237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9047134515422448e-05,
+      "loss": 0.304,
+      "step": 7238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9044545036198268e-05,
+      "loss": 0.8705,
+      "step": 7239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9041955426982477e-05,
+      "loss": 0.5108,
+      "step": 7240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9039365687858303e-05,
+      "loss": 1.1523,
+      "step": 7241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9036775818908986e-05,
+      "loss": 0.2727,
+      "step": 7242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9034185820217765e-05,
+      "loss": 0.7566,
+      "step": 7243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9031595691867877e-05,
+      "loss": 0.6134,
+      "step": 7244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.902900543394257e-05,
+      "loss": 0.6042,
+      "step": 7245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9026415046525096e-05,
+      "loss": 0.1292,
+      "step": 7246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9023824529698713e-05,
+      "loss": 0.8171,
+      "step": 7247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.902123388354667e-05,
+      "loss": 0.3722,
+      "step": 7248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.901864310815224e-05,
+      "loss": 0.4859,
+      "step": 7249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9016052203598688e-05,
+      "loss": 0.7379,
+      "step": 7250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9013461169969284e-05,
+      "loss": 0.2518,
+      "step": 7251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.901087000734731e-05,
+      "loss": 0.2995,
+      "step": 7252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.900827871581604e-05,
+      "loss": 0.3926,
+      "step": 7253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9005687295458754e-05,
+      "loss": 0.6684,
+      "step": 7254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.900309574635875e-05,
+      "loss": 1.5707,
+      "step": 7255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9000504068599308e-05,
+      "loss": 1.4101,
+      "step": 7256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.899791226226374e-05,
+      "loss": 0.2365,
+      "step": 7257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8995320327435337e-05,
+      "loss": 1.1168,
+      "step": 7258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89927282641974e-05,
+      "loss": 0.3907,
+      "step": 7259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.899013607263325e-05,
+      "loss": 0.0312,
+      "step": 7260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8987543752826192e-05,
+      "loss": 1.5577,
+      "step": 7261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8984951304859546e-05,
+      "loss": 0.666,
+      "step": 7262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8982358728816626e-05,
+      "loss": 0.0537,
+      "step": 7263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8979766024780767e-05,
+      "loss": 0.561,
+      "step": 7264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8977173192835292e-05,
+      "loss": 0.2241,
+      "step": 7265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.897458023306354e-05,
+      "loss": 0.2124,
+      "step": 7266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8971987145548844e-05,
+      "loss": 0.0749,
+      "step": 7267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896939393037455e-05,
+      "loss": 0.3427,
+      "step": 7268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8966800587623997e-05,
+      "loss": 0.0631,
+      "step": 7269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8964207117380545e-05,
+      "loss": 0.2496,
+      "step": 7270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961613519727535e-05,
+      "loss": 0.0888,
+      "step": 7271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895901979474834e-05,
+      "loss": 0.0428,
+      "step": 7272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895642594252631e-05,
+      "loss": 0.4027,
+      "step": 7273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953831963144813e-05,
+      "loss": 1.6534,
+      "step": 7274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951237856687233e-05,
+      "loss": 0.1613,
+      "step": 7275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894864362323693e-05,
+      "loss": 0.4061,
+      "step": 7276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946049262877284e-05,
+      "loss": 0.1584,
+      "step": 7277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943454775691682e-05,
+      "loss": 0.0923,
+      "step": 7278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8940860161763507e-05,
+      "loss": 0.1028,
+      "step": 7279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938265421176158e-05,
+      "loss": 1.3246,
+      "step": 7280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893567055401302e-05,
+      "loss": 0.5358,
+      "step": 7281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933075560357497e-05,
+      "loss": 0.2302,
+      "step": 7282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893048044029299e-05,
+      "loss": 0.726,
+      "step": 7283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927885193902912e-05,
+      "loss": 1.1487,
+      "step": 7284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925289821270664e-05,
+      "loss": 1.0058,
+      "step": 7285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892269432247967e-05,
+      "loss": 0.283,
+      "step": 7286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892009869761334e-05,
+      "loss": 0.7243,
+      "step": 7287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8917502946755107e-05,
+      "loss": 0.2621,
+      "step": 7288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8914907069988405e-05,
+      "loss": 0.7946,
+      "step": 7289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8912311067396645e-05,
+      "loss": 0.1425,
+      "step": 7290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.890971493906327e-05,
+      "loss": 0.8206,
+      "step": 7291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8907118685071726e-05,
+      "loss": 0.383,
+      "step": 7292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8904522305505457e-05,
+      "loss": 0.2932,
+      "step": 7293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89019258004479e-05,
+      "loss": 0.1882,
+      "step": 7294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.889932916998251e-05,
+      "loss": 1.0683,
+      "step": 7295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.889673241419275e-05,
+      "loss": 0.5834,
+      "step": 7296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8894135533162078e-05,
+      "loss": 0.6345,
+      "step": 7297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8891538526973956e-05,
+      "loss": 0.8628,
+      "step": 7298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8888941395711845e-05,
+      "loss": 0.7176,
+      "step": 7299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8886344139459227e-05,
+      "loss": 0.5614,
+      "step": 7300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.888374675829957e-05,
+      "loss": 1.8765,
+      "step": 7301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8881149252316355e-05,
+      "loss": 0.3498,
+      "step": 7302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8878551621593077e-05,
+      "loss": 0.0494,
+      "step": 7303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8875953866213207e-05,
+      "loss": 0.0319,
+      "step": 7304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8873355986260246e-05,
+      "loss": 0.4972,
+      "step": 7305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.887075798181769e-05,
+      "loss": 0.4685,
+      "step": 7306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8868159852969037e-05,
+      "loss": 0.4669,
+      "step": 7307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.886556159979779e-05,
+      "loss": 0.0934,
+      "step": 7308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8862963222387454e-05,
+      "loss": 0.0981,
+      "step": 7309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8860364720821546e-05,
+      "loss": 0.0363,
+      "step": 7310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8857766095183587e-05,
+      "loss": 0.4303,
+      "step": 7311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8855167345557084e-05,
+      "loss": 1.2635,
+      "step": 7312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8852568472025568e-05,
+      "loss": 1.745,
+      "step": 7313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8849969474672567e-05,
+      "loss": 0.719,
+      "step": 7314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.884737035358161e-05,
+      "loss": 0.2738,
+      "step": 7315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8844771108836236e-05,
+      "loss": 0.9028,
+      "step": 7316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8842171740519982e-05,
+      "loss": 1.3099,
+      "step": 7317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.883957224871639e-05,
+      "loss": 0.6109,
+      "step": 7318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8836972633509012e-05,
+      "loss": 0.3623,
+      "step": 7319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.88343728949814e-05,
+      "loss": 0.9725,
+      "step": 7320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8831773033217103e-05,
+      "loss": 0.7675,
+      "step": 7321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8829173048299688e-05,
+      "loss": 0.4476,
+      "step": 7322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8826572940312712e-05,
+      "loss": 1.093,
+      "step": 7323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.882397270933975e-05,
+      "loss": 0.8932,
+      "step": 7324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8821372355464365e-05,
+      "loss": 0.6611,
+      "step": 7325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8818771878770138e-05,
+      "loss": 0.6086,
+      "step": 7326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8816171279340643e-05,
+      "loss": 0.9124,
+      "step": 7327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8813570557259473e-05,
+      "loss": 0.8622,
+      "step": 7328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8810969712610206e-05,
+      "loss": 0.2554,
+      "step": 7329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8808368745476438e-05,
+      "loss": 0.3276,
+      "step": 7330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8805767655941757e-05,
+      "loss": 1.135,
+      "step": 7331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8803166444089772e-05,
+      "loss": 1.3823,
+      "step": 7332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8800565110004076e-05,
+      "loss": 0.6777,
+      "step": 7333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8797963653768286e-05,
+      "loss": 1.0704,
+      "step": 7334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8795362075466006e-05,
+      "loss": 0.6537,
+      "step": 7335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.879276037518085e-05,
+      "loss": 0.7835,
+      "step": 7336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.879015855299644e-05,
+      "loss": 0.194,
+      "step": 7337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8787556608996403e-05,
+      "loss": 0.4352,
+      "step": 7338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8784954543264355e-05,
+      "loss": 1.2295,
+      "step": 7339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8782352355883932e-05,
+      "loss": 0.8942,
+      "step": 7340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.877975004693876e-05,
+      "loss": 0.879,
+      "step": 7341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8777147616512496e-05,
+      "loss": 0.5827,
+      "step": 7342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8774545064688768e-05,
+      "loss": 1.0882,
+      "step": 7343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8771942391551223e-05,
+      "loss": 0.9737,
+      "step": 7344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8769339597183513e-05,
+      "loss": 0.512,
+      "step": 7345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.876673668166929e-05,
+      "loss": 0.4282,
+      "step": 7346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8764133645092214e-05,
+      "loss": 0.4507,
+      "step": 7347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8761530487535944e-05,
+      "loss": 0.7775,
+      "step": 7348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.875892720908415e-05,
+      "loss": 0.129,
+      "step": 7349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8756323809820494e-05,
+      "loss": 0.7929,
+      "step": 7350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8753720289828655e-05,
+      "loss": 0.1706,
+      "step": 7351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8751116649192313e-05,
+      "loss": 1.1952,
+      "step": 7352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8748512887995143e-05,
+      "loss": 0.0344,
+      "step": 7353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8745909006320825e-05,
+      "loss": 0.0721,
+      "step": 7354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8743305004253062e-05,
+      "loss": 1.3858,
+      "step": 7355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.874070088187553e-05,
+      "loss": 0.0349,
+      "step": 7356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8738096639271942e-05,
+      "loss": 0.505,
+      "step": 7357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.873549227652599e-05,
+      "loss": 0.4759,
+      "step": 7358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8732887793721374e-05,
+      "loss": 0.241,
+      "step": 7359
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.873028319094181e-05,
+      "loss": 0.5119,
+      "step": 7360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8727678468271008e-05,
+      "loss": 1.7732,
+      "step": 7361
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.872507362579268e-05,
+      "loss": 0.2537,
+      "step": 7362
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8722468663590546e-05,
+      "loss": 1.3971,
+      "step": 7363
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8719863581748332e-05,
+      "loss": 0.4006,
+      "step": 7364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8717258380349767e-05,
+      "loss": 0.7142,
+      "step": 7365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.871465305947858e-05,
+      "loss": 0.764,
+      "step": 7366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8712047619218505e-05,
+      "loss": 0.3168,
+      "step": 7367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.870944205965328e-05,
+      "loss": 0.7915,
+      "step": 7368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.870683638086665e-05,
+      "loss": 0.3336,
+      "step": 7369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8704230582942364e-05,
+      "loss": 0.5293,
+      "step": 7370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.870162466596416e-05,
+      "loss": 0.0967,
+      "step": 7371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8699018630015806e-05,
+      "loss": 0.1831,
+      "step": 7372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8696412475181057e-05,
+      "loss": 0.6923,
+      "step": 7373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.869380620154367e-05,
+      "loss": 1.0218,
+      "step": 7374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.869119980918742e-05,
+      "loss": 0.8679,
+      "step": 7375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8688593298196062e-05,
+      "loss": 1.6964,
+      "step": 7376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.868598666865338e-05,
+      "loss": 0.362,
+      "step": 7377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8683379920643147e-05,
+      "loss": 1.228,
+      "step": 7378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8680773054249142e-05,
+      "loss": 0.3032,
+      "step": 7379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8678166069555158e-05,
+      "loss": 0.6916,
+      "step": 7380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.867555896664498e-05,
+      "loss": 1.4424,
+      "step": 7381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.867295174560239e-05,
+      "loss": 1.0416,
+      "step": 7382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8670344406511203e-05,
+      "loss": 0.4846,
+      "step": 7383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8667736949455198e-05,
+      "loss": 0.6249,
+      "step": 7384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8665129374518194e-05,
+      "loss": 0.1668,
+      "step": 7385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8662521681783992e-05,
+      "loss": 0.4003,
+      "step": 7386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.86599138713364e-05,
+      "loss": 0.5786,
+      "step": 7387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8657305943259243e-05,
+      "loss": 0.5815,
+      "step": 7388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8654697897636333e-05,
+      "loss": 0.6812,
+      "step": 7389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8652089734551492e-05,
+      "loss": 0.5333,
+      "step": 7390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.864948145408855e-05,
+      "loss": 0.5441,
+      "step": 7391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8646873056331335e-05,
+      "loss": 0.622,
+      "step": 7392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.864426454136368e-05,
+      "loss": 0.3374,
+      "step": 7393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.864165590926942e-05,
+      "loss": 0.5561,
+      "step": 7394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8639047160132403e-05,
+      "loss": 0.796,
+      "step": 7395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.863643829403647e-05,
+      "loss": 0.7939,
+      "step": 7396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8633829311065474e-05,
+      "loss": 0.7849,
+      "step": 7397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.863122021130326e-05,
+      "loss": 0.4897,
+      "step": 7398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8628610994833695e-05,
+      "loss": 0.6481,
+      "step": 7399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8626001661740628e-05,
+      "loss": 0.1741,
+      "step": 7400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8623392212107933e-05,
+      "loss": 0.0433,
+      "step": 7401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8620782646019466e-05,
+      "loss": 0.0392,
+      "step": 7402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.861817296355911e-05,
+      "loss": 0.0364,
+      "step": 7403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8615563164810735e-05,
+      "loss": 0.0772,
+      "step": 7404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8612953249858217e-05,
+      "loss": 0.0366,
+      "step": 7405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8610343218785447e-05,
+      "loss": 1.7956,
+      "step": 7406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8607733071676306e-05,
+      "loss": 0.5071,
+      "step": 7407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.860512280861468e-05,
+      "loss": 0.5162,
+      "step": 7408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8602512429684472e-05,
+      "loss": 1.5273,
+      "step": 7409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8599901934969567e-05,
+      "loss": 0.269,
+      "step": 7410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8597291324553885e-05,
+      "loss": 0.4821,
+      "step": 7411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8594680598521313e-05,
+      "loss": 0.0772,
+      "step": 7412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8592069756955767e-05,
+      "loss": 0.4759,
+      "step": 7413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.858945879994116e-05,
+      "loss": 0.4405,
+      "step": 7414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8586847727561408e-05,
+      "loss": 0.3275,
+      "step": 7415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.858423653990043e-05,
+      "loss": 0.2981,
+      "step": 7416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8581625237042147e-05,
+      "loss": 0.2798,
+      "step": 7417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8579013819070487e-05,
+      "loss": 0.2484,
+      "step": 7418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8576402286069382e-05,
+      "loss": 0.2525,
+      "step": 7419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8573790638122775e-05,
+      "loss": 0.2919,
+      "step": 7420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8571178875314594e-05,
+      "loss": 0.0424,
+      "step": 7421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8568566997728783e-05,
+      "loss": 0.5549,
+      "step": 7422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8565955005449287e-05,
+      "loss": 0.2218,
+      "step": 7423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8563342898560056e-05,
+      "loss": 0.2204,
+      "step": 7424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8560730677145042e-05,
+      "loss": 0.7588,
+      "step": 7425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8558118341288206e-05,
+      "loss": 0.6389,
+      "step": 7426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8555505891073506e-05,
+      "loss": 0.4448,
+      "step": 7427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8552893326584907e-05,
+      "loss": 0.2913,
+      "step": 7428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8550280647906377e-05,
+      "loss": 0.4783,
+      "step": 7429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8547667855121883e-05,
+      "loss": 0.5361,
+      "step": 7430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8545054948315408e-05,
+      "loss": 0.3231,
+      "step": 7431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.854244192757092e-05,
+      "loss": 0.9011,
+      "step": 7432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.853982879297241e-05,
+      "loss": 0.8942,
+      "step": 7433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8537215544603862e-05,
+      "loss": 0.8044,
+      "step": 7434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8534602182549267e-05,
+      "loss": 1.4048,
+      "step": 7435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8531988706892615e-05,
+      "loss": 0.7552,
+      "step": 7436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8529375117717906e-05,
+      "loss": 0.163,
+      "step": 7437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.852676141510914e-05,
+      "loss": 0.7921,
+      "step": 7438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.852414759915032e-05,
+      "loss": 0.0748,
+      "step": 7439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8521533669925455e-05,
+      "loss": 0.1751,
+      "step": 7440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8518919627518557e-05,
+      "loss": 0.2684,
+      "step": 7441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.851630547201364e-05,
+      "loss": 1.1558,
+      "step": 7442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8513691203494726e-05,
+      "loss": 0.2478,
+      "step": 7443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8511076822045835e-05,
+      "loss": 0.8264,
+      "step": 7444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8508462327750993e-05,
+      "loss": 0.74,
+      "step": 7445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.850584772069423e-05,
+      "loss": 0.8855,
+      "step": 7446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.850323300095958e-05,
+      "loss": 1.824,
+      "step": 7447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.850061816863108e-05,
+      "loss": 0.2941,
+      "step": 7448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.849800322379277e-05,
+      "loss": 0.6648,
+      "step": 7449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8495388166528696e-05,
+      "loss": 0.4431,
+      "step": 7450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.84927729969229e-05,
+      "loss": 1.4286,
+      "step": 7451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8490157715059448e-05,
+      "loss": 0.0378,
+      "step": 7452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.848754232102238e-05,
+      "loss": 0.2162,
+      "step": 7453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8484926814895756e-05,
+      "loss": 0.0328,
+      "step": 7454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8482311196763647e-05,
+      "loss": 0.2053,
+      "step": 7455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8479695466710107e-05,
+      "loss": 1.4511,
+      "step": 7456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8477079624819226e-05,
+      "loss": 0.9978,
+      "step": 7457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8474463671175056e-05,
+      "loss": 0.272,
+      "step": 7458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.847184760586168e-05,
+      "loss": 0.3219,
+      "step": 7459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8469231428963188e-05,
+      "loss": 0.3484,
+      "step": 7460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.846661514056365e-05,
+      "loss": 0.0428,
+      "step": 7461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8463998740747157e-05,
+      "loss": 1.6232,
+      "step": 7462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8461382229597804e-05,
+      "loss": 1.1935,
+      "step": 7463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8458765607199683e-05,
+      "loss": 0.4413,
+      "step": 7464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.84561488736369e-05,
+      "loss": 1.8644,
+      "step": 7465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8453532028993544e-05,
+      "loss": 0.4864,
+      "step": 7466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.845091507335373e-05,
+      "loss": 0.8258,
+      "step": 7467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8448298006801555e-05,
+      "loss": 0.2864,
+      "step": 7468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.844568082942115e-05,
+      "loss": 0.3611,
+      "step": 7469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8443063541296617e-05,
+      "loss": 0.1442,
+      "step": 7470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8440446142512073e-05,
+      "loss": 0.4914,
+      "step": 7471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8437828633151654e-05,
+      "loss": 0.6854,
+      "step": 7472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8435211013299477e-05,
+      "loss": 0.5626,
+      "step": 7473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8432593283039682e-05,
+      "loss": 0.4256,
+      "step": 7474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.842997544245639e-05,
+      "loss": 1.3076,
+      "step": 7475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8427357491633747e-05,
+      "loss": 0.9569,
+      "step": 7476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8424739430655887e-05,
+      "loss": 0.9995,
+      "step": 7477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8422121259606964e-05,
+      "loss": 0.1859,
+      "step": 7478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8419502978571117e-05,
+      "loss": 0.7835,
+      "step": 7479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8416884587632505e-05,
+      "loss": 1.2627,
+      "step": 7480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8414266086875278e-05,
+      "loss": 0.7001,
+      "step": 7481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8411647476383595e-05,
+      "loss": 0.2097,
+      "step": 7482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8409028756241623e-05,
+      "loss": 0.5018,
+      "step": 7483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.840640992653352e-05,
+      "loss": 0.3596,
+      "step": 7484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.840379098734346e-05,
+      "loss": 1.4895,
+      "step": 7485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.840117193875561e-05,
+      "loss": 0.7425,
+      "step": 7486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8398552780854157e-05,
+      "loss": 0.7742,
+      "step": 7487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8395933513723275e-05,
+      "loss": 0.733,
+      "step": 7488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8393314137447142e-05,
+      "loss": 1.0487,
+      "step": 7489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.839069465210995e-05,
+      "loss": 0.8421,
+      "step": 7490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8388075057795893e-05,
+      "loss": 0.8988,
+      "step": 7491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8385455354589156e-05,
+      "loss": 0.3045,
+      "step": 7492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8382835542573943e-05,
+      "loss": 0.1379,
+      "step": 7493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8380215621834444e-05,
+      "loss": 1.3592,
+      "step": 7494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.837759559245488e-05,
+      "loss": 1.1959,
+      "step": 7495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8374975454519446e-05,
+      "loss": 0.6034,
+      "step": 7496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8372355208112367e-05,
+      "loss": 0.5349,
+      "step": 7497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8369734853317838e-05,
+      "loss": 0.5491,
+      "step": 7498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.836711439022009e-05,
+      "loss": 0.1791,
+      "step": 7499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.836449381890334e-05,
+      "loss": 0.5373,
+      "step": 7500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8361873139451815e-05,
+      "loss": 0.037,
+      "step": 7501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8359252351949743e-05,
+      "loss": 1.1064,
+      "step": 7502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8356631456481357e-05,
+      "loss": 0.0367,
+      "step": 7503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8354010453130895e-05,
+      "loss": 0.0354,
+      "step": 7504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.835138934198259e-05,
+      "loss": 0.0864,
+      "step": 7505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.834876812312069e-05,
+      "loss": 0.1461,
+      "step": 7506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8346146796629438e-05,
+      "loss": 1.3503,
+      "step": 7507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.834352536259308e-05,
+      "loss": 1.2815,
+      "step": 7508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8340903821095878e-05,
+      "loss": 0.9776,
+      "step": 7509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8338282172222077e-05,
+      "loss": 0.452,
+      "step": 7510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8335660416055948e-05,
+      "loss": 0.114,
+      "step": 7511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8333038552681747e-05,
+      "loss": 0.2526,
+      "step": 7512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8330416582183743e-05,
+      "loss": 0.3648,
+      "step": 7513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8327794504646205e-05,
+      "loss": 0.5305,
+      "step": 7514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.832517232015341e-05,
+      "loss": 0.9937,
+      "step": 7515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8322550028789633e-05,
+      "loss": 0.3009,
+      "step": 7516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8319927630639148e-05,
+      "loss": 1.0894,
+      "step": 7517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8317305125786245e-05,
+      "loss": 0.2329,
+      "step": 7518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8314682514315214e-05,
+      "loss": 0.4121,
+      "step": 7519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8312059796310343e-05,
+      "loss": 0.087,
+      "step": 7520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8309436971855923e-05,
+      "loss": 0.2647,
+      "step": 7521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8306814041036253e-05,
+      "loss": 0.3317,
+      "step": 7522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8304191003935636e-05,
+      "loss": 1.1905,
+      "step": 7523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8301567860638378e-05,
+      "loss": 0.6732,
+      "step": 7524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.829894461122878e-05,
+      "loss": 0.6118,
+      "step": 7525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8296321255791157e-05,
+      "loss": 0.8028,
+      "step": 7526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8293697794409825e-05,
+      "loss": 0.9491,
+      "step": 7527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.82910742271691e-05,
+      "loss": 0.5325,
+      "step": 7528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.828845055415331e-05,
+      "loss": 1.0578,
+      "step": 7529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.828582677544677e-05,
+      "loss": 0.8216,
+      "step": 7530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8283202891133813e-05,
+      "loss": 0.6635,
+      "step": 7531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8280578901298773e-05,
+      "loss": 1.4901,
+      "step": 7532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8277954806025976e-05,
+      "loss": 0.4899,
+      "step": 7533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8275330605399767e-05,
+      "loss": 0.6715,
+      "step": 7534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.827270629950449e-05,
+      "loss": 0.4664,
+      "step": 7535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8270081888424484e-05,
+      "loss": 1.1667,
+      "step": 7536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8267457372244106e-05,
+      "loss": 1.0729,
+      "step": 7537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.82648327510477e-05,
+      "loss": 1.4324,
+      "step": 7538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8262208024919623e-05,
+      "loss": 0.8134,
+      "step": 7539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8259583193944233e-05,
+      "loss": 1.3543,
+      "step": 7540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8256958258205895e-05,
+      "loss": 0.4723,
+      "step": 7541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8254333217788972e-05,
+      "loss": 0.5001,
+      "step": 7542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8251708072777835e-05,
+      "loss": 0.8685,
+      "step": 7543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8249082823256856e-05,
+      "loss": 0.4597,
+      "step": 7544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.824645746931041e-05,
+      "loss": 0.9354,
+      "step": 7545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8243832011022874e-05,
+      "loss": 0.5211,
+      "step": 7546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8241206448478633e-05,
+      "loss": 0.5185,
+      "step": 7547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8238580781762063e-05,
+      "loss": 0.7309,
+      "step": 7548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8235955010957567e-05,
+      "loss": 0.5122,
+      "step": 7549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8233329136149528e-05,
+      "loss": 0.271,
+      "step": 7550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8230703157422354e-05,
+      "loss": 1.2996,
+      "step": 7551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8228077074860426e-05,
+      "loss": 0.491,
+      "step": 7552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8225450888548156e-05,
+      "loss": 0.0337,
+      "step": 7553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8222824598569947e-05,
+      "loss": 1.7538,
+      "step": 7554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8220198205010213e-05,
+      "loss": 1.6294,
+      "step": 7555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8217571707953357e-05,
+      "loss": 0.2102,
+      "step": 7556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8214945107483802e-05,
+      "loss": 0.1206,
+      "step": 7557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8212318403685966e-05,
+      "loss": 0.1704,
+      "step": 7558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8209691596644274e-05,
+      "loss": 1.6948,
+      "step": 7559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8207064686443146e-05,
+      "loss": 0.9115,
+      "step": 7560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.820443767316701e-05,
+      "loss": 0.3867,
+      "step": 7561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.82018105569003e-05,
+      "loss": 1.0529,
+      "step": 7562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8199183337727456e-05,
+      "loss": 0.3058,
+      "step": 7563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8196556015732906e-05,
+      "loss": 1.4856,
+      "step": 7564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8193928591001107e-05,
+      "loss": 0.2102,
+      "step": 7565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8191301063616493e-05,
+      "loss": 0.5977,
+      "step": 7566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8188673433663515e-05,
+      "loss": 0.3043,
+      "step": 7567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8186045701226625e-05,
+      "loss": 0.1812,
+      "step": 7568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8183417866390285e-05,
+      "loss": 0.4992,
+      "step": 7569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8180789929238945e-05,
+      "loss": 0.2278,
+      "step": 7570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8178161889857068e-05,
+      "loss": 0.6494,
+      "step": 7571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8175533748329122e-05,
+      "loss": 0.7221,
+      "step": 7572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8172905504739575e-05,
+      "loss": 0.6604,
+      "step": 7573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.81702771591729e-05,
+      "loss": 1.0936,
+      "step": 7574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8167648711713566e-05,
+      "loss": 0.492,
+      "step": 7575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8165020162446057e-05,
+      "loss": 1.1268,
+      "step": 7576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8162391511454854e-05,
+      "loss": 0.0483,
+      "step": 7577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8159762758824442e-05,
+      "loss": 1.4848,
+      "step": 7578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8157133904639302e-05,
+      "loss": 0.0918,
+      "step": 7579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.815450494898393e-05,
+      "loss": 0.3451,
+      "step": 7580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8151875891942827e-05,
+      "loss": 0.2093,
+      "step": 7581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.814924673360049e-05,
+      "loss": 1.5038,
+      "step": 7582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8146617474041413e-05,
+      "loss": 0.3123,
+      "step": 7583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8143988113350096e-05,
+      "loss": 1.6325,
+      "step": 7584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8141358651611056e-05,
+      "loss": 0.8409,
+      "step": 7585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8138729088908806e-05,
+      "loss": 1.2108,
+      "step": 7586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8136099425327845e-05,
+      "loss": 0.2521,
+      "step": 7587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8133469660952714e-05,
+      "loss": 0.4264,
+      "step": 7588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8130839795867915e-05,
+      "loss": 0.1423,
+      "step": 7589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8128209830157976e-05,
+      "loss": 0.2221,
+      "step": 7590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8125579763907425e-05,
+      "loss": 1.2187,
+      "step": 7591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.81229495972008e-05,
+      "loss": 0.9696,
+      "step": 7592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.812031933012262e-05,
+      "loss": 0.5734,
+      "step": 7593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8117688962757426e-05,
+      "loss": 0.2657,
+      "step": 7594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8115058495189763e-05,
+      "loss": 0.3626,
+      "step": 7595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8112427927504178e-05,
+      "loss": 0.6199,
+      "step": 7596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8109797259785207e-05,
+      "loss": 0.8072,
+      "step": 7597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8107166492117403e-05,
+      "loss": 0.481,
+      "step": 7598
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.810453562458532e-05,
+      "loss": 0.5816,
+      "step": 7599
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8101904657273516e-05,
+      "loss": 0.446,
+      "step": 7600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8099273590266552e-05,
+      "loss": 0.2228,
+      "step": 7601
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8096642423648977e-05,
+      "loss": 0.0441,
+      "step": 7602
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.809401115750537e-05,
+      "loss": 0.8884,
+      "step": 7603
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8091379791920295e-05,
+      "loss": 0.0472,
+      "step": 7604
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8088748326978326e-05,
+      "loss": 1.3729,
+      "step": 7605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8086116762764045e-05,
+      "loss": 0.2473,
+      "step": 7606
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8083485099362013e-05,
+      "loss": 0.2212,
+      "step": 7607
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8080853336856824e-05,
+      "loss": 0.3666,
+      "step": 7608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8078221475333064e-05,
+      "loss": 0.2705,
+      "step": 7609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.807558951487531e-05,
+      "loss": 0.2374,
+      "step": 7610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8072957455568163e-05,
+      "loss": 0.7373,
+      "step": 7611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.807032529749622e-05,
+      "loss": 1.0105,
+      "step": 7612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8067693040744068e-05,
+      "loss": 0.3969,
+      "step": 7613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8065060685396318e-05,
+      "loss": 0.564,
+      "step": 7614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8062428231537562e-05,
+      "loss": 0.0584,
+      "step": 7615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.805979567925242e-05,
+      "loss": 0.3161,
+      "step": 7616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.805716302862549e-05,
+      "loss": 0.1139,
+      "step": 7617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8054530279741393e-05,
+      "loss": 0.2886,
+      "step": 7618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8051897432684746e-05,
+      "loss": 0.7647,
+      "step": 7619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8049264487540162e-05,
+      "loss": 1.0001,
+      "step": 7620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.804663144439227e-05,
+      "loss": 0.2049,
+      "step": 7621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.80439983033257e-05,
+      "loss": 0.7964,
+      "step": 7622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8041365064425076e-05,
+      "loss": 0.102,
+      "step": 7623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8038731727775023e-05,
+      "loss": 0.8341,
+      "step": 7624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.803609829346018e-05,
+      "loss": 0.2853,
+      "step": 7625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8033464761565193e-05,
+      "loss": 1.4607,
+      "step": 7626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.80308311321747e-05,
+      "loss": 0.3356,
+      "step": 7627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.802819740537335e-05,
+      "loss": 1.1312,
+      "step": 7628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.802556358124578e-05,
+      "loss": 0.9883,
+      "step": 7629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.802292965987665e-05,
+      "loss": 1.0165,
+      "step": 7630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8020295641350604e-05,
+      "loss": 1.0063,
+      "step": 7631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8017661525752316e-05,
+      "loss": 0.8384,
+      "step": 7632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.801502731316643e-05,
+      "loss": 0.4558,
+      "step": 7633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8012393003677615e-05,
+      "loss": 1.0332,
+      "step": 7634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8009758597370545e-05,
+      "loss": 0.257,
+      "step": 7635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.800712409432988e-05,
+      "loss": 0.3156,
+      "step": 7636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8004489494640302e-05,
+      "loss": 0.7758,
+      "step": 7637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8001854798386482e-05,
+      "loss": 1.07,
+      "step": 7638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7999220005653095e-05,
+      "loss": 0.7352,
+      "step": 7639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7996585116524827e-05,
+      "loss": 0.6881,
+      "step": 7640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7993950131086362e-05,
+      "loss": 1.1435,
+      "step": 7641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7991315049422393e-05,
+      "loss": 1.12,
+      "step": 7642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.798867987161761e-05,
+      "loss": 0.2702,
+      "step": 7643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.79860445977567e-05,
+      "loss": 0.4785,
+      "step": 7644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.798340922792437e-05,
+      "loss": 0.0885,
+      "step": 7645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.798077376220532e-05,
+      "loss": 0.7679,
+      "step": 7646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7978138200684247e-05,
+      "loss": 0.6831,
+      "step": 7647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.797550254344586e-05,
+      "loss": 0.0992,
+      "step": 7648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7972866790574877e-05,
+      "loss": 0.3458,
+      "step": 7649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7970230942156e-05,
+      "loss": 0.2271,
+      "step": 7650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7967594998273954e-05,
+      "loss": 0.4803,
+      "step": 7651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.796495895901345e-05,
+      "loss": 0.0411,
+      "step": 7652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7962322824459217e-05,
+      "loss": 0.1919,
+      "step": 7653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.795968659469598e-05,
+      "loss": 1.1799,
+      "step": 7654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.795705026980847e-05,
+      "loss": 0.0314,
+      "step": 7655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7954413849881398e-05,
+      "loss": 1.3884,
+      "step": 7656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7951777334999525e-05,
+      "loss": 0.3873,
+      "step": 7657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7949140725247573e-05,
+      "loss": 1.0414,
+      "step": 7658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.794650402071029e-05,
+      "loss": 1.2617,
+      "step": 7659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7943867221472423e-05,
+      "loss": 0.3928,
+      "step": 7660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7941230327618706e-05,
+      "loss": 0.6607,
+      "step": 7661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7938593339233896e-05,
+      "loss": 1.0995,
+      "step": 7662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7935956256402747e-05,
+      "loss": 1.4483,
+      "step": 7663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7933319079210003e-05,
+      "loss": 1.1327,
+      "step": 7664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7930681807740446e-05,
+      "loss": 0.0292,
+      "step": 7665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.792804444207882e-05,
+      "loss": 0.3638,
+      "step": 7666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7925406982309895e-05,
+      "loss": 0.8601,
+      "step": 7667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7922769428518436e-05,
+      "loss": 1.7399,
+      "step": 7668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7920131780789224e-05,
+      "loss": 0.3022,
+      "step": 7669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.791749403920702e-05,
+      "loss": 0.325,
+      "step": 7670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.79148562038566e-05,
+      "loss": 0.5506,
+      "step": 7671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7912218274822757e-05,
+      "loss": 0.3436,
+      "step": 7672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7909580252190268e-05,
+      "loss": 0.6162,
+      "step": 7673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7906942136043918e-05,
+      "loss": 0.5595,
+      "step": 7674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7904303926468492e-05,
+      "loss": 0.1715,
+      "step": 7675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7901665623548792e-05,
+      "loss": 1.0229,
+      "step": 7676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7899027227369603e-05,
+      "loss": 0.2376,
+      "step": 7677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7896388738015732e-05,
+      "loss": 0.0574,
+      "step": 7678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7893750155571968e-05,
+      "loss": 0.3875,
+      "step": 7679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.789111148012313e-05,
+      "loss": 0.0687,
+      "step": 7680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7888472711754013e-05,
+      "loss": 0.8778,
+      "step": 7681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7885833850549436e-05,
+      "loss": 0.1883,
+      "step": 7682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7883194896594202e-05,
+      "loss": 1.0937,
+      "step": 7683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7880555849973134e-05,
+      "loss": 0.0897,
+      "step": 7684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.787791671077105e-05,
+      "loss": 0.2937,
+      "step": 7685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7875277479072774e-05,
+      "loss": 0.7325,
+      "step": 7686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7872638154963115e-05,
+      "loss": 0.3762,
+      "step": 7687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7869998738526923e-05,
+      "loss": 1.132,
+      "step": 7688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.786735922984902e-05,
+      "loss": 0.6741,
+      "step": 7689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7864719629014236e-05,
+      "loss": 1.2945,
+      "step": 7690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.786207993610742e-05,
+      "loss": 1.1808,
+      "step": 7691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7859440151213396e-05,
+      "loss": 0.7401,
+      "step": 7692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.785680027441701e-05,
+      "loss": 1.5176,
+      "step": 7693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7854160305803112e-05,
+      "loss": 0.7881,
+      "step": 7694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.785152024545655e-05,
+      "loss": 0.3284,
+      "step": 7695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.784888009346218e-05,
+      "loss": 0.6308,
+      "step": 7696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.784623984990485e-05,
+      "loss": 0.9868,
+      "step": 7697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.784359951486941e-05,
+      "loss": 0.4151,
+      "step": 7698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7840959088440738e-05,
+      "loss": 0.5659,
+      "step": 7699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7838318570703692e-05,
+      "loss": 0.8401,
+      "step": 7700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7835677961743126e-05,
+      "loss": 0.0299,
+      "step": 7701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7833037261643915e-05,
+      "loss": 0.7803,
+      "step": 7702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.783039647049094e-05,
+      "loss": 1.4718,
+      "step": 7703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7827755588369068e-05,
+      "loss": 0.1636,
+      "step": 7704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7825114615363183e-05,
+      "loss": 1.037,
+      "step": 7705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7822473551558155e-05,
+      "loss": 0.1527,
+      "step": 7706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.781983239703888e-05,
+      "loss": 0.5815,
+      "step": 7707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7817191151890234e-05,
+      "loss": 0.722,
+      "step": 7708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7814549816197114e-05,
+      "loss": 1.4749,
+      "step": 7709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7811908390044408e-05,
+      "loss": 0.2445,
+      "step": 7710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7809266873517014e-05,
+      "loss": 0.2917,
+      "step": 7711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.780662526669983e-05,
+      "loss": 0.137,
+      "step": 7712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7803983569677756e-05,
+      "loss": 0.3916,
+      "step": 7713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7801341782535702e-05,
+      "loss": 0.4741,
+      "step": 7714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7798699905358565e-05,
+      "loss": 1.0686,
+      "step": 7715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.779605793823126e-05,
+      "loss": 0.5449,
+      "step": 7716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7793415881238695e-05,
+      "loss": 0.1356,
+      "step": 7717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7790773734465798e-05,
+      "loss": 0.2532,
+      "step": 7718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7788131497997478e-05,
+      "loss": 0.4652,
+      "step": 7719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.778548917191866e-05,
+      "loss": 0.0689,
+      "step": 7720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.778284675631426e-05,
+      "loss": 0.4933,
+      "step": 7721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7780204251269214e-05,
+      "loss": 0.2252,
+      "step": 7722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7777561656868454e-05,
+      "loss": 0.6316,
+      "step": 7723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7774918973196906e-05,
+      "loss": 0.6636,
+      "step": 7724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7772276200339508e-05,
+      "loss": 0.266,
+      "step": 7725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.77696333383812e-05,
+      "loss": 0.9654,
+      "step": 7726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.776699038740692e-05,
+      "loss": 0.5322,
+      "step": 7727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.776434734750162e-05,
+      "loss": 0.323,
+      "step": 7728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7761704218750244e-05,
+      "loss": 1.2487,
+      "step": 7729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7759061001237733e-05,
+      "loss": 0.6528,
+      "step": 7730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.775641769504905e-05,
+      "loss": 1.743,
+      "step": 7731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7753774300269156e-05,
+      "loss": 0.2138,
+      "step": 7732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.775113081698299e-05,
+      "loss": 0.5051,
+      "step": 7733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.774848724527553e-05,
+      "loss": 0.196,
+      "step": 7734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7745843585231736e-05,
+      "loss": 0.4718,
+      "step": 7735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7743199836936576e-05,
+      "loss": 0.8658,
+      "step": 7736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7740556000475027e-05,
+      "loss": 0.3105,
+      "step": 7737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7737912075932043e-05,
+      "loss": 0.683,
+      "step": 7738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7735268063392612e-05,
+      "loss": 0.2014,
+      "step": 7739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7732623962941715e-05,
+      "loss": 1.0586,
+      "step": 7740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.772997977466432e-05,
+      "loss": 0.1769,
+      "step": 7741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.772733549864543e-05,
+      "loss": 1.1405,
+      "step": 7742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.772469113497002e-05,
+      "loss": 0.3777,
+      "step": 7743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7722046683723084e-05,
+      "loss": 1.6499,
+      "step": 7744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7719402144989612e-05,
+      "loss": 1.0749,
+      "step": 7745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7716757518854604e-05,
+      "loss": 0.4902,
+      "step": 7746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.771411280540305e-05,
+      "loss": 0.2278,
+      "step": 7747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7711468004719953e-05,
+      "loss": 0.4594,
+      "step": 7748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7708823116890325e-05,
+      "loss": 0.6718,
+      "step": 7749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7706178141999165e-05,
+      "loss": 0.8349,
+      "step": 7750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.770353308013149e-05,
+      "loss": 0.124,
+      "step": 7751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7700887931372302e-05,
+      "loss": 0.2948,
+      "step": 7752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.769824269580662e-05,
+      "loss": 1.8417,
+      "step": 7753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7695597373519465e-05,
+      "loss": 1.7777,
+      "step": 7754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7692951964595856e-05,
+      "loss": 0.9456,
+      "step": 7755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.769030646912081e-05,
+      "loss": 0.292,
+      "step": 7756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7687660887179365e-05,
+      "loss": 1.5942,
+      "step": 7757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7685015218856544e-05,
+      "loss": 1.4735,
+      "step": 7758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.768236946423738e-05,
+      "loss": 0.0597,
+      "step": 7759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.76797236234069e-05,
+      "loss": 0.2951,
+      "step": 7760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.767707769645015e-05,
+      "loss": 0.6652,
+      "step": 7761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.767443168345217e-05,
+      "loss": 0.3508,
+      "step": 7762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7671785584498004e-05,
+      "loss": 0.83,
+      "step": 7763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7669139399672686e-05,
+      "loss": 0.1149,
+      "step": 7764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.766649312906127e-05,
+      "loss": 0.6595,
+      "step": 7765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7663846772748816e-05,
+      "loss": 0.2864,
+      "step": 7766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.766120033082037e-05,
+      "loss": 0.0587,
+      "step": 7767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7658553803360994e-05,
+      "loss": 0.4674,
+      "step": 7768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7655907190455738e-05,
+      "loss": 0.5271,
+      "step": 7769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.765326049218967e-05,
+      "loss": 1.0089,
+      "step": 7770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.765061370864785e-05,
+      "loss": 0.32,
+      "step": 7771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7647966839915355e-05,
+      "loss": 0.6743,
+      "step": 7772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.764531988607725e-05,
+      "loss": 0.5372,
+      "step": 7773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7642672847218606e-05,
+      "loss": 0.5246,
+      "step": 7774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7640025723424503e-05,
+      "loss": 0.1614,
+      "step": 7775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7637378514780016e-05,
+      "loss": 0.5694,
+      "step": 7776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7634731221370232e-05,
+      "loss": 0.8074,
+      "step": 7777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7632083843280222e-05,
+      "loss": 0.6891,
+      "step": 7778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7629436380595084e-05,
+      "loss": 0.4738,
+      "step": 7779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7626788833399904e-05,
+      "loss": 0.0837,
+      "step": 7780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7624141201779777e-05,
+      "loss": 0.5469,
+      "step": 7781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.76214934858198e-05,
+      "loss": 1.035,
+      "step": 7782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.761884568560506e-05,
+      "loss": 1.2237,
+      "step": 7783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7616197801220666e-05,
+      "loss": 0.2427,
+      "step": 7784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7613549832751715e-05,
+      "loss": 0.0945,
+      "step": 7785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7610901780283323e-05,
+      "loss": 0.3351,
+      "step": 7786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7608253643900582e-05,
+      "loss": 0.6534,
+      "step": 7787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7605605423688614e-05,
+      "loss": 0.6653,
+      "step": 7788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7602957119732535e-05,
+      "loss": 0.1036,
+      "step": 7789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7600308732117458e-05,
+      "loss": 0.6771,
+      "step": 7790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7597660260928506e-05,
+      "loss": 0.5481,
+      "step": 7791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.759501170625079e-05,
+      "loss": 0.3846,
+      "step": 7792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7592363068169442e-05,
+      "loss": 0.427,
+      "step": 7793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7589714346769587e-05,
+      "loss": 1.0091,
+      "step": 7794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7587065542136357e-05,
+      "loss": 1.2051,
+      "step": 7795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7584416654354888e-05,
+      "loss": 0.6938,
+      "step": 7796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7581767683510307e-05,
+      "loss": 0.8496,
+      "step": 7797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7579118629687757e-05,
+      "loss": 0.8953,
+      "step": 7798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7576469492972374e-05,
+      "loss": 1.0845,
+      "step": 7799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7573820273449312e-05,
+      "loss": 0.6473,
+      "step": 7800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7571170971203705e-05,
+      "loss": 1.209,
+      "step": 7801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7568521586320703e-05,
+      "loss": 0.359,
+      "step": 7802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.756587211888546e-05,
+      "loss": 1.3539,
+      "step": 7803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7563222568983132e-05,
+      "loss": 0.3979,
+      "step": 7804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7560572936698875e-05,
+      "loss": 0.1579,
+      "step": 7805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7557923222117846e-05,
+      "loss": 0.5156,
+      "step": 7806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7555273425325208e-05,
+      "loss": 0.1645,
+      "step": 7807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7552623546406126e-05,
+      "loss": 0.2346,
+      "step": 7808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7549973585445767e-05,
+      "loss": 1.0095,
+      "step": 7809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7547323542529292e-05,
+      "loss": 0.6282,
+      "step": 7810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7544673417741882e-05,
+      "loss": 0.2115,
+      "step": 7811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7542023211168714e-05,
+      "loss": 0.2308,
+      "step": 7812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7539372922894963e-05,
+      "loss": 1.505,
+      "step": 7813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7536722553005813e-05,
+      "loss": 0.6029,
+      "step": 7814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7534072101586436e-05,
+      "loss": 0.5536,
+      "step": 7815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7531421568722024e-05,
+      "loss": 0.5759,
+      "step": 7816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7528770954497764e-05,
+      "loss": 0.1289,
+      "step": 7817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7526120258998853e-05,
+      "loss": 1.408,
+      "step": 7818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7523469482310478e-05,
+      "loss": 0.9034,
+      "step": 7819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7520818624517832e-05,
+      "loss": 0.287,
+      "step": 7820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.751816768570612e-05,
+      "loss": 0.114,
+      "step": 7821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.751551666596054e-05,
+      "loss": 0.1484,
+      "step": 7822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.75128655653663e-05,
+      "loss": 0.6039,
+      "step": 7823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.75102143840086e-05,
+      "loss": 0.5859,
+      "step": 7824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.750756312197265e-05,
+      "loss": 0.7321,
+      "step": 7825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7504911779343664e-05,
+      "loss": 0.2637,
+      "step": 7826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7502260356206855e-05,
+      "loss": 1.5765,
+      "step": 7827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7499608852647444e-05,
+      "loss": 1.2735,
+      "step": 7828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.749695726875064e-05,
+      "loss": 0.3083,
+      "step": 7829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7494305604601672e-05,
+      "loss": 0.3337,
+      "step": 7830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7491653860285772e-05,
+      "loss": 0.835,
+      "step": 7831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.748900203588815e-05,
+      "loss": 1.6156,
+      "step": 7832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7486350131494044e-05,
+      "loss": 1.4477,
+      "step": 7833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7483698147188686e-05,
+      "loss": 1.3435,
+      "step": 7834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7481046083057317e-05,
+      "loss": 0.3424,
+      "step": 7835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7478393939185168e-05,
+      "loss": 0.7891,
+      "step": 7836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7475741715657475e-05,
+      "loss": 0.6768,
+      "step": 7837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7473089412559483e-05,
+      "loss": 0.2485,
+      "step": 7838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.747043702997644e-05,
+      "loss": 0.2657,
+      "step": 7839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7467784567993598e-05,
+      "loss": 0.112,
+      "step": 7840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7465132026696196e-05,
+      "loss": 0.684,
+      "step": 7841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7462479406169493e-05,
+      "loss": 0.5831,
+      "step": 7842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.745982670649874e-05,
+      "loss": 0.8269,
+      "step": 7843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.74571739277692e-05,
+      "loss": 0.6496,
+      "step": 7844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7454521070066136e-05,
+      "loss": 0.8062,
+      "step": 7845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.74518681334748e-05,
+      "loss": 0.1303,
+      "step": 7846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7449215118080462e-05,
+      "loss": 0.6989,
+      "step": 7847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.744656202396839e-05,
+      "loss": 0.779,
+      "step": 7848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.744390885122386e-05,
+      "loss": 0.7419,
+      "step": 7849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.744125559993214e-05,
+      "loss": 0.459,
+      "step": 7850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7438602270178504e-05,
+      "loss": 0.0332,
+      "step": 7851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.743594886204823e-05,
+      "loss": 0.0296,
+      "step": 7852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7433295375626607e-05,
+      "loss": 0.1017,
+      "step": 7853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.743064181099891e-05,
+      "loss": 1.3747,
+      "step": 7854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.742798816825042e-05,
+      "loss": 0.0413,
+      "step": 7855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7425334447466432e-05,
+      "loss": 0.348,
+      "step": 7856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7422680648732238e-05,
+      "loss": 0.7438,
+      "step": 7857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7420026772133124e-05,
+      "loss": 0.1723,
+      "step": 7858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.74173728177544e-05,
+      "loss": 0.3115,
+      "step": 7859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7414718785681345e-05,
+      "loss": 0.3394,
+      "step": 7860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7412064675999272e-05,
+      "loss": 0.2577,
+      "step": 7861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7409410488793478e-05,
+      "loss": 2.2946,
+      "step": 7862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7406756224149276e-05,
+      "loss": 0.7767,
+      "step": 7863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7404101882151964e-05,
+      "loss": 1.9501,
+      "step": 7864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7401447462886857e-05,
+      "loss": 1.3112,
+      "step": 7865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.739879296643927e-05,
+      "loss": 0.7727,
+      "step": 7866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.739613839289452e-05,
+      "loss": 0.4438,
+      "step": 7867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7393483742337922e-05,
+      "loss": 0.3787,
+      "step": 7868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7390829014854795e-05,
+      "loss": 0.1967,
+      "step": 7869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7388174210530465e-05,
+      "loss": 0.8166,
+      "step": 7870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7385519329450246e-05,
+      "loss": 0.7127,
+      "step": 7871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7382864371699488e-05,
+      "loss": 1.1422,
+      "step": 7872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7380209337363504e-05,
+      "loss": 0.1041,
+      "step": 7873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7377554226527634e-05,
+      "loss": 0.5444,
+      "step": 7874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.737489903927721e-05,
+      "loss": 0.664,
+      "step": 7875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7372243775697568e-05,
+      "loss": 0.6361,
+      "step": 7876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7369588435874055e-05,
+      "loss": 0.1344,
+      "step": 7877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7366933019892005e-05,
+      "loss": 0.0748,
+      "step": 7878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.736427752783677e-05,
+      "loss": 0.5314,
+      "step": 7879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.736162195979369e-05,
+      "loss": 0.1305,
+      "step": 7880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.735896631584812e-05,
+      "loss": 0.417,
+      "step": 7881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7356310596085422e-05,
+      "loss": 0.4911,
+      "step": 7882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7353654800590934e-05,
+      "loss": 0.394,
+      "step": 7883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.735099892945002e-05,
+      "loss": 0.7411,
+      "step": 7884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.734834298274804e-05,
+      "loss": 0.8265,
+      "step": 7885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.734568696057036e-05,
+      "loss": 0.6446,
+      "step": 7886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.734303086300233e-05,
+      "loss": 1.0198,
+      "step": 7887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7340374690129333e-05,
+      "loss": 0.3444,
+      "step": 7888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.733771844203673e-05,
+      "loss": 0.1763,
+      "step": 7889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.73350621188099e-05,
+      "loss": 0.6147,
+      "step": 7890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.733240572053421e-05,
+      "loss": 0.8837,
+      "step": 7891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7329749247295042e-05,
+      "loss": 0.3679,
+      "step": 7892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7327092699177766e-05,
+      "loss": 0.1352,
+      "step": 7893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7324436076267773e-05,
+      "loss": 0.3943,
+      "step": 7894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7321779378650443e-05,
+      "loss": 0.9034,
+      "step": 7895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7319122606411167e-05,
+      "loss": 0.5425,
+      "step": 7896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7316465759635326e-05,
+      "loss": 0.2348,
+      "step": 7897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.731380883840831e-05,
+      "loss": 1.2428,
+      "step": 7898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.731115184281552e-05,
+      "loss": 0.3452,
+      "step": 7899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7308494772942354e-05,
+      "loss": 0.538,
+      "step": 7900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7305837628874195e-05,
+      "loss": 1.2665,
+      "step": 7901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7303180410696456e-05,
+      "loss": 0.2494,
+      "step": 7902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7300523118494537e-05,
+      "loss": 0.1425,
+      "step": 7903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7297865752353844e-05,
+      "loss": 0.2517,
+      "step": 7904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7295208312359786e-05,
+      "loss": 1.4272,
+      "step": 7905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7292550798597768e-05,
+      "loss": 0.0401,
+      "step": 7906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.728989321115321e-05,
+      "loss": 1.0853,
+      "step": 7907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7287235550111523e-05,
+      "loss": 1.4156,
+      "step": 7908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.728457781555812e-05,
+      "loss": 0.1131,
+      "step": 7909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7281920007578423e-05,
+      "loss": 0.5985,
+      "step": 7910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7279262126257855e-05,
+      "loss": 1.2728,
+      "step": 7911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.727660417168184e-05,
+      "loss": 0.6964,
+      "step": 7912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.727394614393581e-05,
+      "loss": 0.1586,
+      "step": 7913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7271288043105184e-05,
+      "loss": 0.1388,
+      "step": 7914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.72686298692754e-05,
+      "loss": 1.2936,
+      "step": 7915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7265971622531894e-05,
+      "loss": 0.0769,
+      "step": 7916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7263313302960087e-05,
+      "loss": 0.2282,
+      "step": 7917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.726065491064544e-05,
+      "loss": 1.1684,
+      "step": 7918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7257996445673374e-05,
+      "loss": 0.049,
+      "step": 7919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7255337908129348e-05,
+      "loss": 0.0717,
+      "step": 7920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.725267929809879e-05,
+      "loss": 0.0657,
+      "step": 7921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7250020615667165e-05,
+      "loss": 0.1534,
+      "step": 7922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.724736186091991e-05,
+      "loss": 0.1086,
+      "step": 7923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7244703033942486e-05,
+      "loss": 1.7635,
+      "step": 7924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.724204413482034e-05,
+      "loss": 1.1333,
+      "step": 7925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.723938516363893e-05,
+      "loss": 0.7631,
+      "step": 7926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7236726120483726e-05,
+      "loss": 0.8355,
+      "step": 7927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7234067005440182e-05,
+      "loss": 0.1898,
+      "step": 7928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7231407818593757e-05,
+      "loss": 0.3542,
+      "step": 7929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7228748560029922e-05,
+      "loss": 0.5188,
+      "step": 7930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.722608922983415e-05,
+      "loss": 0.4869,
+      "step": 7931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7223429828091906e-05,
+      "loss": 0.4645,
+      "step": 7932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7220770354888655e-05,
+      "loss": 1.5387,
+      "step": 7933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.721811081030989e-05,
+      "loss": 1.1287,
+      "step": 7934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7215451194441083e-05,
+      "loss": 0.8053,
+      "step": 7935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7212791507367712e-05,
+      "loss": 0.4419,
+      "step": 7936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7210131749175252e-05,
+      "loss": 0.9283,
+      "step": 7937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.72074719199492e-05,
+      "loss": 0.657,
+      "step": 7938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7204812019775036e-05,
+      "loss": 1.0167,
+      "step": 7939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7202152048738243e-05,
+      "loss": 0.8912,
+      "step": 7940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7199492006924333e-05,
+      "loss": 1.6426,
+      "step": 7941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7196831894418778e-05,
+      "loss": 1.2162,
+      "step": 7942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7194171711307084e-05,
+      "loss": 0.7632,
+      "step": 7943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.719151145767475e-05,
+      "loss": 0.3337,
+      "step": 7944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.718885113360728e-05,
+      "loss": 0.7087,
+      "step": 7945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7186190739190163e-05,
+      "loss": 0.6183,
+      "step": 7946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7183530274508912e-05,
+      "loss": 1.0143,
+      "step": 7947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7180869739649033e-05,
+      "loss": 0.5997,
+      "step": 7948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7178209134696042e-05,
+      "loss": 0.9727,
+      "step": 7949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7175548459735446e-05,
+      "loss": 0.2465,
+      "step": 7950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.717288771485276e-05,
+      "loss": 1.1988,
+      "step": 7951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7170226900133488e-05,
+      "loss": 1.3562,
+      "step": 7952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7167566015663172e-05,
+      "loss": 0.0354,
+      "step": 7953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7164905061527317e-05,
+      "loss": 1.2832,
+      "step": 7954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.716224403781145e-05,
+      "loss": 0.0615,
+      "step": 7955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.715958294460109e-05,
+      "loss": 0.7154,
+      "step": 7956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7156921781981775e-05,
+      "loss": 1.0401,
+      "step": 7957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7154260550039032e-05,
+      "loss": 0.0486,
+      "step": 7958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.715159924885839e-05,
+      "loss": 0.0447,
+      "step": 7959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.714893787852538e-05,
+      "loss": 1.4604,
+      "step": 7960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7146276439125546e-05,
+      "loss": 0.2825,
+      "step": 7961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7143614930744422e-05,
+      "loss": 1.187,
+      "step": 7962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7140953353467556e-05,
+      "loss": 0.1135,
+      "step": 7963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7138291707380474e-05,
+      "loss": 0.3819,
+      "step": 7964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7135629992568734e-05,
+      "loss": 0.4279,
+      "step": 7965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7132968209117885e-05,
+      "loss": 0.406,
+      "step": 7966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7130306357113472e-05,
+      "loss": 0.571,
+      "step": 7967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7127644436641055e-05,
+      "loss": 0.0769,
+      "step": 7968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7124982447786177e-05,
+      "loss": 0.1228,
+      "step": 7969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7122320390634395e-05,
+      "loss": 0.2936,
+      "step": 7970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.711965826527127e-05,
+      "loss": 0.2271,
+      "step": 7971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.711699607178237e-05,
+      "loss": 0.3483,
+      "step": 7972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.711433381025325e-05,
+      "loss": 0.3445,
+      "step": 7973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7111671480769475e-05,
+      "loss": 0.0823,
+      "step": 7974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7109009083416614e-05,
+      "loss": 0.8415,
+      "step": 7975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7106346618280236e-05,
+      "loss": 0.7483,
+      "step": 7976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.710368408544592e-05,
+      "loss": 1.4825,
+      "step": 7977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7101021484999226e-05,
+      "loss": 0.4278,
+      "step": 7978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.709835881702573e-05,
+      "loss": 0.5843,
+      "step": 7979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7095696081611023e-05,
+      "loss": 0.3216,
+      "step": 7980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7093033278840686e-05,
+      "loss": 1.5935,
+      "step": 7981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7090370408800285e-05,
+      "loss": 1.4891,
+      "step": 7982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.708770747157542e-05,
+      "loss": 0.4437,
+      "step": 7983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7085044467251668e-05,
+      "loss": 0.8758,
+      "step": 7984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7082381395914623e-05,
+      "loss": 0.618,
+      "step": 7985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7079718257649873e-05,
+      "loss": 0.2034,
+      "step": 7986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7077055052543012e-05,
+      "loss": 0.6401,
+      "step": 7987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7074391780679638e-05,
+      "loss": 0.4512,
+      "step": 7988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.707172844214535e-05,
+      "loss": 0.3565,
+      "step": 7989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.706906503702574e-05,
+      "loss": 0.2808,
+      "step": 7990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7066401565406418e-05,
+      "loss": 0.3476,
+      "step": 7991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7063738027372982e-05,
+      "loss": 0.1343,
+      "step": 7992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.706107442301104e-05,
+      "loss": 0.2958,
+      "step": 7993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7058410752406194e-05,
+      "loss": 0.1837,
+      "step": 7994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7055747015644074e-05,
+      "loss": 0.4088,
+      "step": 7995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.705308321281027e-05,
+      "loss": 0.9112,
+      "step": 7996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.705041934399041e-05,
+      "loss": 0.8382,
+      "step": 7997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7047755409270104e-05,
+      "loss": 0.4969,
+      "step": 7998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.704509140873498e-05,
+      "loss": 0.5031,
+      "step": 7999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7042427342470645e-05,
+      "loss": 0.6092,
+      "step": 8000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7039763210562732e-05,
+      "loss": 0.0353,
+      "step": 8001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7037099013096858e-05,
+      "loss": 0.0419,
+      "step": 8002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.703443475015866e-05,
+      "loss": 0.036,
+      "step": 8003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7031770421833764e-05,
+      "loss": 0.0399,
+      "step": 8004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7029106028207802e-05,
+      "loss": 0.1329,
+      "step": 8005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7026441569366405e-05,
+      "loss": 0.1567,
+      "step": 8006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7023777045395208e-05,
+      "loss": 0.7979,
+      "step": 8007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7021112456379856e-05,
+      "loss": 0.0484,
+      "step": 8008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7018447802405978e-05,
+      "loss": 0.4618,
+      "step": 8009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7015783083559218e-05,
+      "loss": 0.3502,
+      "step": 8010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7013118299925226e-05,
+      "loss": 0.486,
+      "step": 8011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.701045345158965e-05,
+      "loss": 1.0187,
+      "step": 8012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.700778853863813e-05,
+      "loss": 0.2845,
+      "step": 8013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7005123561156323e-05,
+      "loss": 0.257,
+      "step": 8014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.700245851922988e-05,
+      "loss": 0.3061,
+      "step": 8015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6999793412944446e-05,
+      "loss": 0.2675,
+      "step": 8016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6997128242385685e-05,
+      "loss": 0.1895,
+      "step": 8017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.699446300763926e-05,
+      "loss": 1.6586,
+      "step": 8018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6991797708790832e-05,
+      "loss": 0.6774,
+      "step": 8019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.698913234592605e-05,
+      "loss": 0.2154,
+      "step": 8020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.698646691913059e-05,
+      "loss": 0.0832,
+      "step": 8021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.698380142849012e-05,
+      "loss": 0.845,
+      "step": 8022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6981135874090305e-05,
+      "loss": 0.8026,
+      "step": 8023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.697847025601681e-05,
+      "loss": 0.7983,
+      "step": 8024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6975804574355317e-05,
+      "loss": 1.2507,
+      "step": 8025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.69731388291915e-05,
+      "loss": 0.6426,
+      "step": 8026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6970473020611035e-05,
+      "loss": 1.2624,
+      "step": 8027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6967807148699598e-05,
+      "loss": 0.1395,
+      "step": 8028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.696514121354287e-05,
+      "loss": 0.8578,
+      "step": 8029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.696247521522654e-05,
+      "loss": 0.1819,
+      "step": 8030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.695980915383629e-05,
+      "loss": 0.7677,
+      "step": 8031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6957143029457802e-05,
+      "loss": 0.6556,
+      "step": 8032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6954476842176768e-05,
+      "loss": 0.8184,
+      "step": 8033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.695181059207889e-05,
+      "loss": 0.4093,
+      "step": 8034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6949144279249843e-05,
+      "loss": 0.5367,
+      "step": 8035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.694647790377534e-05,
+      "loss": 1.0768,
+      "step": 8036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6943811465741065e-05,
+      "loss": 1.7715,
+      "step": 8037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6941144965232722e-05,
+      "loss": 0.1071,
+      "step": 8038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.693847840233601e-05,
+      "loss": 0.145,
+      "step": 8039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6935811777136634e-05,
+      "loss": 0.6579,
+      "step": 8040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6933145089720307e-05,
+      "loss": 0.5898,
+      "step": 8041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6930478340172724e-05,
+      "loss": 0.9757,
+      "step": 8042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.69278115285796e-05,
+      "loss": 0.5856,
+      "step": 8043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6925144655026647e-05,
+      "loss": 0.6901,
+      "step": 8044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6922477719599578e-05,
+      "loss": 0.3892,
+      "step": 8045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.69198107223841e-05,
+      "loss": 0.3999,
+      "step": 8046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6917143663465948e-05,
+      "loss": 0.1354,
+      "step": 8047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.691447654293082e-05,
+      "loss": 0.5948,
+      "step": 8048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6911809360864453e-05,
+      "loss": 0.0799,
+      "step": 8049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6909142117352565e-05,
+      "loss": 0.2943,
+      "step": 8050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6906474812480885e-05,
+      "loss": 0.1192,
+      "step": 8051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6903807446335133e-05,
+      "loss": 0.3558,
+      "step": 8052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.690114001900104e-05,
+      "loss": 0.0496,
+      "step": 8053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6898472530564343e-05,
+      "loss": 1.4672,
+      "step": 8054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6895804981110764e-05,
+      "loss": 0.1066,
+      "step": 8055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6893137370726047e-05,
+      "loss": 0.0376,
+      "step": 8056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6890469699495924e-05,
+      "loss": 0.0453,
+      "step": 8057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6887801967506142e-05,
+      "loss": 0.7611,
+      "step": 8058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6885134174842435e-05,
+      "loss": 1.3186,
+      "step": 8059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6882466321590546e-05,
+      "loss": 0.9975,
+      "step": 8060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6879798407836222e-05,
+      "loss": 0.9208,
+      "step": 8061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.687713043366521e-05,
+      "loss": 0.1419,
+      "step": 8062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.687446239916325e-05,
+      "loss": 0.4284,
+      "step": 8063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6871794304416106e-05,
+      "loss": 0.118,
+      "step": 8064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6869126149509523e-05,
+      "loss": 0.338,
+      "step": 8065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6866457934529256e-05,
+      "loss": 0.6342,
+      "step": 8066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6863789659561065e-05,
+      "loss": 0.0945,
+      "step": 8067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.68611213246907e-05,
+      "loss": 0.4136,
+      "step": 8068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6858452930003934e-05,
+      "loss": 0.0399,
+      "step": 8069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6855784475586518e-05,
+      "loss": 1.7892,
+      "step": 8070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6853115961524216e-05,
+      "loss": 0.9191,
+      "step": 8071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6850447387902803e-05,
+      "loss": 1.5571,
+      "step": 8072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6847778754808042e-05,
+      "loss": 0.7252,
+      "step": 8073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.68451100623257e-05,
+      "loss": 0.1217,
+      "step": 8074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6842441310541556e-05,
+      "loss": 0.5722,
+      "step": 8075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6839772499541375e-05,
+      "loss": 0.2086,
+      "step": 8076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.683710362941094e-05,
+      "loss": 0.848,
+      "step": 8077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.683443470023602e-05,
+      "loss": 0.3679,
+      "step": 8078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.68317657121024e-05,
+      "loss": 0.0915,
+      "step": 8079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.682909666509586e-05,
+      "loss": 1.2835,
+      "step": 8080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6826427559302193e-05,
+      "loss": 0.2038,
+      "step": 8081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6823758394807168e-05,
+      "loss": 0.821,
+      "step": 8082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6821089171696577e-05,
+      "loss": 1.3936,
+      "step": 8083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.681841989005621e-05,
+      "loss": 0.7921,
+      "step": 8084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6815750549971865e-05,
+      "loss": 1.6422,
+      "step": 8085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.681308115152932e-05,
+      "loss": 1.2751,
+      "step": 8086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.681041169481438e-05,
+      "loss": 1.0194,
+      "step": 8087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6807742179912834e-05,
+      "loss": 0.4976,
+      "step": 8088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6805072606910494e-05,
+      "loss": 0.1314,
+      "step": 8089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6802402975893146e-05,
+      "loss": 0.304,
+      "step": 8090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6799733286946595e-05,
+      "loss": 1.015,
+      "step": 8091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6797063540156645e-05,
+      "loss": 1.247,
+      "step": 8092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6794393735609104e-05,
+      "loss": 0.6336,
+      "step": 8093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.679172387338978e-05,
+      "loss": 0.5867,
+      "step": 8094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.678905395358448e-05,
+      "loss": 0.5349,
+      "step": 8095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6786383976279017e-05,
+      "loss": 0.4149,
+      "step": 8096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.67837139415592e-05,
+      "loss": 0.4151,
+      "step": 8097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6781043849510846e-05,
+      "loss": 0.6453,
+      "step": 8098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.677837370021978e-05,
+      "loss": 0.9731,
+      "step": 8099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6775703493771806e-05,
+      "loss": 0.5854,
+      "step": 8100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6773033230252754e-05,
+      "loss": 1.3867,
+      "step": 8101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6770362909748436e-05,
+      "loss": 1.3818,
+      "step": 8102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6767692532344693e-05,
+      "loss": 0.0398,
+      "step": 8103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.676502209812734e-05,
+      "loss": 1.5719,
+      "step": 8104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6762351607182205e-05,
+      "loss": 1.343,
+      "step": 8105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.675968105959512e-05,
+      "loss": 1.6382,
+      "step": 8106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6757010455451916e-05,
+      "loss": 1.9072,
+      "step": 8107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6754339794838425e-05,
+      "loss": 0.456,
+      "step": 8108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.675166907784048e-05,
+      "loss": 1.2762,
+      "step": 8109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6748998304543916e-05,
+      "loss": 0.0632,
+      "step": 8110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.674632747503458e-05,
+      "loss": 0.9288,
+      "step": 8111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6743656589398308e-05,
+      "loss": 0.2292,
+      "step": 8112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6740985647720948e-05,
+      "loss": 0.2937,
+      "step": 8113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6738314650088334e-05,
+      "loss": 0.7566,
+      "step": 8114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6735643596586315e-05,
+      "loss": 0.0558,
+      "step": 8115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.673297248730074e-05,
+      "loss": 1.1451,
+      "step": 8116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6730301322317454e-05,
+      "loss": 1.3432,
+      "step": 8117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.672763010172232e-05,
+      "loss": 0.1575,
+      "step": 8118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.672495882560118e-05,
+      "loss": 0.1216,
+      "step": 8119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6722287494039893e-05,
+      "loss": 0.3453,
+      "step": 8120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6719616107124313e-05,
+      "loss": 0.1776,
+      "step": 8121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6716944664940304e-05,
+      "loss": 1.0998,
+      "step": 8122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6714273167573715e-05,
+      "loss": 0.0833,
+      "step": 8123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.671160161511042e-05,
+      "loss": 1.6576,
+      "step": 8124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6708930007636274e-05,
+      "loss": 0.4246,
+      "step": 8125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.670625834523715e-05,
+      "loss": 0.1465,
+      "step": 8126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6703586627998913e-05,
+      "loss": 0.4605,
+      "step": 8127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6700914856007427e-05,
+      "loss": 1.055,
+      "step": 8128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6698243029348567e-05,
+      "loss": 0.0514,
+      "step": 8129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6695571148108204e-05,
+      "loss": 1.7501,
+      "step": 8130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6692899212372215e-05,
+      "loss": 0.1074,
+      "step": 8131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.669022722222647e-05,
+      "loss": 0.1718,
+      "step": 8132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6687555177756846e-05,
+      "loss": 0.1647,
+      "step": 8133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6684883079049235e-05,
+      "loss": 0.2021,
+      "step": 8134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.668221092618951e-05,
+      "loss": 0.217,
+      "step": 8135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6679538719263552e-05,
+      "loss": 1.3278,
+      "step": 8136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6676866458357246e-05,
+      "loss": 1.0595,
+      "step": 8137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.667419414355648e-05,
+      "loss": 0.9908,
+      "step": 8138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6671521774947145e-05,
+      "loss": 0.4956,
+      "step": 8139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6668849352615123e-05,
+      "loss": 0.3096,
+      "step": 8140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6666176876646314e-05,
+      "loss": 0.7713,
+      "step": 8141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6663504347126606e-05,
+      "loss": 0.9782,
+      "step": 8142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6660831764141896e-05,
+      "loss": 0.1797,
+      "step": 8143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665815912777809e-05,
+      "loss": 0.3465,
+      "step": 8144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6655486438121068e-05,
+      "loss": 0.2631,
+      "step": 8145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665281369525674e-05,
+      "loss": 1.6994,
+      "step": 8146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665014089927101e-05,
+      "loss": 0.8161,
+      "step": 8147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6647468050249774e-05,
+      "loss": 0.6133,
+      "step": 8148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664479514827895e-05,
+      "loss": 0.629,
+      "step": 8149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6642122193444434e-05,
+      "loss": 0.7062,
+      "step": 8150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6639449185832136e-05,
+      "loss": 0.0476,
+      "step": 8151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663677612552797e-05,
+      "loss": 0.0359,
+      "step": 8152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663410301261785e-05,
+      "loss": 1.4321,
+      "step": 8153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6631429847187684e-05,
+      "loss": 1.5582,
+      "step": 8154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6628756629323388e-05,
+      "loss": 1.4075,
+      "step": 8155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6626083359110877e-05,
+      "loss": 1.6761,
+      "step": 8156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6623410036636084e-05,
+      "loss": 1.2918,
+      "step": 8157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662073666198492e-05,
+      "loss": 1.3299,
+      "step": 8158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6618063235243297e-05,
+      "loss": 1.3477,
+      "step": 8159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6615389756497157e-05,
+      "loss": 1.3069,
+      "step": 8160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6612716225832412e-05,
+      "loss": 0.2488,
+      "step": 8161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6610042643334998e-05,
+      "loss": 0.1695,
+      "step": 8162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660736900909084e-05,
+      "loss": 0.3902,
+      "step": 8163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660469532318586e-05,
+      "loss": 1.5815,
+      "step": 8164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660202158570601e-05,
+      "loss": 0.0939,
+      "step": 8165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.659934779673721e-05,
+      "loss": 0.2131,
+      "step": 8166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.65966739563654e-05,
+      "loss": 0.995,
+      "step": 8167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6594000064676513e-05,
+      "loss": 2.0555,
+      "step": 8168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6591326121756493e-05,
+      "loss": 0.4766,
+      "step": 8169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6588652127691275e-05,
+      "loss": 0.0932,
+      "step": 8170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.65859780825668e-05,
+      "loss": 0.9188,
+      "step": 8171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6583303986469027e-05,
+      "loss": 0.1861,
+      "step": 8172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6580629839483882e-05,
+      "loss": 0.2188,
+      "step": 8173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6577955641697324e-05,
+      "loss": 0.3073,
+      "step": 8174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6575281393195298e-05,
+      "loss": 0.5236,
+      "step": 8175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6572607094063758e-05,
+      "loss": 0.5448,
+      "step": 8176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6569932744388647e-05,
+      "loss": 0.0699,
+      "step": 8177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6567258344255923e-05,
+      "loss": 2.0167,
+      "step": 8178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.656458389375154e-05,
+      "loss": 1.22,
+      "step": 8179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.656190939296146e-05,
+      "loss": 0.3757,
+      "step": 8180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.655923484197164e-05,
+      "loss": 0.3247,
+      "step": 8181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.655656024086804e-05,
+      "loss": 0.7359,
+      "step": 8182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6553885589736616e-05,
+      "loss": 0.2116,
+      "step": 8183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6551210888663335e-05,
+      "loss": 0.7145,
+      "step": 8184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6548536137734165e-05,
+      "loss": 0.751,
+      "step": 8185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6545861337035067e-05,
+      "loss": 0.6802,
+      "step": 8186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.654318648665201e-05,
+      "loss": 0.8966,
+      "step": 8187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6540511586670966e-05,
+      "loss": 0.5482,
+      "step": 8188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6537836637177903e-05,
+      "loss": 0.2391,
+      "step": 8189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6535161638258805e-05,
+      "loss": 0.1481,
+      "step": 8190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.653248658999963e-05,
+      "loss": 0.5281,
+      "step": 8191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6529811492486364e-05,
+      "loss": 0.2664,
+      "step": 8192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6527136345804983e-05,
+      "loss": 0.6233,
+      "step": 8193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.652446115004146e-05,
+      "loss": 0.7274,
+      "step": 8194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6521785905281784e-05,
+      "loss": 0.9566,
+      "step": 8195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.651911061161194e-05,
+      "loss": 0.3764,
+      "step": 8196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6516435269117898e-05,
+      "loss": 1.2817,
+      "step": 8197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6513759877885657e-05,
+      "loss": 1.1883,
+      "step": 8198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6511084438001203e-05,
+      "loss": 0.7443,
+      "step": 8199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.650840894955051e-05,
+      "loss": 0.7142,
+      "step": 8200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6505733412619585e-05,
+      "loss": 0.93,
+      "step": 8201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6503057827294407e-05,
+      "loss": 1.3738,
+      "step": 8202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.650038219366098e-05,
+      "loss": 0.7465,
+      "step": 8203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.64977065118053e-05,
+      "loss": 1.2828,
+      "step": 8204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649503078181335e-05,
+      "loss": 0.1546,
+      "step": 8205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6492355003771138e-05,
+      "loss": 0.321,
+      "step": 8206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6489679177764665e-05,
+      "loss": 0.6711,
+      "step": 8207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6487003303879922e-05,
+      "loss": 1.1275,
+      "step": 8208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.648432738220292e-05,
+      "loss": 0.8972,
+      "step": 8209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6481651412819655e-05,
+      "loss": 0.5233,
+      "step": 8210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6478975395816146e-05,
+      "loss": 0.2723,
+      "step": 8211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.647629933127839e-05,
+      "loss": 1.7325,
+      "step": 8212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6473623219292402e-05,
+      "loss": 0.0691,
+      "step": 8213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6470947059944183e-05,
+      "loss": 1.7238,
+      "step": 8214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6468270853319752e-05,
+      "loss": 1.165,
+      "step": 8215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6465594599505126e-05,
+      "loss": 0.0735,
+      "step": 8216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.64629182985863e-05,
+      "loss": 0.1147,
+      "step": 8217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6460241950649316e-05,
+      "loss": 0.0879,
+      "step": 8218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6457565555780174e-05,
+      "loss": 0.2996,
+      "step": 8219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6454889114064903e-05,
+      "loss": 0.1305,
+      "step": 8220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6452212625589523e-05,
+      "loss": 1.14,
+      "step": 8221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6449536090440046e-05,
+      "loss": 1.1032,
+      "step": 8222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6446859508702507e-05,
+      "loss": 0.6838,
+      "step": 8223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6444182880462927e-05,
+      "loss": 0.2596,
+      "step": 8224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6441506205807334e-05,
+      "loss": 0.9035,
+      "step": 8225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.643882948482176e-05,
+      "loss": 0.9096,
+      "step": 8226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.643615271759222e-05,
+      "loss": 0.6548,
+      "step": 8227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6433475904204764e-05,
+      "loss": 0.656,
+      "step": 8228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6430799044745413e-05,
+      "loss": 0.7509,
+      "step": 8229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6428122139300206e-05,
+      "loss": 0.3437,
+      "step": 8230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6425445187955178e-05,
+      "loss": 0.2956,
+      "step": 8231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6422768190796364e-05,
+      "loss": 0.7391,
+      "step": 8232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.64200911479098e-05,
+      "loss": 0.4016,
+      "step": 8233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6417414059381538e-05,
+      "loss": 0.3889,
+      "step": 8234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6414736925297608e-05,
+      "loss": 1.5623,
+      "step": 8235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6412059745744056e-05,
+      "loss": 0.4101,
+      "step": 8236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6409382520806932e-05,
+      "loss": 0.0838,
+      "step": 8237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6406705250572275e-05,
+      "loss": 0.2512,
+      "step": 8238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6404027935126134e-05,
+      "loss": 0.1933,
+      "step": 8239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6401350574554556e-05,
+      "loss": 0.1327,
+      "step": 8240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6398673168943593e-05,
+      "loss": 0.2834,
+      "step": 8241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.63959957183793e-05,
+      "loss": 0.4629,
+      "step": 8242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.639331822294773e-05,
+      "loss": 0.9016,
+      "step": 8243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.639064068273494e-05,
+      "loss": 0.2381,
+      "step": 8244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6387963097826974e-05,
+      "loss": 0.5441,
+      "step": 8245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.63852854683099e-05,
+      "loss": 0.5657,
+      "step": 8246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.638260779426977e-05,
+      "loss": 0.418,
+      "step": 8247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.637993007579265e-05,
+      "loss": 0.3646,
+      "step": 8248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.637725231296461e-05,
+      "loss": 0.555,
+      "step": 8249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.637457450587169e-05,
+      "loss": 0.2567,
+      "step": 8250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6371896654599974e-05,
+      "loss": 1.2256,
+      "step": 8251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6369218759235524e-05,
+      "loss": 0.0347,
+      "step": 8252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.636654081986441e-05,
+      "loss": 0.0349,
+      "step": 8253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.636386283657269e-05,
+      "loss": 0.5954,
+      "step": 8254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6361184809446444e-05,
+      "loss": 0.6323,
+      "step": 8255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6358506738571736e-05,
+      "loss": 0.2684,
+      "step": 8256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6355828624034648e-05,
+      "loss": 0.1511,
+      "step": 8257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6353150465921255e-05,
+      "loss": 0.184,
+      "step": 8258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6350472264317623e-05,
+      "loss": 1.6893,
+      "step": 8259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6347794019309834e-05,
+      "loss": 1.689,
+      "step": 8260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6345115730983973e-05,
+      "loss": 0.6197,
+      "step": 8261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6342437399426114e-05,
+      "loss": 0.1389,
+      "step": 8262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6339759024722334e-05,
+      "loss": 1.9177,
+      "step": 8263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6337080606958723e-05,
+      "loss": 0.289,
+      "step": 8264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6334402146221366e-05,
+      "loss": 0.1709,
+      "step": 8265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6331723642596346e-05,
+      "loss": 0.4722,
+      "step": 8266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.632904509616975e-05,
+      "loss": 0.4899,
+      "step": 8267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6326366507027668e-05,
+      "loss": 1.6094,
+      "step": 8268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6323687875256185e-05,
+      "loss": 0.1077,
+      "step": 8269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.63210092009414e-05,
+      "loss": 0.2027,
+      "step": 8270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6318330484169397e-05,
+      "loss": 0.401,
+      "step": 8271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6315651725026277e-05,
+      "loss": 0.3262,
+      "step": 8272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6312972923598134e-05,
+      "loss": 1.1248,
+      "step": 8273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.631029407997106e-05,
+      "loss": 0.2345,
+      "step": 8274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.630761519423116e-05,
+      "loss": 0.4252,
+      "step": 8275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.630493626646453e-05,
+      "loss": 0.4443,
+      "step": 8276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.630225729675727e-05,
+      "loss": 0.2944,
+      "step": 8277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6299578285195477e-05,
+      "loss": 0.2816,
+      "step": 8278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6296899231865258e-05,
+      "loss": 0.1007,
+      "step": 8279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6294220136852727e-05,
+      "loss": 0.4584,
+      "step": 8280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.629154100024398e-05,
+      "loss": 0.6723,
+      "step": 8281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6288861822125123e-05,
+      "loss": 1.0066,
+      "step": 8282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6286182602582272e-05,
+      "loss": 0.4271,
+      "step": 8283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6283503341701533e-05,
+      "loss": 0.7109,
+      "step": 8284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.628082403956902e-05,
+      "loss": 0.6564,
+      "step": 8285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.627814469627084e-05,
+      "loss": 0.4843,
+      "step": 8286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.627546531189311e-05,
+      "loss": 0.353,
+      "step": 8287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.627278588652195e-05,
+      "loss": 0.5743,
+      "step": 8288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6270106420243476e-05,
+      "loss": 1.1859,
+      "step": 8289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6267426913143798e-05,
+      "loss": 0.4055,
+      "step": 8290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.626474736530904e-05,
+      "loss": 1.3516,
+      "step": 8291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6262067776825325e-05,
+      "loss": 0.1452,
+      "step": 8292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6259388147778775e-05,
+      "loss": 0.328,
+      "step": 8293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6256708478255502e-05,
+      "loss": 1.3019,
+      "step": 8294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6254028768341646e-05,
+      "loss": 0.6754,
+      "step": 8295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6251349018123322e-05,
+      "loss": 0.459,
+      "step": 8296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6248669227686662e-05,
+      "loss": 0.3506,
+      "step": 8297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.62459893971178e-05,
+      "loss": 0.2223,
+      "step": 8298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6243309526502858e-05,
+      "loss": 0.5492,
+      "step": 8299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6240629615927964e-05,
+      "loss": 0.1839,
+      "step": 8300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6237949665479258e-05,
+      "loss": 0.5025,
+      "step": 8301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6235269675242867e-05,
+      "loss": 1.1791,
+      "step": 8302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6232589645304933e-05,
+      "loss": 1.5244,
+      "step": 8303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6229909575751588e-05,
+      "loss": 0.36,
+      "step": 8304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6227229466668972e-05,
+      "loss": 1.459,
+      "step": 8305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6224549318143218e-05,
+      "loss": 0.1071,
+      "step": 8306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6221869130260474e-05,
+      "loss": 0.1046,
+      "step": 8307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.621918890310687e-05,
+      "loss": 0.4642,
+      "step": 8308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.621650863676856e-05,
+      "loss": 1.6714,
+      "step": 8309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6213828331331676e-05,
+      "loss": 0.2771,
+      "step": 8310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6211147986882373e-05,
+      "loss": 0.3735,
+      "step": 8311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.62084676035068e-05,
+      "loss": 1.3294,
+      "step": 8312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6205787181291093e-05,
+      "loss": 1.7869,
+      "step": 8313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6203106720321403e-05,
+      "loss": 1.4682,
+      "step": 8314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.620042622068389e-05,
+      "loss": 0.1904,
+      "step": 8315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6197745682464697e-05,
+      "loss": 1.7693,
+      "step": 8316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6195065105749974e-05,
+      "loss": 0.9943,
+      "step": 8317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6192384490625878e-05,
+      "loss": 0.0795,
+      "step": 8318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6189703837178566e-05,
+      "loss": 1.9461,
+      "step": 8319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6187023145494194e-05,
+      "loss": 0.2228,
+      "step": 8320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.618434241565892e-05,
+      "loss": 0.1167,
+      "step": 8321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.61816616477589e-05,
+      "loss": 0.682,
+      "step": 8322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.617898084188029e-05,
+      "loss": 0.7953,
+      "step": 8323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6176299998109256e-05,
+      "loss": 0.3062,
+      "step": 8324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.617361911653196e-05,
+      "loss": 0.5392,
+      "step": 8325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6170938197234565e-05,
+      "loss": 0.3428,
+      "step": 8326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6168257240303242e-05,
+      "loss": 0.3055,
+      "step": 8327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6165576245824144e-05,
+      "loss": 0.5064,
+      "step": 8328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6162895213883446e-05,
+      "loss": 0.4329,
+      "step": 8329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6160214144567318e-05,
+      "loss": 0.068,
+      "step": 8330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6157533037961926e-05,
+      "loss": 1.6427,
+      "step": 8331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.615485189415344e-05,
+      "loss": 0.6936,
+      "step": 8332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.615217071322803e-05,
+      "loss": 1.7955,
+      "step": 8333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6149489495271875e-05,
+      "loss": 1.0887,
+      "step": 8334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6146808240371153e-05,
+      "loss": 0.3009,
+      "step": 8335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.614412694861203e-05,
+      "loss": 0.8891,
+      "step": 8336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6141445620080685e-05,
+      "loss": 0.8672,
+      "step": 8337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6138764254863293e-05,
+      "loss": 0.2478,
+      "step": 8338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6136082853046045e-05,
+      "loss": 0.0692,
+      "step": 8339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6133401414715103e-05,
+      "loss": 0.469,
+      "step": 8340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6130719939956666e-05,
+      "loss": 1.184,
+      "step": 8341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6128038428856907e-05,
+      "loss": 1.0111,
+      "step": 8342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.612535688150201e-05,
+      "loss": 0.7996,
+      "step": 8343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6122675297978167e-05,
+      "loss": 0.3164,
+      "step": 8344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6119993678371553e-05,
+      "loss": 0.0805,
+      "step": 8345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6117312022768363e-05,
+      "loss": 1.069,
+      "step": 8346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6114630331254784e-05,
+      "loss": 0.2652,
+      "step": 8347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6111948603916997e-05,
+      "loss": 0.7188,
+      "step": 8348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6109266840841213e-05,
+      "loss": 0.6017,
+      "step": 8349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6106585042113606e-05,
+      "loss": 0.5771,
+      "step": 8350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6103903207820376e-05,
+      "loss": 0.0377,
+      "step": 8351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6101221338047714e-05,
+      "loss": 0.1593,
+      "step": 8352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6098539432881817e-05,
+      "loss": 0.3444,
+      "step": 8353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6095857492408874e-05,
+      "loss": 0.5581,
+      "step": 8354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6093175516715097e-05,
+      "loss": 0.5804,
+      "step": 8355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6090493505886672e-05,
+      "loss": 0.106,
+      "step": 8356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.608781146000981e-05,
+      "loss": 0.5822,
+      "step": 8357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6085129379170703e-05,
+      "loss": 0.8694,
+      "step": 8358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6082447263455557e-05,
+      "loss": 1.2135,
+      "step": 8359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6079765112950572e-05,
+      "loss": 1.6178,
+      "step": 8360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.607708292774196e-05,
+      "loss": 0.2804,
+      "step": 8361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6074400707915914e-05,
+      "loss": 1.5538,
+      "step": 8362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6071718453558645e-05,
+      "loss": 1.3379,
+      "step": 8363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.606903616475637e-05,
+      "loss": 0.2299,
+      "step": 8364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6066353841595284e-05,
+      "loss": 0.1714,
+      "step": 8365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.606367148416161e-05,
+      "loss": 0.0508,
+      "step": 8366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.606098909254155e-05,
+      "loss": 0.0486,
+      "step": 8367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.605830666682132e-05,
+      "loss": 0.2562,
+      "step": 8368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6055624207087126e-05,
+      "loss": 0.2449,
+      "step": 8369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6052941713425195e-05,
+      "loss": 1.7573,
+      "step": 8370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6050259185921725e-05,
+      "loss": 0.4468,
+      "step": 8371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6047576624662946e-05,
+      "loss": 0.6991,
+      "step": 8372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6044894029735074e-05,
+      "loss": 0.3756,
+      "step": 8373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6042211401224325e-05,
+      "loss": 1.1946,
+      "step": 8374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6039528739216923e-05,
+      "loss": 0.3673,
+      "step": 8375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.603684604379908e-05,
+      "loss": 0.5494,
+      "step": 8376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.603416331505702e-05,
+      "loss": 0.1979,
+      "step": 8377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6031480553076972e-05,
+      "loss": 0.9398,
+      "step": 8378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6028797757945152e-05,
+      "loss": 0.6615,
+      "step": 8379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6026114929747795e-05,
+      "loss": 1.8828,
+      "step": 8380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.602343206857112e-05,
+      "loss": 0.4175,
+      "step": 8381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6020749174501355e-05,
+      "loss": 1.168,
+      "step": 8382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6018066247624728e-05,
+      "loss": 0.9076,
+      "step": 8383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.601538328802747e-05,
+      "loss": 1.9681,
+      "step": 8384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.601270029579581e-05,
+      "loss": 1.1176,
+      "step": 8385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6010017271015978e-05,
+      "loss": 0.6078,
+      "step": 8386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6007334213774204e-05,
+      "loss": 0.807,
+      "step": 8387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.600465112415673e-05,
+      "loss": 1.168,
+      "step": 8388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6001968002249787e-05,
+      "loss": 0.6914,
+      "step": 8389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5999284848139607e-05,
+      "loss": 1.2604,
+      "step": 8390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5996601661912428e-05,
+      "loss": 0.5377,
+      "step": 8391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.599391844365449e-05,
+      "loss": 0.5534,
+      "step": 8392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.599123519345203e-05,
+      "loss": 0.5268,
+      "step": 8393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5988551911391284e-05,
+      "loss": 0.4553,
+      "step": 8394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5985868597558496e-05,
+      "loss": 0.2528,
+      "step": 8395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5983185252039914e-05,
+      "loss": 0.3523,
+      "step": 8396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5980501874921768e-05,
+      "loss": 0.4128,
+      "step": 8397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5977818466290313e-05,
+      "loss": 0.7086,
+      "step": 8398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.597513502623179e-05,
+      "loss": 0.4257,
+      "step": 8399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5972451554832435e-05,
+      "loss": 0.1116,
+      "step": 8400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.596976805217851e-05,
+      "loss": 1.3941,
+      "step": 8401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.596708451835625e-05,
+      "loss": 0.0599,
+      "step": 8402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5964400953451916e-05,
+      "loss": 1.2619,
+      "step": 8403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5961717357551753e-05,
+      "loss": 0.5616,
+      "step": 8404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5959033730742003e-05,
+      "loss": 0.0419,
+      "step": 8405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.595635007310893e-05,
+      "loss": 0.2816,
+      "step": 8406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5953666384738783e-05,
+      "loss": 0.3727,
+      "step": 8407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5950982665717814e-05,
+      "loss": 0.0426,
+      "step": 8408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.594829891613227e-05,
+      "loss": 0.0552,
+      "step": 8409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.594561513606842e-05,
+      "loss": 0.3142,
+      "step": 8410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5942931325612514e-05,
+      "loss": 0.4886,
+      "step": 8411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5940247484850816e-05,
+      "loss": 0.0922,
+      "step": 8412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5937563613869575e-05,
+      "loss": 1.0805,
+      "step": 8413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5934879712755055e-05,
+      "loss": 1.4443,
+      "step": 8414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.593219578159352e-05,
+      "loss": 1.3039,
+      "step": 8415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.592951182047123e-05,
+      "loss": 0.2719,
+      "step": 8416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5926827829474434e-05,
+      "loss": 0.0328,
+      "step": 8417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5924143808689416e-05,
+      "loss": 1.1929,
+      "step": 8418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.592145975820243e-05,
+      "loss": 0.3217,
+      "step": 8419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5918775678099743e-05,
+      "loss": 0.4954,
+      "step": 8420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5916091568467623e-05,
+      "loss": 0.6163,
+      "step": 8421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5913407429392333e-05,
+      "loss": 0.0313,
+      "step": 8422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5910723260960144e-05,
+      "loss": 0.5391,
+      "step": 8423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5908039063257326e-05,
+      "loss": 0.1593,
+      "step": 8424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.590535483637014e-05,
+      "loss": 0.8222,
+      "step": 8425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5902670580384875e-05,
+      "loss": 1.2731,
+      "step": 8426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5899986295387787e-05,
+      "loss": 1.271,
+      "step": 8427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5897301981465154e-05,
+      "loss": 0.5032,
+      "step": 8428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5894617638703258e-05,
+      "loss": 0.763,
+      "step": 8429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5891933267188357e-05,
+      "loss": 1.4187,
+      "step": 8430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5889248867006738e-05,
+      "loss": 0.1414,
+      "step": 8431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5886564438244675e-05,
+      "loss": 0.7068,
+      "step": 8432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5883879980988448e-05,
+      "loss": 0.734,
+      "step": 8433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.588119549532433e-05,
+      "loss": 0.283,
+      "step": 8434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5878510981338608e-05,
+      "loss": 1.1711,
+      "step": 8435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5875826439117556e-05,
+      "loss": 0.909,
+      "step": 8436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5873141868747454e-05,
+      "loss": 0.7049,
+      "step": 8437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5870457270314592e-05,
+      "loss": 1.321,
+      "step": 8438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5867772643905245e-05,
+      "loss": 0.4128,
+      "step": 8439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5865087989605695e-05,
+      "loss": 0.0778,
+      "step": 8440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5862403307502237e-05,
+      "loss": 0.8594,
+      "step": 8441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5859718597681148e-05,
+      "loss": 0.3287,
+      "step": 8442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5857033860228722e-05,
+      "loss": 0.3913,
+      "step": 8443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5854349095231242e-05,
+      "loss": 0.267,
+      "step": 8444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5851664302774994e-05,
+      "loss": 0.9081,
+      "step": 8445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.584897948294627e-05,
+      "loss": 0.2854,
+      "step": 8446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.584629463583136e-05,
+      "loss": 0.8885,
+      "step": 8447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.584360976151655e-05,
+      "loss": 0.8284,
+      "step": 8448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5840924860088142e-05,
+      "loss": 0.1178,
+      "step": 8449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.583823993163242e-05,
+      "loss": 0.4043,
+      "step": 8450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5835554976235682e-05,
+      "loss": 0.0345,
+      "step": 8451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5832869993984226e-05,
+      "loss": 1.1956,
+      "step": 8452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5830184984964337e-05,
+      "loss": 1.2176,
+      "step": 8453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5827499949262314e-05,
+      "loss": 0.5526,
+      "step": 8454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.582481488696446e-05,
+      "loss": 0.1142,
+      "step": 8455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5822129798157064e-05,
+      "loss": 0.0321,
+      "step": 8456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.581944468292644e-05,
+      "loss": 0.116,
+      "step": 8457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5816759541358867e-05,
+      "loss": 0.7266,
+      "step": 8458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5814074373540663e-05,
+      "loss": 0.2381,
+      "step": 8459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5811389179558122e-05,
+      "loss": 0.1646,
+      "step": 8460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5808703959497545e-05,
+      "loss": 1.7439,
+      "step": 8461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5806018713445236e-05,
+      "loss": 0.2575,
+      "step": 8462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5803333441487496e-05,
+      "loss": 0.3727,
+      "step": 8463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5800648143710636e-05,
+      "loss": 0.3891,
+      "step": 8464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5797962820200956e-05,
+      "loss": 2.0099,
+      "step": 8465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5795277471044768e-05,
+      "loss": 0.9613,
+      "step": 8466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5792592096328375e-05,
+      "loss": 0.3315,
+      "step": 8467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.578990669613808e-05,
+      "loss": 0.0717,
+      "step": 8468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.57872212705602e-05,
+      "loss": 0.6128,
+      "step": 8469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5784535819681044e-05,
+      "loss": 0.1918,
+      "step": 8470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5781850343586916e-05,
+      "loss": 1.2836,
+      "step": 8471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5779164842364133e-05,
+      "loss": 0.7194,
+      "step": 8472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5776479316099004e-05,
+      "loss": 0.4005,
+      "step": 8473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.577379376487784e-05,
+      "loss": 1.3189,
+      "step": 8474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5771108188786968e-05,
+      "loss": 0.2638,
+      "step": 8475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5768422587912683e-05,
+      "loss": 1.3241,
+      "step": 8476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.576573696234131e-05,
+      "loss": 1.3056,
+      "step": 8477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5763051312159164e-05,
+      "loss": 0.0938,
+      "step": 8478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5760365637452558e-05,
+      "loss": 0.2729,
+      "step": 8479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.575767993830782e-05,
+      "loss": 0.0473,
+      "step": 8480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.575499421481126e-05,
+      "loss": 0.2196,
+      "step": 8481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5752308467049193e-05,
+      "loss": 0.7287,
+      "step": 8482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5749622695107948e-05,
+      "loss": 0.1434,
+      "step": 8483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5746936899073847e-05,
+      "loss": 1.5412,
+      "step": 8484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.57442510790332e-05,
+      "loss": 0.5526,
+      "step": 8485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5741565235072337e-05,
+      "loss": 0.6177,
+      "step": 8486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5738879367277577e-05,
+      "loss": 0.8783,
+      "step": 8487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5736193475735253e-05,
+      "loss": 0.9517,
+      "step": 8488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.573350756053168e-05,
+      "loss": 0.981,
+      "step": 8489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5730821621753187e-05,
+      "loss": 0.5692,
+      "step": 8490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.57281356594861e-05,
+      "loss": 0.0831,
+      "step": 8491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.572544967381674e-05,
+      "loss": 0.1934,
+      "step": 8492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5722763664831444e-05,
+      "loss": 1.1875,
+      "step": 8493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5720077632616534e-05,
+      "loss": 1.0821,
+      "step": 8494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.571739157725834e-05,
+      "loss": 0.076,
+      "step": 8495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5714705498843196e-05,
+      "loss": 0.6348,
+      "step": 8496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.571201939745743e-05,
+      "loss": 1.0109,
+      "step": 8497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.570933327318737e-05,
+      "loss": 0.0947,
+      "step": 8498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.570664712611935e-05,
+      "loss": 0.1019,
+      "step": 8499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5703960956339704e-05,
+      "loss": 0.8525,
+      "step": 8500
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 14019,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 7827671395074048.0,
+  "trial_name": null,
+  "trial_params": null
+}