diff --git "a/7b_model/trainer_state.json" "b/7b_model/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/7b_model/trainer_state.json"
@@ -0,0 +1,42037 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 7002,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.478672985781992e-08,
+      "loss": 0.291,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8957345971563984e-07,
+      "loss": 1.5023,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.843601895734597e-07,
+      "loss": 1.5691,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.791469194312797e-07,
+      "loss": 1.5803,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7393364928909956e-07,
+      "loss": 1.5388,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.687203791469194e-07,
+      "loss": 1.5141,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.635071090047394e-07,
+      "loss": 1.5756,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.582938388625594e-07,
+      "loss": 1.5102,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.530805687203792e-07,
+      "loss": 1.6015,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.478672985781991e-07,
+      "loss": 1.5072,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.042654028436019e-06,
+      "loss": 1.5701,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1374407582938388e-06,
+      "loss": 1.5184,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2322274881516587e-06,
+      "loss": 1.5082,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3270142180094788e-06,
+      "loss": 1.4829,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4218009478672987e-06,
+      "loss": 1.4011,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5165876777251187e-06,
+      "loss": 1.4312,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6113744075829384e-06,
+      "loss": 1.3164,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7061611374407585e-06,
+      "loss": 1.3462,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8009478672985784e-06,
+      "loss": 1.378,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8957345971563982e-06,
+      "loss": 1.4261,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.990521327014218e-06,
+      "loss": 1.3852,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.085308056872038e-06,
+      "loss": 1.3224,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.180094786729858e-06,
+      "loss": 1.3063,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2748815165876777e-06,
+      "loss": 1.2438,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.369668246445498e-06,
+      "loss": 0.3191,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4644549763033174e-06,
+      "loss": 1.4079,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5592417061611373e-06,
+      "loss": 0.3075,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6540284360189576e-06,
+      "loss": 0.3028,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7488151658767775e-06,
+      "loss": 1.3607,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8436018957345973e-06,
+      "loss": 1.2609,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.938388625592417e-06,
+      "loss": 1.2312,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0331753554502375e-06,
+      "loss": 1.2696,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1279620853080574e-06,
+      "loss": 1.3535,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.222748815165877e-06,
+      "loss": 1.3544,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3175355450236967e-06,
+      "loss": 1.3323,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.412322274881517e-06,
+      "loss": 1.2493,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.507109004739337e-06,
+      "loss": 1.2508,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6018957345971567e-06,
+      "loss": 1.2871,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6966824644549766e-06,
+      "loss": 1.1818,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7914691943127964e-06,
+      "loss": 1.2141,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.886255924170616e-06,
+      "loss": 1.2039,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.981042654028436e-06,
+      "loss": 1.2268,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.075829383886256e-06,
+      "loss": 1.2447,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.170616113744076e-06,
+      "loss": 1.2165,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.265402843601897e-06,
+      "loss": 1.1967,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.360189573459716e-06,
+      "loss": 1.2533,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4549763033175355e-06,
+      "loss": 1.2422,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.549763033175355e-06,
+      "loss": 1.2157,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.644549763033176e-06,
+      "loss": 1.1797,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.739336492890996e-06,
+      "loss": 1.1988,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.834123222748816e-06,
+      "loss": 1.2278,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.928909952606635e-06,
+      "loss": 1.1843,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.023696682464455e-06,
+      "loss": 1.1932,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.118483412322275e-06,
+      "loss": 0.2782,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.213270142180096e-06,
+      "loss": 1.1883,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.308056872037915e-06,
+      "loss": 1.1787,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.402843601895735e-06,
+      "loss": 1.2025,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.497630331753555e-06,
+      "loss": 1.1554,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.592417061611375e-06,
+      "loss": 1.1928,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.687203791469195e-06,
+      "loss": 1.1872,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7819905213270145e-06,
+      "loss": 1.1871,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.876777251184834e-06,
+      "loss": 1.0827,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.971563981042654e-06,
+      "loss": 1.171,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.066350710900475e-06,
+      "loss": 1.1623,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.161137440758295e-06,
+      "loss": 1.1254,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.255924170616115e-06,
+      "loss": 1.1463,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.350710900473935e-06,
+      "loss": 1.1616,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.445497630331754e-06,
+      "loss": 1.1152,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5402843601895735e-06,
+      "loss": 1.2683,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.635071090047393e-06,
+      "loss": 1.1138,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.729857819905213e-06,
+      "loss": 1.1346,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.824644549763034e-06,
+      "loss": 1.1221,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.919431279620854e-06,
+      "loss": 1.0776,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.014218009478674e-06,
+      "loss": 1.1674,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1090047393364935e-06,
+      "loss": 1.13,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.203791469194313e-06,
+      "loss": 1.1337,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.298578199052133e-06,
+      "loss": 1.1914,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.393364928909953e-06,
+      "loss": 1.155,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.488151658767773e-06,
+      "loss": 1.1966,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.582938388625593e-06,
+      "loss": 1.1829,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.677725118483414e-06,
+      "loss": 1.1488,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.772511848341233e-06,
+      "loss": 1.0676,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.867298578199053e-06,
+      "loss": 1.1027,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.962085308056872e-06,
+      "loss": 1.162,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.056872037914693e-06,
+      "loss": 1.1125,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.151658767772512e-06,
+      "loss": 1.1654,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.246445497630333e-06,
+      "loss": 1.179,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.341232227488152e-06,
+      "loss": 1.1225,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.436018957345973e-06,
+      "loss": 1.1709,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.530805687203793e-06,
+      "loss": 1.1753,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.625592417061612e-06,
+      "loss": 1.1695,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.720379146919431e-06,
+      "loss": 1.2186,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.815165876777252e-06,
+      "loss": 1.1182,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.909952606635071e-06,
+      "loss": 1.0952,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.004739336492892e-06,
+      "loss": 1.1037,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.09952606635071e-06,
+      "loss": 1.137,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.194312796208532e-06,
+      "loss": 1.1593,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.289099526066352e-06,
+      "loss": 1.0799,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.383886255924171e-06,
+      "loss": 1.0745,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.478672985781992e-06,
+      "loss": 1.1455,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.573459715639811e-06,
+      "loss": 1.0874,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.668246445497632e-06,
+      "loss": 1.1317,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.76303317535545e-06,
+      "loss": 1.1139,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.85781990521327e-06,
+      "loss": 0.3148,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.95260663507109e-06,
+      "loss": 1.1293,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.004739336492891e-05,
+      "loss": 1.1323,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0142180094786732e-05,
+      "loss": 1.1513,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.023696682464455e-05,
+      "loss": 1.0873,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.033175355450237e-05,
+      "loss": 1.1491,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0426540284360192e-05,
+      "loss": 1.1472,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.052132701421801e-05,
+      "loss": 1.1506,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.061611374407583e-05,
+      "loss": 1.1677,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.071090047393365e-05,
+      "loss": 1.1641,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.080568720379147e-05,
+      "loss": 1.0745,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0900473933649289e-05,
+      "loss": 1.1235,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.099526066350711e-05,
+      "loss": 1.1201,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1090047393364929e-05,
+      "loss": 1.042,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.118483412322275e-05,
+      "loss": 1.1797,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.127962085308057e-05,
+      "loss": 1.1463,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.137440758293839e-05,
+      "loss": 1.098,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.146919431279621e-05,
+      "loss": 1.1733,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1563981042654029e-05,
+      "loss": 1.1337,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.165876777251185e-05,
+      "loss": 1.1842,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1753554502369669e-05,
+      "loss": 0.9859,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.184834123222749e-05,
+      "loss": 1.0932,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1943127962085309e-05,
+      "loss": 1.1653,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.203791469194313e-05,
+      "loss": 1.1006,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.213270142180095e-05,
+      "loss": 1.1369,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2227488151658769e-05,
+      "loss": 1.1147,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.232227488151659e-05,
+      "loss": 1.1291,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2417061611374409e-05,
+      "loss": 1.0869,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.251184834123223e-05,
+      "loss": 1.1287,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2606635071090047e-05,
+      "loss": 1.1265,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.270142180094787e-05,
+      "loss": 1.127,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2796208530805687e-05,
+      "loss": 1.1226,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2890995260663507e-05,
+      "loss": 1.0994,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.298578199052133e-05,
+      "loss": 1.1796,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3080568720379147e-05,
+      "loss": 1.0504,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3175355450236968e-05,
+      "loss": 1.0645,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3270142180094787e-05,
+      "loss": 1.0959,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3364928909952607e-05,
+      "loss": 1.0283,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3459715639810426e-05,
+      "loss": 1.1248,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3554502369668247e-05,
+      "loss": 0.3253,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3649289099526068e-05,
+      "loss": 0.3012,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3744075829383887e-05,
+      "loss": 1.069,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3838862559241708e-05,
+      "loss": 1.1421,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3933649289099527e-05,
+      "loss": 1.1275,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4028436018957347e-05,
+      "loss": 1.0801,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4123222748815166e-05,
+      "loss": 1.1286,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4218009478672987e-05,
+      "loss": 1.1525,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4312796208530806e-05,
+      "loss": 1.0526,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4407582938388627e-05,
+      "loss": 1.1604,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4502369668246448e-05,
+      "loss": 1.1082,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4597156398104267e-05,
+      "loss": 1.1349,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4691943127962087e-05,
+      "loss": 1.0715,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4786729857819906e-05,
+      "loss": 1.0367,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4881516587677727e-05,
+      "loss": 1.1615,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4976303317535546e-05,
+      "loss": 1.1017,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5071090047393367e-05,
+      "loss": 1.0954,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5165876777251186e-05,
+      "loss": 0.2974,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5260663507109007e-05,
+      "loss": 1.1559,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5355450236966827e-05,
+      "loss": 1.0991,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5450236966824645e-05,
+      "loss": 1.0605,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5545023696682465e-05,
+      "loss": 1.0889,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5639810426540286e-05,
+      "loss": 1.0973,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5734597156398107e-05,
+      "loss": 1.1107,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5829383886255924e-05,
+      "loss": 1.139,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5924170616113745e-05,
+      "loss": 1.0995,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6018957345971565e-05,
+      "loss": 1.1316,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6113744075829386e-05,
+      "loss": 1.1303,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6208530805687207e-05,
+      "loss": 1.1039,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6303317535545024e-05,
+      "loss": 1.1314,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6398104265402845e-05,
+      "loss": 1.0804,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6492890995260666e-05,
+      "loss": 1.1244,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6587677725118486e-05,
+      "loss": 1.1196,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6682464454976304e-05,
+      "loss": 1.1195,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6777251184834124e-05,
+      "loss": 1.0742,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6872037914691945e-05,
+      "loss": 1.1274,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6966824644549766e-05,
+      "loss": 1.1569,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7061611374407587e-05,
+      "loss": 1.0681,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7156398104265404e-05,
+      "loss": 1.0368,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7251184834123225e-05,
+      "loss": 0.3123,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7345971563981042e-05,
+      "loss": 1.1691,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7440758293838863e-05,
+      "loss": 1.124,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7535545023696683e-05,
+      "loss": 1.1034,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7630331753554504e-05,
+      "loss": 1.0987,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7725118483412325e-05,
+      "loss": 1.1379,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7819905213270142e-05,
+      "loss": 1.1159,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7914691943127963e-05,
+      "loss": 1.1735,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8009478672985784e-05,
+      "loss": 1.06,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8104265402843604e-05,
+      "loss": 1.0849,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.819905213270142e-05,
+      "loss": 1.1024,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8293838862559242e-05,
+      "loss": 1.1022,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8388625592417063e-05,
+      "loss": 1.0883,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8483412322274884e-05,
+      "loss": 1.0663,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8578199052132704e-05,
+      "loss": 1.0752,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8672985781990522e-05,
+      "loss": 1.1079,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8767772511848342e-05,
+      "loss": 1.0815,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8862559241706163e-05,
+      "loss": 1.0921,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8957345971563984e-05,
+      "loss": 1.0686,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.90521327014218e-05,
+      "loss": 1.1067,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9146919431279622e-05,
+      "loss": 0.3157,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9241706161137443e-05,
+      "loss": 1.1561,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9336492890995263e-05,
+      "loss": 1.0911,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9431279620853084e-05,
+      "loss": 1.1446,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.95260663507109e-05,
+      "loss": 1.0748,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9620853080568722e-05,
+      "loss": 1.138,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.971563981042654e-05,
+      "loss": 1.1386,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9810426540284364e-05,
+      "loss": 1.146,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.990521327014218e-05,
+      "loss": 1.1289,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.1502,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998929954413e-05,
+      "loss": 1.0806,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999995719817877e-05,
+      "loss": 1.1496,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999990369591077e-05,
+      "loss": 1.1146,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998287927516e-05,
+      "loss": 1.1049,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997324887173e-05,
+      "loss": 1.096,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999961478382845e-05,
+      "loss": 1.0893,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999994756781103e-05,
+      "loss": 1.0753,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999931517159255e-05,
+      "loss": 1.0572,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991332643096e-05,
+      "loss": 1.1155,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999892995630034e-05,
+      "loss": 1.1166,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987052476083e-05,
+      "loss": 1.0625,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984591382816e-05,
+      "loss": 1.0833,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999819162837286e-05,
+      "loss": 1.1129,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999790271793934e-05,
+      "loss": 1.0703,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999975924070429e-05,
+      "loss": 1.0991,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972606957499e-05,
+      "loss": 1.1391,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999969075841314e-05,
+      "loss": 1.1785,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999965330722629e-05,
+      "loss": 1.049,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999961371602246e-05,
+      "loss": 0.325,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999957198481012e-05,
+      "loss": 1.1002,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999528113598198e-05,
+      "loss": 1.1334,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999948210239609e-05,
+      "loss": 1.0793,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999433951213633e-05,
+      "loss": 1.0882,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999938366006114e-05,
+      "loss": 1.0859,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999331228949375e-05,
+      "loss": 1.1124,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999276657889556e-05,
+      "loss": 1.1184,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999219946893356e-05,
+      "loss": 1.1211,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999161095972922e-05,
+      "loss": 1.1497,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999100105140837e-05,
+      "loss": 1.0876,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999036974410163e-05,
+      "loss": 1.0883,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999897170379441e-05,
+      "loss": 1.1041,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999890429330754e-05,
+      "loss": 1.0891,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998834742963988e-05,
+      "loss": 1.0842,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998763052778627e-05,
+      "loss": 1.0822,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999868922276681e-05,
+      "loss": 1.0893,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998613252944334e-05,
+      "loss": 1.1645,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999853514332746e-05,
+      "loss": 1.1127,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998454893932893e-05,
+      "loss": 1.1394,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999837250477782e-05,
+      "loss": 1.0415,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998287975879868e-05,
+      "loss": 1.1097,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998201307257123e-05,
+      "loss": 1.1119,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999811249892814e-05,
+      "loss": 1.081,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999802155091192e-05,
+      "loss": 1.1086,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999792846322793e-05,
+      "loss": 1.1164,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997833235896088e-05,
+      "loss": 1.1356,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997735868936778e-05,
+      "loss": 1.1784,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997636362370834e-05,
+      "loss": 1.14,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999753471621955e-05,
+      "loss": 1.0898,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997430930504685e-05,
+      "loss": 1.1269,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997325005248444e-05,
+      "loss": 1.1087,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99972169404735e-05,
+      "loss": 1.038,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997106736202977e-05,
+      "loss": 1.1052,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999699439246046e-05,
+      "loss": 1.094,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996879909269995e-05,
+      "loss": 1.0925,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999676328665608e-05,
+      "loss": 0.9878,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999664452464367e-05,
+      "loss": 1.1476,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996523623258186e-05,
+      "loss": 1.1389,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996400582525504e-05,
+      "loss": 1.1346,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999627540247195e-05,
+      "loss": 1.1267,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999614808312431e-05,
+      "loss": 1.0809,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996018624509846e-05,
+      "loss": 1.0766,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999588702665625e-05,
+      "loss": 0.2856,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995753289591692e-05,
+      "loss": 1.105,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995617413344792e-05,
+      "loss": 1.0756,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995479397944625e-05,
+      "loss": 1.1143,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995339243420733e-05,
+      "loss": 1.1027,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995196949803106e-05,
+      "loss": 1.0746,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995052517122196e-05,
+      "loss": 0.3291,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994905945408917e-05,
+      "loss": 1.1525,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994757234694635e-05,
+      "loss": 1.0735,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994606385011172e-05,
+      "loss": 0.3193,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994453396390817e-05,
+      "loss": 1.0707,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994298268866307e-05,
+      "loss": 1.1078,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999414100247084e-05,
+      "loss": 1.0047,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993981597238073e-05,
+      "loss": 1.0924,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993820053202122e-05,
+      "loss": 1.0592,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993656370397558e-05,
+      "loss": 1.0129,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993490548859413e-05,
+      "loss": 1.1076,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999332258862317e-05,
+      "loss": 1.0679,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999315248972478e-05,
+      "loss": 1.0981,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992980252200636e-05,
+      "loss": 1.0924,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992805876087607e-05,
+      "loss": 1.056,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999262936142301e-05,
+      "loss": 1.037,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999245070824462e-05,
+      "loss": 1.1322,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992269916590663e-05,
+      "loss": 1.0815,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992086986499842e-05,
+      "loss": 1.124,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991901918011296e-05,
+      "loss": 1.0748,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999171471116464e-05,
+      "loss": 1.0433,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999152536599993e-05,
+      "loss": 1.0482,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991333882557692e-05,
+      "loss": 1.0732,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991140260878906e-05,
+      "loss": 1.0417,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990944501005005e-05,
+      "loss": 1.124,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990746602977887e-05,
+      "loss": 1.0691,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99905465668399e-05,
+      "loss": 1.0844,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990344392633857e-05,
+      "loss": 1.091,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990140080403027e-05,
+      "loss": 1.1413,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998993363019113e-05,
+      "loss": 1.1338,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998972504204235e-05,
+      "loss": 1.0385,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989514316001325e-05,
+      "loss": 1.1102,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989301452113158e-05,
+      "loss": 1.0669,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99890864504234e-05,
+      "loss": 1.0676,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998886931097806e-05,
+      "loss": 1.1292,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988650033823612e-05,
+      "loss": 1.1089,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988428619006982e-05,
+      "loss": 1.1421,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988205066575553e-05,
+      "loss": 1.1119,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987979376577175e-05,
+      "loss": 1.1852,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998775154906014e-05,
+      "loss": 1.0914,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987521584073207e-05,
+      "loss": 1.0808,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998728948166559e-05,
+      "loss": 1.1597,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987055241886962e-05,
+      "loss": 0.3612,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986818864787455e-05,
+      "loss": 1.0995,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998658035041765e-05,
+      "loss": 1.1165,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986339698828597e-05,
+      "loss": 1.0705,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986096910071793e-05,
+      "loss": 1.0726,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99858519841992e-05,
+      "loss": 1.0496,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985604921263237e-05,
+      "loss": 1.151,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985355721316768e-05,
+      "loss": 1.0945,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985104384413134e-05,
+      "loss": 1.0449,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998485091060612e-05,
+      "loss": 1.0935,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998459529994997e-05,
+      "loss": 1.0804,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998433755249939e-05,
+      "loss": 1.072,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998407766830954e-05,
+      "loss": 1.0711,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998381564743603e-05,
+      "loss": 1.159,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983551489934948e-05,
+      "loss": 1.0984,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983285195862814e-05,
+      "loss": 1.0552,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983016765276625e-05,
+      "loss": 1.0671,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982746198233824e-05,
+      "loss": 1.0432,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998247349479232e-05,
+      "loss": 1.0577,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982198655010467e-05,
+      "loss": 0.9995,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981921678947085e-05,
+      "loss": 1.0666,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981642566661454e-05,
+      "loss": 1.1055,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981361318213302e-05,
+      "loss": 1.0505,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998107793366282e-05,
+      "loss": 1.0832,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998079241307065e-05,
+      "loss": 1.0724,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998050475649791e-05,
+      "loss": 1.1146,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980214964006145e-05,
+      "loss": 1.0164,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979923035657385e-05,
+      "loss": 1.0545,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99796289715141e-05,
+      "loss": 1.0618,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997933277163922e-05,
+      "loss": 1.0296,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997903443609614e-05,
+      "loss": 1.1016,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99787339649487e-05,
+      "loss": 1.1171,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997843135826121e-05,
+      "loss": 1.109,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997812661609843e-05,
+      "loss": 1.081,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977819738525576e-05,
+      "loss": 1.1392,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977510725608318e-05,
+      "loss": 1.1003,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977199577412795e-05,
+      "loss": 1.1169,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997688629400559e-05,
+      "loss": 1.1034,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976570875453754e-05,
+      "loss": 1.0754,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976253321824787e-05,
+      "loss": 1.0591,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975933633186645e-05,
+      "loss": 1.0782,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975611809607748e-05,
+      "loss": 1.0895,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997528785115697e-05,
+      "loss": 1.0846,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974961757903635e-05,
+      "loss": 1.0149,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997463352991754e-05,
+      "loss": 1.0112,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997430316726892e-05,
+      "loss": 1.0159,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973970670028476e-05,
+      "loss": 1.0761,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973636038267373e-05,
+      "loss": 1.0484,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997329927205722e-05,
+      "loss": 1.0676,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972960371470088e-05,
+      "loss": 1.0728,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972619336578504e-05,
+      "loss": 1.1035,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972276167455452e-05,
+      "loss": 1.1275,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971930864174376e-05,
+      "loss": 1.052,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971583426809172e-05,
+      "loss": 1.1074,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99712338554342e-05,
+      "loss": 1.1065,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970882150124263e-05,
+      "loss": 1.0423,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997052831095464e-05,
+      "loss": 1.0963,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970172338001044e-05,
+      "loss": 1.0657,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996981423133966e-05,
+      "loss": 1.0261,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969453991047133e-05,
+      "loss": 1.1315,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996909161720055e-05,
+      "loss": 0.3809,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968727109877464e-05,
+      "loss": 1.0708,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968360469155887e-05,
+      "loss": 1.1336,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967991695114276e-05,
+      "loss": 1.007,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996762078783156e-05,
+      "loss": 1.0968,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996724774738711e-05,
+      "loss": 1.1415,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966872573860766e-05,
+      "loss": 1.1429,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966495267332813e-05,
+      "loss": 1.0643,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966115827884002e-05,
+      "loss": 1.0799,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965734255595533e-05,
+      "loss": 1.0055,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965350550549068e-05,
+      "loss": 1.0583,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996496471282672e-05,
+      "loss": 1.1311,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964576742511073e-05,
+      "loss": 1.1169,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964186639685143e-05,
+      "loss": 1.0275,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963794404432422e-05,
+      "loss": 1.0812,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963400036836852e-05,
+      "loss": 1.0949,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996300353698283e-05,
+      "loss": 1.0587,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962604904955207e-05,
+      "loss": 1.0467,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99622041408393e-05,
+      "loss": 0.9934,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996180124472087e-05,
+      "loss": 1.1241,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961396216686153e-05,
+      "loss": 1.0148,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960989056821813e-05,
+      "loss": 1.0459,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960579765214995e-05,
+      "loss": 1.0912,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960168341953284e-05,
+      "loss": 1.1069,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995975478712474e-05,
+      "loss": 1.092,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959339100817856e-05,
+      "loss": 1.1431,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958921283121603e-05,
+      "loss": 1.0327,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958501334125385e-05,
+      "loss": 1.024,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995807925391909e-05,
+      "loss": 1.0812,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995765504259303e-05,
+      "loss": 1.0834,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957228700238003e-05,
+      "loss": 1.0663,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956800226945248e-05,
+      "loss": 1.1208,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956369622806456e-05,
+      "loss": 1.0946,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955936887913788e-05,
+      "loss": 1.0242,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995550202235985e-05,
+      "loss": 1.1111,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995506502623771e-05,
+      "loss": 1.0973,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995462589964088e-05,
+      "loss": 1.0602,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954184642663347e-05,
+      "loss": 1.0643,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995374125539954e-05,
+      "loss": 1.0337,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953295737944347e-05,
+      "loss": 1.0981,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952848090393114e-05,
+      "loss": 1.1169,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952398312841645e-05,
+      "loss": 1.1616,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951946405386188e-05,
+      "loss": 1.0596,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951492368123465e-05,
+      "loss": 0.3803,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951036201150638e-05,
+      "loss": 1.0267,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950577904565335e-05,
+      "loss": 1.0733,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995011747846563e-05,
+      "loss": 1.0589,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949654922950065e-05,
+      "loss": 1.1852,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949190238117622e-05,
+      "loss": 1.114,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948723424067754e-05,
+      "loss": 1.1161,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948254480900368e-05,
+      "loss": 1.1114,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947783408715815e-05,
+      "loss": 1.0506,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947310207614913e-05,
+      "loss": 1.019,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946834877698925e-05,
+      "loss": 1.0375,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994635741906958e-05,
+      "loss": 1.0609,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994587783182906e-05,
+      "loss": 1.0465,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994539611608e-05,
+      "loss": 1.1269,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944912271925493e-05,
+      "loss": 1.0966,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994442629946908e-05,
+      "loss": 1.0549,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994393819881477e-05,
+      "loss": 1.079,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994344797006702e-05,
+      "loss": 1.0306,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942955613330742e-05,
+      "loss": 1.0836,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942461128711302e-05,
+      "loss": 1.0883,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941964516314528e-05,
+      "loss": 0.9876,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99414657762467e-05,
+      "loss": 1.0638,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940964908614556e-05,
+      "loss": 1.0598,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940461913525274e-05,
+      "loss": 1.0306,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993995679108651e-05,
+      "loss": 1.064,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993944954140637e-05,
+      "loss": 1.1055,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938940164593392e-05,
+      "loss": 1.0893,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99384286607566e-05,
+      "loss": 1.0313,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993791503000546e-05,
+      "loss": 1.0813,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937399272449896e-05,
+      "loss": 1.065,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936881388200272e-05,
+      "loss": 1.0688,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936361377367436e-05,
+      "loss": 1.0859,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935839240062666e-05,
+      "loss": 1.0945,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99353149763977e-05,
+      "loss": 1.1321,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934788586484747e-05,
+      "loss": 1.0988,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993426007043645e-05,
+      "loss": 1.1128,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993372942836592e-05,
+      "loss": 1.1217,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933196660386723e-05,
+      "loss": 1.1085,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932661766612867e-05,
+      "loss": 1.0626,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932124747158832e-05,
+      "loss": 1.0105,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993158560213954e-05,
+      "loss": 1.0304,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931044331670377e-05,
+      "loss": 1.1548,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930500935867176e-05,
+      "loss": 1.0935,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992995541484623e-05,
+      "loss": 1.0465,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992940776872429e-05,
+      "loss": 1.1351,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992885799761855e-05,
+      "loss": 1.0887,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992830610164667e-05,
+      "loss": 1.0823,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927752080926762e-05,
+      "loss": 1.1161,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927195935577385e-05,
+      "loss": 1.0291,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926637665717567e-05,
+      "loss": 0.3289,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926077271466778e-05,
+      "loss": 1.0924,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925514752944954e-05,
+      "loss": 1.0682,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924950110272468e-05,
+      "loss": 1.0222,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992438334357017e-05,
+      "loss": 1.0808,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992381445295934e-05,
+      "loss": 1.1083,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992324343856174e-05,
+      "loss": 1.0672,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922670300499567e-05,
+      "loss": 1.0199,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922095038895475e-05,
+      "loss": 1.0102,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921517653872578e-05,
+      "loss": 1.1198,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920938145554438e-05,
+      "loss": 1.0585,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992035651406508e-05,
+      "loss": 1.0912,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991977275952898e-05,
+      "loss": 1.0934,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919186882071057e-05,
+      "loss": 1.0872,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918598881816705e-05,
+      "loss": 1.085,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918008758891752e-05,
+      "loss": 1.0183,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917416513422496e-05,
+      "loss": 1.0572,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916822145535682e-05,
+      "loss": 1.1213,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991622565535851e-05,
+      "loss": 1.1134,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991562704301863e-05,
+      "loss": 1.0011,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915026308644162e-05,
+      "loss": 1.0525,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914423452363654e-05,
+      "loss": 1.0903,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913818474306133e-05,
+      "loss": 1.0933,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913211374601064e-05,
+      "loss": 1.0394,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912602153378377e-05,
+      "loss": 1.0765,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911990810768446e-05,
+      "loss": 1.0719,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911377346902107e-05,
+      "loss": 1.0776,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910761761910645e-05,
+      "loss": 1.0419,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910144055925806e-05,
+      "loss": 1.0325,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909524229079775e-05,
+      "loss": 1.062,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908902281505207e-05,
+      "loss": 1.0601,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908278213335202e-05,
+      "loss": 1.081,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907652024703324e-05,
+      "loss": 1.0974,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907023715743573e-05,
+      "loss": 1.1004,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906393286590417e-05,
+      "loss": 1.0086,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990576073737877e-05,
+      "loss": 1.0405,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905126068244013e-05,
+      "loss": 1.1139,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904489279321964e-05,
+      "loss": 1.0643,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903850370748897e-05,
+      "loss": 1.0529,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903209342661555e-05,
+      "loss": 1.0784,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990256619519711e-05,
+      "loss": 1.0466,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990192092849322e-05,
+      "loss": 1.0701,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901273542687963e-05,
+      "loss": 1.063,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990062403791989e-05,
+      "loss": 1.0394,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899972414328e-05,
+      "loss": 1.0436,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989931867205175e-05,
+      "loss": 1.0714,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898662811231042e-05,
+      "loss": 1.055,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989800483200624e-05,
+      "loss": 1.0625,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897344734518156e-05,
+      "loss": 1.0974,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896682518908057e-05,
+      "loss": 1.0369,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896018185317666e-05,
+      "loss": 1.1203,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895351733889148e-05,
+      "loss": 1.0672,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894683164765138e-05,
+      "loss": 1.0456,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894012478088715e-05,
+      "loss": 1.1193,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893339674003406e-05,
+      "loss": 1.0096,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892664752653204e-05,
+      "loss": 1.0479,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891987714182544e-05,
+      "loss": 1.0084,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891308558736324e-05,
+      "loss": 1.1091,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989062728645988e-05,
+      "loss": 1.1086,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889943897499023e-05,
+      "loss": 1.078,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988925839199999e-05,
+      "loss": 1.0364,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98885707701095e-05,
+      "loss": 1.0384,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98878810319747e-05,
+      "loss": 1.0519,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98871891777432e-05,
+      "loss": 1.0451,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886495207563072e-05,
+      "loss": 1.1076,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885799121582824e-05,
+      "loss": 1.0981,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988510091995143e-05,
+      "loss": 0.392,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884400602818308e-05,
+      "loss": 1.0276,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883698170333334e-05,
+      "loss": 1.0858,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988299362264683e-05,
+      "loss": 1.0783,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882286959909584e-05,
+      "loss": 0.3669,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881578182272823e-05,
+      "loss": 1.0939,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880867289888232e-05,
+      "loss": 1.0684,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880154282907948e-05,
+      "loss": 1.0835,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879439161484563e-05,
+      "loss": 1.0117,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987872192577112e-05,
+      "loss": 1.0574,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878002575921113e-05,
+      "loss": 1.1067,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987728111208849e-05,
+      "loss": 1.0932,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987655753442765e-05,
+      "loss": 1.0198,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875831843093443e-05,
+      "loss": 1.0463,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987510403824118e-05,
+      "loss": 1.002,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874374120026608e-05,
+      "loss": 1.0267,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873642088605945e-05,
+      "loss": 1.019,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872907944135852e-05,
+      "loss": 1.0886,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872171686773436e-05,
+      "loss": 1.041,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871433316676265e-05,
+      "loss": 1.1126,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870692834002363e-05,
+      "loss": 1.0367,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869950238910194e-05,
+      "loss": 1.0603,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986920553155868e-05,
+      "loss": 1.0629,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98684587121072e-05,
+      "loss": 1.0731,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867709780715578e-05,
+      "loss": 1.0536,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866958737544086e-05,
+      "loss": 1.1063,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866205582753464e-05,
+      "loss": 1.0578,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865450316504887e-05,
+      "loss": 1.0509,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864692938959994e-05,
+      "loss": 1.0404,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863933450280863e-05,
+      "loss": 1.0428,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863171850630043e-05,
+      "loss": 1.0266,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862408140170512e-05,
+      "loss": 1.1149,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986164231906572e-05,
+      "loss": 1.0558,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986087438747955e-05,
+      "loss": 1.038,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986010434557635e-05,
+      "loss": 1.0575,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859332193520925e-05,
+      "loss": 1.083,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985855793147851e-05,
+      "loss": 1.0369,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857781559614813e-05,
+      "loss": 1.1174,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857003078095978e-05,
+      "loss": 1.0182,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856222487088615e-05,
+      "loss": 1.0483,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855439786759766e-05,
+      "loss": 1.0746,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854654977276947e-05,
+      "loss": 1.1146,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985386805880811e-05,
+      "loss": 1.0576,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853079031521664e-05,
+      "loss": 1.04,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852287895586466e-05,
+      "loss": 1.0923,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851494651171826e-05,
+      "loss": 1.07,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985069929844751e-05,
+      "loss": 1.0535,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849901837583726e-05,
+      "loss": 1.0771,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849102268751138e-05,
+      "loss": 1.0255,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848300592120863e-05,
+      "loss": 1.0672,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984749680786447e-05,
+      "loss": 1.0622,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846690916153967e-05,
+      "loss": 1.1097,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984588291716183e-05,
+      "loss": 1.091,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984507281106098e-05,
+      "loss": 1.0618,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984426059802478e-05,
+      "loss": 1.0531,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843446278227053e-05,
+      "loss": 1.0392,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842629851842073e-05,
+      "loss": 1.051,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841811319044562e-05,
+      "loss": 1.1347,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840990680009693e-05,
+      "loss": 0.3622,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984016793491309e-05,
+      "loss": 1.0955,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983934308393083e-05,
+      "loss": 1.111,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838516127239436e-05,
+      "loss": 1.0027,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837687065015883e-05,
+      "loss": 1.0394,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836855897437605e-05,
+      "loss": 0.3294,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836022624682473e-05,
+      "loss": 1.044,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835187246928814e-05,
+      "loss": 1.0835,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983434976435541e-05,
+      "loss": 1.0621,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833510177141494e-05,
+      "loss": 1.0369,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832668485466735e-05,
+      "loss": 1.0859,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831824689511268e-05,
+      "loss": 1.0992,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830978789455676e-05,
+      "loss": 1.1055,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983013078548098e-05,
+      "loss": 1.0701,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982928067776867e-05,
+      "loss": 1.0978,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828428466500676e-05,
+      "loss": 1.0167,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827574151859376e-05,
+      "loss": 1.0427,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826717734027603e-05,
+      "loss": 1.0602,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825859213188635e-05,
+      "loss": 1.0974,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824998589526205e-05,
+      "loss": 0.3361,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824135863224495e-05,
+      "loss": 1.0539,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823271034468137e-05,
+      "loss": 1.0348,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982240410344221e-05,
+      "loss": 1.0786,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821535070332248e-05,
+      "loss": 1.0615,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982066393532423e-05,
+      "loss": 1.0469,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819790698604584e-05,
+      "loss": 1.0791,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818915360360198e-05,
+      "loss": 1.0973,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818037920778397e-05,
+      "loss": 1.0951,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981715838004696e-05,
+      "loss": 1.0647,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981627673835412e-05,
+      "loss": 1.0376,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815392995888554e-05,
+      "loss": 1.0375,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814507152839395e-05,
+      "loss": 1.0574,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981361920939622e-05,
+      "loss": 1.0807,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981272916574905e-05,
+      "loss": 1.0837,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811837022088373e-05,
+      "loss": 1.0635,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810942778605108e-05,
+      "loss": 1.0453,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810046435490634e-05,
+      "loss": 1.0695,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980914799293678e-05,
+      "loss": 1.1008,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980824745113582e-05,
+      "loss": 1.0851,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980734481028047e-05,
+      "loss": 1.066,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980644007056391e-05,
+      "loss": 1.0274,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805533232179766e-05,
+      "loss": 1.0353,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804624295322102e-05,
+      "loss": 1.1366,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803713260185445e-05,
+      "loss": 1.096,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980280012696476e-05,
+      "loss": 0.3814,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980188489585547e-05,
+      "loss": 1.035,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980096756705344e-05,
+      "loss": 1.0759,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800048140754987e-05,
+      "loss": 1.0785,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799126617156875e-05,
+      "loss": 1.0879,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798202996456324e-05,
+      "loss": 1.0229,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797277278850993e-05,
+      "loss": 1.1072,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796349464538994e-05,
+      "loss": 1.1046,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979541955371889e-05,
+      "loss": 0.9992,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979448754658969e-05,
+      "loss": 1.063,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793553443350846e-05,
+      "loss": 1.0643,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792617244202273e-05,
+      "loss": 1.0831,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791678949344324e-05,
+      "loss": 1.0821,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790738558977802e-05,
+      "loss": 1.0554,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789796073303953e-05,
+      "loss": 1.0394,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788851492524487e-05,
+      "loss": 1.0379,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978790481684155e-05,
+      "loss": 1.1069,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786956046457736e-05,
+      "loss": 1.0896,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786005181576095e-05,
+      "loss": 1.0329,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785052222400115e-05,
+      "loss": 1.0464,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978409716913374e-05,
+      "loss": 1.0213,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978314002198137e-05,
+      "loss": 1.0765,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782180781147828e-05,
+      "loss": 1.0723,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978121944683841e-05,
+      "loss": 1.0239,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780256019258846e-05,
+      "loss": 1.0381,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779290498615317e-05,
+      "loss": 1.0625,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778322885114458e-05,
+      "loss": 1.0584,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777353178963347e-05,
+      "loss": 1.083,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776381380369507e-05,
+      "loss": 1.1288,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775407489540913e-05,
+      "loss": 1.0374,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977443150668598e-05,
+      "loss": 1.0741,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977345343201359e-05,
+      "loss": 0.9994,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772473265733054e-05,
+      "loss": 1.0098,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771491008054132e-05,
+      "loss": 1.0406,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977050665918704e-05,
+      "loss": 1.0377,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976952021934244e-05,
+      "loss": 1.0075,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768531688731438e-05,
+      "loss": 1.0632,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767541067565584e-05,
+      "loss": 1.0613,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766548356056882e-05,
+      "loss": 1.0699,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765553554417786e-05,
+      "loss": 1.1873,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764556662861188e-05,
+      "loss": 1.0699,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763557681600434e-05,
+      "loss": 1.0565,
+      "step": 682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762556610849312e-05,
+      "loss": 1.013,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761553450822063e-05,
+      "loss": 0.9936,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760548201733374e-05,
+      "loss": 1.0334,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975954086379837e-05,
+      "loss": 1.0764,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758531437232643e-05,
+      "loss": 1.0363,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757519922252208e-05,
+      "loss": 1.0596,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756506319073543e-05,
+      "loss": 1.0308,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755490627913567e-05,
+      "loss": 1.0316,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975447284898965e-05,
+      "loss": 1.0594,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753452982519603e-05,
+      "loss": 0.3351,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752431028721687e-05,
+      "loss": 1.0342,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751406987814616e-05,
+      "loss": 1.0486,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975038086001753e-05,
+      "loss": 1.0301,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749352645550043e-05,
+      "loss": 1.0604,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97483223446322e-05,
+      "loss": 0.995,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747289957484485e-05,
+      "loss": 1.0797,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746255484327847e-05,
+      "loss": 1.0689,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745218925383676e-05,
+      "loss": 1.0909,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744180280873796e-05,
+      "loss": 1.1228,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743139551020493e-05,
+      "loss": 1.0612,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974209673604649e-05,
+      "loss": 1.0148,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741051836174956e-05,
+      "loss": 1.0029,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740004851629516e-05,
+      "loss": 1.0998,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738955782634227e-05,
+      "loss": 1.0066,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737904629413605e-05,
+      "loss": 1.0757,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736851392192606e-05,
+      "loss": 1.0462,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735796071196627e-05,
+      "loss": 1.1046,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973473866665152e-05,
+      "loss": 1.0274,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973367917878358e-05,
+      "loss": 1.1027,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732617607819548e-05,
+      "loss": 1.0319,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731553953986606e-05,
+      "loss": 1.048,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973048821751239e-05,
+      "loss": 1.0814,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729420398624973e-05,
+      "loss": 0.9957,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728350497552878e-05,
+      "loss": 1.0069,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727278514525078e-05,
+      "loss": 1.0144,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726204449770983e-05,
+      "loss": 1.036,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972512830352046e-05,
+      "loss": 1.0773,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724050076003802e-05,
+      "loss": 1.0939,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972296976745177e-05,
+      "loss": 1.0524,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972188737809555e-05,
+      "loss": 1.106,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720802908166793e-05,
+      "loss": 1.0827,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719716357897584e-05,
+      "loss": 1.0684,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971862772752045e-05,
+      "loss": 1.0241,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971753701726837e-05,
+      "loss": 1.0469,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716444227374763e-05,
+      "loss": 1.0038,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97153493580735e-05,
+      "loss": 1.0711,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714252409598896e-05,
+      "loss": 0.9934,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97131533821857e-05,
+      "loss": 1.0207,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712052276069118e-05,
+      "loss": 1.0362,
+      "step": 731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97109490914848e-05,
+      "loss": 1.0073,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970984382866883e-05,
+      "loss": 0.9979,
+      "step": 733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970873648785775e-05,
+      "loss": 1.0103,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707627069288537e-05,
+      "loss": 1.0561,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970651557319862e-05,
+      "loss": 0.979,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705401999825872e-05,
+      "loss": 1.0205,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97042863494086e-05,
+      "loss": 1.058,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703168622185572e-05,
+      "loss": 1.0666,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702048818395983e-05,
+      "loss": 1.0932,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700926938279488e-05,
+      "loss": 0.9818,
+      "step": 741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969980298207618e-05,
+      "loss": 0.9978,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969867695002659e-05,
+      "loss": 1.046,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96975488423717e-05,
+      "loss": 1.02,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969641865935294e-05,
+      "loss": 0.9944,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695286401212178e-05,
+      "loss": 1.0753,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694152068191726e-05,
+      "loss": 1.0722,
+      "step": 747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969301566053434e-05,
+      "loss": 1.0574,
+      "step": 748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691877178483226e-05,
+      "loss": 0.3819,
+      "step": 749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690736622282028e-05,
+      "loss": 1.053,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689593992174834e-05,
+      "loss": 1.0541,
+      "step": 751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688449288406177e-05,
+      "loss": 1.0647,
+      "step": 752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687302511221036e-05,
+      "loss": 1.0371,
+      "step": 753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686153660864832e-05,
+      "loss": 1.0443,
+      "step": 754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685002737583427e-05,
+      "loss": 1.0547,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683849741623133e-05,
+      "loss": 1.0898,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682694673230695e-05,
+      "loss": 1.0375,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681537532653314e-05,
+      "loss": 1.1312,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680378320138625e-05,
+      "loss": 0.348,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967921703593471e-05,
+      "loss": 1.0761,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678053680290102e-05,
+      "loss": 1.085,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967688825345376e-05,
+      "loss": 1.0737,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675720755675104e-05,
+      "loss": 1.1457,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674551187203984e-05,
+      "loss": 1.0378,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673379548290698e-05,
+      "loss": 1.0398,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672205839185988e-05,
+      "loss": 1.0385,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967103006014104e-05,
+      "loss": 1.0088,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669852211407482e-05,
+      "loss": 1.0255,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668672293237382e-05,
+      "loss": 1.082,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667490305883256e-05,
+      "loss": 1.1414,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666306249598056e-05,
+      "loss": 1.0875,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665120124635186e-05,
+      "loss": 1.1265,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663931931248485e-05,
+      "loss": 1.041,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662741669692234e-05,
+      "loss": 1.0843,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661549340221163e-05,
+      "loss": 1.0611,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660354943090444e-05,
+      "loss": 0.9514,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659158478555683e-05,
+      "loss": 1.0096,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965795994687294e-05,
+      "loss": 0.3441,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965675934829871e-05,
+      "loss": 1.024,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965555668308993e-05,
+      "loss": 1.0507,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965435195150398e-05,
+      "loss": 1.0123,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965314515379869e-05,
+      "loss": 1.0181,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965193629023232e-05,
+      "loss": 1.0361,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965072536106358e-05,
+      "loss": 1.0571,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964951236655162e-05,
+      "loss": 0.9943,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648297306956027e-05,
+      "loss": 1.0605,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647080182536843e-05,
+      "loss": 1.0559,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964586099355454e-05,
+      "loss": 1.0162,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644639740270036e-05,
+      "loss": 0.3271,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964341642294469e-05,
+      "loss": 0.9943,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642191041840302e-05,
+      "loss": 1.1132,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640963597219115e-05,
+      "loss": 1.0841,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639734089343813e-05,
+      "loss": 1.0074,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638502518477524e-05,
+      "loss": 1.0745,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637268884883816e-05,
+      "loss": 1.0401,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636033188826697e-05,
+      "loss": 0.3185,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963479543057061e-05,
+      "loss": 1.0169,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633555610380458e-05,
+      "loss": 0.968,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963231372852157e-05,
+      "loss": 1.0241,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631069785259716e-05,
+      "loss": 1.0763,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629823780861113e-05,
+      "loss": 1.0898,
+      "step": 801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962857571559242e-05,
+      "loss": 1.0789,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627325589720732e-05,
+      "loss": 1.0845,
+      "step": 803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962607340351359e-05,
+      "loss": 1.021,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962481915723897e-05,
+      "loss": 1.0503,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962356285116529e-05,
+      "loss": 0.9677,
+      "step": 806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622304485561418e-05,
+      "loss": 1.0034,
+      "step": 807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621044060696647e-05,
+      "loss": 1.0342,
+      "step": 808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961978157684073e-05,
+      "loss": 1.0015,
+      "step": 809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618517034263843e-05,
+      "loss": 1.0462,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617250433236613e-05,
+      "loss": 1.0872,
+      "step": 811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615981774030096e-05,
+      "loss": 1.0203,
+      "step": 812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961471105691581e-05,
+      "loss": 1.0643,
+      "step": 813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613438282165687e-05,
+      "loss": 1.0573,
+      "step": 814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612163450052122e-05,
+      "loss": 1.0259,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961088656084794e-05,
+      "loss": 1.0052,
+      "step": 816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96096076148264e-05,
+      "loss": 1.013,
+      "step": 817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960832661226121e-05,
+      "loss": 1.0313,
+      "step": 818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960704355342652e-05,
+      "loss": 1.0034,
+      "step": 819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605758438596916e-05,
+      "loss": 1.0511,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604471268047424e-05,
+      "loss": 1.0436,
+      "step": 821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603182042053505e-05,
+      "loss": 1.058,
+      "step": 822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601890760891074e-05,
+      "loss": 0.995,
+      "step": 823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600597424836468e-05,
+      "loss": 1.0313,
+      "step": 824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959930203416648e-05,
+      "loss": 1.0329,
+      "step": 825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959800458915833e-05,
+      "loss": 1.0872,
+      "step": 826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596705090089688e-05,
+      "loss": 1.0548,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595403537238657e-05,
+      "loss": 1.0884,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594099930883774e-05,
+      "loss": 1.0427,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592794271304033e-05,
+      "loss": 1.007,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591486558778855e-05,
+      "loss": 1.0208,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590176793588098e-05,
+      "loss": 1.0099,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588864976012063e-05,
+      "loss": 1.0636,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587551106331496e-05,
+      "loss": 1.0659,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586235184827574e-05,
+      "loss": 1.1025,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584917211781923e-05,
+      "loss": 1.1143,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958359718747659e-05,
+      "loss": 1.0745,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582275112194076e-05,
+      "loss": 1.0931,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580950986217324e-05,
+      "loss": 1.1195,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579624809829702e-05,
+      "loss": 1.0094,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578296583315027e-05,
+      "loss": 1.1006,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576966306957546e-05,
+      "loss": 1.0566,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957563398104196e-05,
+      "loss": 1.0492,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574299605853394e-05,
+      "loss": 0.9833,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572963181677413e-05,
+      "loss": 1.0707,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571624708800026e-05,
+      "loss": 1.1119,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570284187507678e-05,
+      "loss": 1.0504,
+      "step": 847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568941618087257e-05,
+      "loss": 1.0359,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956759700082608e-05,
+      "loss": 1.033,
+      "step": 849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956625033601191e-05,
+      "loss": 1.0484,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956490162393295e-05,
+      "loss": 1.0905,
+      "step": 851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563550864877828e-05,
+      "loss": 1.0061,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956219805913562e-05,
+      "loss": 1.1038,
+      "step": 853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560843206995847e-05,
+      "loss": 1.0641,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955948630874845e-05,
+      "loss": 1.0842,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955812736468382e-05,
+      "loss": 1.064,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955676637509279e-05,
+      "loss": 1.0733,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555403340266614e-05,
+      "loss": 1.0442,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554038260497004e-05,
+      "loss": 1.0682,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552671136076088e-05,
+      "loss": 1.0381,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551301967296457e-05,
+      "loss": 1.0272,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549930754451114e-05,
+      "loss": 1.0126,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954855749783352e-05,
+      "loss": 1.0851,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547182197737553e-05,
+      "loss": 1.0394,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954580485445755e-05,
+      "loss": 1.0109,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954442546828827e-05,
+      "loss": 1.0789,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543044039524915e-05,
+      "loss": 1.0134,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954166056846313e-05,
+      "loss": 1.0429,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540275055398982e-05,
+      "loss": 1.0961,
+      "step": 869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538887500628982e-05,
+      "loss": 1.0061,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537497904450088e-05,
+      "loss": 1.0573,
+      "step": 871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953610626715968e-05,
+      "loss": 1.0557,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953471258905558e-05,
+      "loss": 1.0563,
+      "step": 873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533316870436053e-05,
+      "loss": 1.0296,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531919111599796e-05,
+      "loss": 1.0087,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530519312845935e-05,
+      "loss": 1.0545,
+      "step": 876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529117474474046e-05,
+      "loss": 1.0546,
+      "step": 877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952771359678413e-05,
+      "loss": 1.0217,
+      "step": 878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526307680076633e-05,
+      "loss": 0.9929,
+      "step": 879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524899724652435e-05,
+      "loss": 0.9979,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952348973081285e-05,
+      "loss": 1.0248,
+      "step": 881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522077698859632e-05,
+      "loss": 1.0201,
+      "step": 882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952066362909496e-05,
+      "loss": 1.0057,
+      "step": 883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951924752182147e-05,
+      "loss": 1.003,
+      "step": 884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517829377342215e-05,
+      "loss": 1.0696,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951640919596069e-05,
+      "loss": 1.0064,
+      "step": 886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951498697798083e-05,
+      "loss": 1.0388,
+      "step": 887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513562723706998e-05,
+      "loss": 1.0109,
+      "step": 888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512136433444002e-05,
+      "loss": 0.9826,
+      "step": 889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951070810749708e-05,
+      "loss": 0.9904,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509277746171906e-05,
+      "loss": 1.0525,
+      "step": 891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950784534977459e-05,
+      "loss": 1.0572,
+      "step": 892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950641091861168e-05,
+      "loss": 1.0238,
+      "step": 893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950497445299016e-05,
+      "loss": 1.058,
+      "step": 894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503535953217438e-05,
+      "loss": 1.0373,
+      "step": 895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950209541960137e-05,
+      "loss": 1.0277,
+      "step": 896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500652852450245e-05,
+      "loss": 1.0491,
+      "step": 897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499208252072787e-05,
+      "loss": 0.9765,
+      "step": 898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949776161877815e-05,
+      "loss": 0.983,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949631295287593e-05,
+      "loss": 0.9931,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494862254676147e-05,
+      "loss": 1.0734,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493409524489272e-05,
+      "loss": 1.0409,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491954762626202e-05,
+      "loss": 1.0261,
+      "step": 903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949049796939827e-05,
+      "loss": 1.0721,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948903914511723e-05,
+      "loss": 0.3828,
+      "step": 905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487578290095302e-05,
+      "loss": 1.0009,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486115404645116e-05,
+      "loss": 1.0448,
+      "step": 907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484650489079734e-05,
+      "loss": 1.043,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483183543712672e-05,
+      "loss": 1.0626,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481714568857866e-05,
+      "loss": 0.921,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480243564829692e-05,
+      "loss": 1.0467,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947877053194295e-05,
+      "loss": 1.017,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477295470512894e-05,
+      "loss": 0.9957,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475818380855195e-05,
+      "loss": 1.0453,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474339263285962e-05,
+      "loss": 0.3686,
+      "step": 915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472858118121746e-05,
+      "loss": 1.0509,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947137494567952e-05,
+      "loss": 1.0638,
+      "step": 917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469889746276692e-05,
+      "loss": 0.9831,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946840252023112e-05,
+      "loss": 1.0696,
+      "step": 919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466913267861075e-05,
+      "loss": 1.0309,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465421989485274e-05,
+      "loss": 1.0197,
+      "step": 921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946392868542286e-05,
+      "loss": 1.0901,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462433355993425e-05,
+      "loss": 0.3694,
+      "step": 923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946093600151697e-05,
+      "loss": 1.0464,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459436622313948e-05,
+      "loss": 1.0505,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457935218705243e-05,
+      "loss": 1.0344,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456431791012164e-05,
+      "loss": 0.9962,
+      "step": 927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454926339556457e-05,
+      "loss": 1.0481,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945341886466031e-05,
+      "loss": 0.9947,
+      "step": 929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945190936664633e-05,
+      "loss": 0.329,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450397845837565e-05,
+      "loss": 0.964,
+      "step": 931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944888430255749e-05,
+      "loss": 1.0259,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447368737130028e-05,
+      "loss": 1.0262,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445851149879514e-05,
+      "loss": 1.0203,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944433154113073e-05,
+      "loss": 1.0,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442809911208884e-05,
+      "loss": 0.9712,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944128626043962e-05,
+      "loss": 1.0827,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439760589149013e-05,
+      "loss": 1.0731,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943823289766357e-05,
+      "loss": 1.1119,
+      "step": 939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943670318631023e-05,
+      "loss": 0.3429,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435171455416365e-05,
+      "loss": 1.0515,
+      "step": 941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433637705309784e-05,
+      "loss": 1.0345,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943210193631872e-05,
+      "loss": 0.9953,
+      "step": 943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943056414877184e-05,
+      "loss": 1.0258,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429024342998246e-05,
+      "loss": 1.0584,
+      "step": 945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942748251932747e-05,
+      "loss": 1.0753,
+      "step": 946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942593867808948e-05,
+      "loss": 1.0428,
+      "step": 947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942439281961467e-05,
+      "loss": 1.0121,
+      "step": 948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422844944233864e-05,
+      "loss": 1.0502,
+      "step": 949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421295052278328e-05,
+      "loss": 0.9779,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419743144079743e-05,
+      "loss": 0.9995,
+      "step": 951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941818921997024e-05,
+      "loss": 1.0181,
+      "step": 952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941663328028238e-05,
+      "loss": 1.0928,
+      "step": 953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415075325349128e-05,
+      "loss": 1.0427,
+      "step": 954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413515355503916e-05,
+      "loss": 1.0434,
+      "step": 955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411953371080585e-05,
+      "loss": 1.0382,
+      "step": 956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941038937241342e-05,
+      "loss": 0.9822,
+      "step": 957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940882335983713e-05,
+      "loss": 1.071,
+      "step": 958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407255333686844e-05,
+      "loss": 1.0186,
+      "step": 959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940568529429815e-05,
+      "loss": 1.026,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404113242007042e-05,
+      "loss": 0.9947,
+      "step": 961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940253917714996e-05,
+      "loss": 1.0842,
+      "step": 962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400963100063758e-05,
+      "loss": 1.0529,
+      "step": 963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939938501108574e-05,
+      "loss": 1.0422,
+      "step": 964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397804910553625e-05,
+      "loss": 1.0316,
+      "step": 965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396222798805574e-05,
+      "loss": 0.9904,
+      "step": 966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939463867618017e-05,
+      "loss": 1.038,
+      "step": 967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393052543016435e-05,
+      "loss": 1.0625,
+      "step": 968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391464399653808e-05,
+      "loss": 1.0436,
+      "step": 969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389874246432172e-05,
+      "loss": 1.0432,
+      "step": 970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938828208369183e-05,
+      "loss": 1.0621,
+      "step": 971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386687911773525e-05,
+      "loss": 1.0169,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385091731018416e-05,
+      "loss": 1.0188,
+      "step": 973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383493541768113e-05,
+      "loss": 1.0552,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938189334436463e-05,
+      "loss": 1.0387,
+      "step": 975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938029113915043e-05,
+      "loss": 1.0031,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378686926468398e-05,
+      "loss": 1.0419,
+      "step": 977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377080706661854e-05,
+      "loss": 1.051,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937547248007454e-05,
+      "loss": 0.3748,
+      "step": 979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373862247050627e-05,
+      "loss": 1.0368,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372250007934728e-05,
+      "loss": 1.0585,
+      "step": 981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370635763071873e-05,
+      "loss": 1.0621,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369019512807525e-05,
+      "loss": 1.0317,
+      "step": 983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367401257487577e-05,
+      "loss": 0.9884,
+      "step": 984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365780997458347e-05,
+      "loss": 1.0489,
+      "step": 985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364158733066594e-05,
+      "loss": 1.0415,
+      "step": 986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362534464659486e-05,
+      "loss": 1.0091,
+      "step": 987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936090819258464e-05,
+      "loss": 1.0181,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935927991719009e-05,
+      "loss": 1.0041,
+      "step": 989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357649638824303e-05,
+      "loss": 1.1004,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935601735783617e-05,
+      "loss": 1.0349,
+      "step": 991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354383074575016e-05,
+      "loss": 1.0654,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352746789390594e-05,
+      "loss": 1.0852,
+      "step": 993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935110850263308e-05,
+      "loss": 1.06,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349468214653088e-05,
+      "loss": 1.0149,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347825925801654e-05,
+      "loss": 1.0397,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346181636430235e-05,
+      "loss": 1.0554,
+      "step": 997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344535346890732e-05,
+      "loss": 0.9544,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342887057535464e-05,
+      "loss": 1.0763,
+      "step": 999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934123676871718e-05,
+      "loss": 1.0063,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339584480789053e-05,
+      "loss": 0.9815,
+      "step": 1001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337930194104693e-05,
+      "loss": 1.0171,
+      "step": 1002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933627390901813e-05,
+      "loss": 1.1015,
+      "step": 1003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334615625883826e-05,
+      "loss": 0.9977,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332955345056662e-05,
+      "loss": 1.0312,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331293066891964e-05,
+      "loss": 1.0036,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329628791745466e-05,
+      "loss": 1.0651,
+      "step": 1007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932796251997334e-05,
+      "loss": 1.0204,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326294251932185e-05,
+      "loss": 1.0307,
+      "step": 1009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324623987979024e-05,
+      "loss": 1.0004,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322951728471313e-05,
+      "loss": 1.0633,
+      "step": 1011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321277473766925e-05,
+      "loss": 1.016,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931960122422417e-05,
+      "loss": 0.9836,
+      "step": 1013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317922980201777e-05,
+      "loss": 1.0352,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931624274205891e-05,
+      "loss": 0.3617,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931456051015515e-05,
+      "loss": 1.0772,
+      "step": 1016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312876284850516e-05,
+      "loss": 1.0449,
+      "step": 1017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931119006650544e-05,
+      "loss": 0.9751,
+      "step": 1018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309501855480795e-05,
+      "loss": 1.0602,
+      "step": 1019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307811652137868e-05,
+      "loss": 0.9593,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306119456838384e-05,
+      "loss": 1.0409,
+      "step": 1021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304425269944484e-05,
+      "loss": 0.9831,
+      "step": 1022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930272909181874e-05,
+      "loss": 1.0327,
+      "step": 1023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930103092282415e-05,
+      "loss": 1.0503,
+      "step": 1024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929933076332414e-05,
+      "loss": 1.0629,
+      "step": 1025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297628613682556e-05,
+      "loss": 1.0426,
+      "step": 1026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295924474263673e-05,
+      "loss": 1.0097,
+      "step": 1027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294218345432198e-05,
+      "loss": 1.019,
+      "step": 1028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292510227553253e-05,
+      "loss": 0.3722,
+      "step": 1029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929080012099239e-05,
+      "loss": 1.0424,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928908802611559e-05,
+      "loss": 1.043,
+      "step": 1031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287373943289257e-05,
+      "loss": 1.0053,
+      "step": 1032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285657872880222e-05,
+      "loss": 1.0247,
+      "step": 1033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283939815255737e-05,
+      "loss": 1.0554,
+      "step": 1034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928221977078348e-05,
+      "loss": 1.0206,
+      "step": 1035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928049773983156e-05,
+      "loss": 1.0301,
+      "step": 1036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278773722768506e-05,
+      "loss": 1.0101,
+      "step": 1037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277047719963274e-05,
+      "loss": 0.9933,
+      "step": 1038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275319731785245e-05,
+      "loss": 1.0327,
+      "step": 1039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927358975860422e-05,
+      "loss": 0.9611,
+      "step": 1040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271857800790434e-05,
+      "loss": 1.0221,
+      "step": 1041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927012385871454e-05,
+      "loss": 1.0195,
+      "step": 1042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268387932747617e-05,
+      "loss": 0.9669,
+      "step": 1043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926665002326117e-05,
+      "loss": 1.0709,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264910130627124e-05,
+      "loss": 1.0893,
+      "step": 1045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926316825521784e-05,
+      "loss": 1.026,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261424397406084e-05,
+      "loss": 1.0427,
+      "step": 1047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259678557565066e-05,
+      "loss": 1.0376,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925793073606841e-05,
+      "loss": 1.0487,
+      "step": 1049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925618093329016e-05,
+      "loss": 1.0056,
+      "step": 1050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254429149604798e-05,
+      "loss": 1.0482,
+      "step": 1051
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252675385387215e-05,
+      "loss": 1.06,
+      "step": 1052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250919641012743e-05,
+      "loss": 1.0786,
+      "step": 1053
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249161916857114e-05,
+      "loss": 1.0102,
+      "step": 1054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247402213296503e-05,
+      "loss": 1.0603,
+      "step": 1055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245640530707504e-05,
+      "loss": 1.1372,
+      "step": 1056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924387686946713e-05,
+      "loss": 1.0244,
+      "step": 1057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242111229952823e-05,
+      "loss": 0.9765,
+      "step": 1058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240343612542447e-05,
+      "loss": 1.0594,
+      "step": 1059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238574017614285e-05,
+      "loss": 1.0605,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236802445547046e-05,
+      "loss": 0.989,
+      "step": 1061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923502889671987e-05,
+      "loss": 0.9716,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233253371512302e-05,
+      "loss": 1.0797,
+      "step": 1063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923147587030433e-05,
+      "loss": 0.9945,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229696393476347e-05,
+      "loss": 1.0515,
+      "step": 1065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227914941409183e-05,
+      "loss": 1.0571,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226131514484082e-05,
+      "loss": 1.0542,
+      "step": 1067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224346113082718e-05,
+      "loss": 1.0863,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922255873758718e-05,
+      "loss": 1.0109,
+      "step": 1069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220769388379977e-05,
+      "loss": 1.0422,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921897806584406e-05,
+      "loss": 1.0097,
+      "step": 1071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217184770362777e-05,
+      "loss": 1.0646,
+      "step": 1072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921538950231991e-05,
+      "loss": 0.9894,
+      "step": 1073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213592262099668e-05,
+      "loss": 0.9767,
+      "step": 1074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211793050086673e-05,
+      "loss": 1.0897,
+      "step": 1075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209991866665974e-05,
+      "loss": 1.0443,
+      "step": 1076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208188712223043e-05,
+      "loss": 1.0383,
+      "step": 1077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920638358714377e-05,
+      "loss": 1.0288,
+      "step": 1078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204576491814468e-05,
+      "loss": 1.0959,
+      "step": 1079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202767426621867e-05,
+      "loss": 0.9997,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920095639195313e-05,
+      "loss": 1.0458,
+      "step": 1081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199143388195835e-05,
+      "loss": 1.0771,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197328415737977e-05,
+      "loss": 0.3736,
+      "step": 1083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919551147496798e-05,
+      "loss": 1.0617,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193692566274682e-05,
+      "loss": 1.0357,
+      "step": 1085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191871690047353e-05,
+      "loss": 0.9818,
+      "step": 1086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190048846675668e-05,
+      "loss": 1.0538,
+      "step": 1087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188224036549742e-05,
+      "loss": 0.9952,
+      "step": 1088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918639726006009e-05,
+      "loss": 1.0912,
+      "step": 1089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184568517597668e-05,
+      "loss": 1.0418,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182737809553843e-05,
+      "loss": 1.0433,
+      "step": 1091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180905136320397e-05,
+      "loss": 1.0445,
+      "step": 1092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917907049828954e-05,
+      "loss": 1.0061,
+      "step": 1093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177233895853907e-05,
+      "loss": 1.0909,
+      "step": 1094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175395329406545e-05,
+      "loss": 1.0736,
+      "step": 1095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173554799340923e-05,
+      "loss": 1.0384,
+      "step": 1096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171712306050928e-05,
+      "loss": 1.0523,
+      "step": 1097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169867849930878e-05,
+      "loss": 1.0227,
+      "step": 1098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168021431375497e-05,
+      "loss": 1.0229,
+      "step": 1099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166173050779936e-05,
+      "loss": 1.0997,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916432270853977e-05,
+      "loss": 0.9877,
+      "step": 1101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916247040505099e-05,
+      "loss": 1.0277,
+      "step": 1102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160616140709995e-05,
+      "loss": 1.0301,
+      "step": 1103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158759915913624e-05,
+      "loss": 1.0046,
+      "step": 1104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915690173105912e-05,
+      "loss": 1.0603,
+      "step": 1105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155041586544157e-05,
+      "loss": 1.0224,
+      "step": 1106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153179482766825e-05,
+      "loss": 0.9636,
+      "step": 1107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151315420125623e-05,
+      "loss": 1.0824,
+      "step": 1108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149449399019482e-05,
+      "loss": 1.0061,
+      "step": 1109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147581419847748e-05,
+      "loss": 1.0062,
+      "step": 1110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914571148301018e-05,
+      "loss": 1.0843,
+      "step": 1111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143839588906973e-05,
+      "loss": 0.9973,
+      "step": 1112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141965737938717e-05,
+      "loss": 1.0374,
+      "step": 1113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914008993050644e-05,
+      "loss": 1.0186,
+      "step": 1114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138212167011582e-05,
+      "loss": 1.0386,
+      "step": 1115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136332447855995e-05,
+      "loss": 0.9922,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134450773441967e-05,
+      "loss": 1.0083,
+      "step": 1117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132567144172184e-05,
+      "loss": 1.1002,
+      "step": 1118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130681560449766e-05,
+      "loss": 1.0811,
+      "step": 1119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128794022678242e-05,
+      "loss": 0.9974,
+      "step": 1120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912690453126156e-05,
+      "loss": 1.0756,
+      "step": 1121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125013086604094e-05,
+      "loss": 1.0243,
+      "step": 1122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912311968911063e-05,
+      "loss": 0.9704,
+      "step": 1123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121224339186363e-05,
+      "loss": 0.3571,
+      "step": 1124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119327037236926e-05,
+      "loss": 1.0272,
+      "step": 1125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117427783668355e-05,
+      "loss": 0.9867,
+      "step": 1126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911552657888711e-05,
+      "loss": 1.0357,
+      "step": 1127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911362342330006e-05,
+      "loss": 1.0048,
+      "step": 1128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91117183173145e-05,
+      "loss": 1.0892,
+      "step": 1129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109811261338143e-05,
+      "loss": 1.0096,
+      "step": 1130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107902255779118e-05,
+      "loss": 0.9503,
+      "step": 1131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910599130104596e-05,
+      "loss": 1.0468,
+      "step": 1132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910407839754764e-05,
+      "loss": 1.0038,
+      "step": 1133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102163545693536e-05,
+      "loss": 1.0389,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100246745893442e-05,
+      "loss": 1.0822,
+      "step": 1135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909832799855757e-05,
+      "loss": 0.983,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096407304096546e-05,
+      "loss": 0.9999,
+      "step": 1137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094484662921423e-05,
+      "loss": 1.0203,
+      "step": 1138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092560075443664e-05,
+      "loss": 1.0501,
+      "step": 1139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909063354207514e-05,
+      "loss": 0.9897,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908870506322815e-05,
+      "loss": 1.0213,
+      "step": 1141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086774639315407e-05,
+      "loss": 1.0215,
+      "step": 1142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084842270750045e-05,
+      "loss": 1.012,
+      "step": 1143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082907957945602e-05,
+      "loss": 1.0714,
+      "step": 1144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908097170131604e-05,
+      "loss": 1.0587,
+      "step": 1145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079033501275733e-05,
+      "loss": 1.0194,
+      "step": 1146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907709335823948e-05,
+      "loss": 0.9956,
+      "step": 1147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075151272622483e-05,
+      "loss": 0.9945,
+      "step": 1148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073207244840367e-05,
+      "loss": 1.0544,
+      "step": 1149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071261275309176e-05,
+      "loss": 1.0466,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906931336444536e-05,
+      "loss": 1.0019,
+      "step": 1151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067363512665795e-05,
+      "loss": 1.0677,
+      "step": 1152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906541172038776e-05,
+      "loss": 1.0229,
+      "step": 1153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063457988028964e-05,
+      "loss": 0.9848,
+      "step": 1154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061502316007517e-05,
+      "loss": 1.0212,
+      "step": 1155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059544704741956e-05,
+      "loss": 1.0598,
+      "step": 1156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057585154651222e-05,
+      "loss": 1.0546,
+      "step": 1157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055623666154683e-05,
+      "loss": 1.077,
+      "step": 1158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905366023967211e-05,
+      "loss": 1.0735,
+      "step": 1159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90516948756237e-05,
+      "loss": 1.0415,
+      "step": 1160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049727574430052e-05,
+      "loss": 1.0502,
+      "step": 1161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047758336512193e-05,
+      "loss": 1.0407,
+      "step": 1162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045787162291548e-05,
+      "loss": 1.0671,
+      "step": 1163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904381405218998e-05,
+      "loss": 0.9658,
+      "step": 1164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041839006629738e-05,
+      "loss": 1.0807,
+      "step": 1165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039862026033513e-05,
+      "loss": 1.0584,
+      "step": 1166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037883110824388e-05,
+      "loss": 1.0492,
+      "step": 1167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035902261425874e-05,
+      "loss": 1.0335,
+      "step": 1168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033919478261886e-05,
+      "loss": 1.0284,
+      "step": 1169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031934761756763e-05,
+      "loss": 1.0446,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029948112335244e-05,
+      "loss": 0.3557,
+      "step": 1171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027959530422504e-05,
+      "loss": 1.0483,
+      "step": 1172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025969016444103e-05,
+      "loss": 0.9553,
+      "step": 1173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902397657082604e-05,
+      "loss": 1.0028,
+      "step": 1174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021982193994712e-05,
+      "loss": 1.1179,
+      "step": 1175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901998588637693e-05,
+      "loss": 1.0776,
+      "step": 1176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901798764839993e-05,
+      "loss": 1.031,
+      "step": 1177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901598748049135e-05,
+      "loss": 1.0446,
+      "step": 1178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013985383079236e-05,
+      "loss": 0.9951,
+      "step": 1179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901198135659207e-05,
+      "loss": 1.0031,
+      "step": 1180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900997540145872e-05,
+      "loss": 1.0022,
+      "step": 1181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007967518108486e-05,
+      "loss": 1.0399,
+      "step": 1182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900595770697107e-05,
+      "loss": 1.0329,
+      "step": 1183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900394596847659e-05,
+      "loss": 1.0275,
+      "step": 1184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001932303055574e-05,
+      "loss": 0.985,
+      "step": 1185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899991671113897e-05,
+      "loss": 1.0691,
+      "step": 1186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997899193158132e-05,
+      "loss": 0.9454,
+      "step": 1187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995879749544823e-05,
+      "loss": 0.9508,
+      "step": 1188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993858380731226e-05,
+      "loss": 0.9872,
+      "step": 1189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899183508714993e-05,
+      "loss": 1.0093,
+      "step": 1190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989809869233942e-05,
+      "loss": 1.0446,
+      "step": 1191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987782727416674e-05,
+      "loss": 1.0565,
+      "step": 1192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985753662131955e-05,
+      "loss": 0.9902,
+      "step": 1193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898372267381402e-05,
+      "loss": 1.0516,
+      "step": 1194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981689762897525e-05,
+      "loss": 0.9681,
+      "step": 1195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897965492981752e-05,
+      "loss": 1.0845,
+      "step": 1196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977618175009494e-05,
+      "loss": 1.0201,
+      "step": 1197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897557949890932e-05,
+      "loss": 1.0351,
+      "step": 1198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897353890195329e-05,
+      "loss": 1.0484,
+      "step": 1199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971496384578124e-05,
+      "loss": 1.0626,
+      "step": 1200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896945194722093e-05,
+      "loss": 1.0239,
+      "step": 1201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967405590319234e-05,
+      "loss": 1.048,
+      "step": 1202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896535731431098e-05,
+      "loss": 0.9354,
+      "step": 1203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963307119634518e-05,
+      "loss": 0.9964,
+      "step": 1204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961255006728607e-05,
+      "loss": 1.0544,
+      "step": 1205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959200976032416e-05,
+      "loss": 1.0371,
+      "step": 1206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895714502798553e-05,
+      "loss": 0.987,
+      "step": 1207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895508716302794e-05,
+      "loss": 0.9788,
+      "step": 1208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953027381600042e-05,
+      "loss": 1.0241,
+      "step": 1209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950965684142656e-05,
+      "loss": 1.0453,
+      "step": 1210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948902071097e-05,
+      "loss": 0.9966,
+      "step": 1211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946836542904705e-05,
+      "loss": 1.0307,
+      "step": 1212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944769100007815e-05,
+      "loss": 0.9696,
+      "step": 1213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942699742848784e-05,
+      "loss": 0.9454,
+      "step": 1214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940628471870467e-05,
+      "loss": 1.0069,
+      "step": 1215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893855528751614e-05,
+      "loss": 1.0238,
+      "step": 1216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893648019022948e-05,
+      "loss": 1.0586,
+      "step": 1217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893440318045458e-05,
+      "loss": 1.0348,
+      "step": 1218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932324258635938e-05,
+      "loss": 1.0451,
+      "step": 1219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930243425218462e-05,
+      "loss": 1.0062,
+      "step": 1220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928160680647467e-05,
+      "loss": 0.9926,
+      "step": 1221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892607602536868e-05,
+      "loss": 0.9945,
+      "step": 1222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892398945982824e-05,
+      "loss": 1.0348,
+      "step": 1223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892190098447269e-05,
+      "loss": 1.0384,
+      "step": 1224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891981059974898e-05,
+      "loss": 1.0076,
+      "step": 1225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8917718306104472e-05,
+      "loss": 0.9782,
+      "step": 1226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915624103986936e-05,
+      "loss": 1.0908,
+      "step": 1227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891352799384455e-05,
+      "loss": 1.0486,
+      "step": 1228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911429976125904e-05,
+      "loss": 0.9939,
+      "step": 1229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909330051279992e-05,
+      "loss": 1.0429,
+      "step": 1230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907228219756214e-05,
+      "loss": 1.0551,
+      "step": 1231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890512448200438e-05,
+      "loss": 0.9638,
+      "step": 1232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903018838474717e-05,
+      "loss": 1.0313,
+      "step": 1233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890091128961784e-05,
+      "loss": 1.0125,
+      "step": 1234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88988018358848e-05,
+      "loss": 1.0699,
+      "step": 1235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896690477727024e-05,
+      "loss": 0.3481,
+      "step": 1236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894577215596367e-05,
+      "loss": 0.9917,
+      "step": 1237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892462049945086e-05,
+      "loss": 1.0648,
+      "step": 1238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889034498122585e-05,
+      "loss": 1.0566,
+      "step": 1239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888226009891726e-05,
+      "loss": 1.0057,
+      "step": 1240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886105136396197e-05,
+      "loss": 1.0326,
+      "step": 1241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883982361193146e-05,
+      "loss": 0.9993,
+      "step": 1242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881857684736866e-05,
+      "loss": 1.0507,
+      "step": 1243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887973110748206e-05,
+      "loss": 1.0656,
+      "step": 1244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877602629883833e-05,
+      "loss": 1.0637,
+      "step": 1245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88754722523977e-05,
+      "loss": 1.0692,
+      "step": 1246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873339975479582e-05,
+      "loss": 0.9877,
+      "step": 1247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88712057995858e-05,
+      "loss": 1.0182,
+      "step": 1248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869069725173093e-05,
+      "loss": 1.0911,
+      "step": 1249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88669317526986e-05,
+      "loss": 1.0101,
+      "step": 1250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864791882619862e-05,
+      "loss": 1.0264,
+      "step": 1251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862650115394837e-05,
+      "loss": 1.0293,
+      "step": 1252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860506451481876e-05,
+      "loss": 1.0167,
+      "step": 1253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885836089133975e-05,
+      "loss": 1.0216,
+      "step": 1254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856213435427623e-05,
+      "loss": 1.0682,
+      "step": 1255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885406408420507e-05,
+      "loss": 1.0958,
+      "step": 1256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851912838132075e-05,
+      "loss": 0.9866,
+      "step": 1257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849759697669022e-05,
+      "loss": 1.003,
+      "step": 1258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847604663276708e-05,
+      "loss": 1.0465,
+      "step": 1259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845447735416322e-05,
+      "loss": 1.1026,
+      "step": 1260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884328891454947e-05,
+      "loss": 1.032,
+      "step": 1261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884112820113816e-05,
+      "loss": 1.0782,
+      "step": 1262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88389655956448e-05,
+      "loss": 0.3768,
+      "step": 1263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836801098532214e-05,
+      "loss": 1.0191,
+      "step": 1264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883463471026362e-05,
+      "loss": 1.0257,
+      "step": 1265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832466431302644e-05,
+      "loss": 1.0279,
+      "step": 1266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830296262113323e-05,
+      "loss": 0.9804,
+      "step": 1267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828124203160086e-05,
+      "loss": 1.0606,
+      "step": 1268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825950254907778e-05,
+      "loss": 0.9912,
+      "step": 1269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823774417821644e-05,
+      "loss": 1.0232,
+      "step": 1270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882159669236733e-05,
+      "loss": 1.0376,
+      "step": 1271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881941707901089e-05,
+      "loss": 1.0065,
+      "step": 1272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817235578218777e-05,
+      "loss": 1.0569,
+      "step": 1273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815052190457863e-05,
+      "loss": 0.3714,
+      "step": 1274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812866916195404e-05,
+      "loss": 1.05,
+      "step": 1275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810679755899072e-05,
+      "loss": 0.3671,
+      "step": 1276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880849071003694e-05,
+      "loss": 1.0137,
+      "step": 1277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806299779077476e-05,
+      "loss": 1.0234,
+      "step": 1278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880410696348957e-05,
+      "loss": 1.0027,
+      "step": 1279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801912263742502e-05,
+      "loss": 1.0786,
+      "step": 1280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879971568030595e-05,
+      "loss": 1.0021,
+      "step": 1281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797517213650014e-05,
+      "loss": 1.0251,
+      "step": 1282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795316864245175e-05,
+      "loss": 1.0175,
+      "step": 1283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793114632562338e-05,
+      "loss": 1.0017,
+      "step": 1284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790910519072792e-05,
+      "loss": 1.0189,
+      "step": 1285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788704524248243e-05,
+      "loss": 1.0615,
+      "step": 1286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786496648560794e-05,
+      "loss": 0.9753,
+      "step": 1287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784286892482947e-05,
+      "loss": 1.0807,
+      "step": 1288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782075256487613e-05,
+      "loss": 0.9914,
+      "step": 1289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87798617410481e-05,
+      "loss": 1.0156,
+      "step": 1290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877764634663812e-05,
+      "loss": 1.0236,
+      "step": 1291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877542907373179e-05,
+      "loss": 1.075,
+      "step": 1292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877320992280363e-05,
+      "loss": 1.0007,
+      "step": 1293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770988894328547e-05,
+      "loss": 1.0189,
+      "step": 1294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768765988781875e-05,
+      "loss": 0.9916,
+      "step": 1295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766541206639325e-05,
+      "loss": 1.0194,
+      "step": 1296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876431454837703e-05,
+      "loss": 0.371,
+      "step": 1297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762086014471507e-05,
+      "loss": 1.0292,
+      "step": 1298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875985560539969e-05,
+      "loss": 1.0133,
+      "step": 1299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87576233216389e-05,
+      "loss": 1.0301,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755389163666867e-05,
+      "loss": 1.0165,
+      "step": 1301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753153131961723e-05,
+      "loss": 1.0293,
+      "step": 1302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750915227002006e-05,
+      "loss": 0.9921,
+      "step": 1303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748675449266635e-05,
+      "loss": 1.0139,
+      "step": 1304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874643379923495e-05,
+      "loss": 1.0469,
+      "step": 1305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744190277386687e-05,
+      "loss": 1.0608,
+      "step": 1306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741944884201974e-05,
+      "loss": 0.9896,
+      "step": 1307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873969762016135e-05,
+      "loss": 1.0591,
+      "step": 1308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873744848574575e-05,
+      "loss": 1.0682,
+      "step": 1309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735197481436507e-05,
+      "loss": 1.0207,
+      "step": 1310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732944607715357e-05,
+      "loss": 0.9791,
+      "step": 1311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730689865064432e-05,
+      "loss": 1.0545,
+      "step": 1312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728433253966277e-05,
+      "loss": 1.0693,
+      "step": 1313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726174774903822e-05,
+      "loss": 1.0733,
+      "step": 1314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723914428360398e-05,
+      "loss": 1.0358,
+      "step": 1315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721652214819745e-05,
+      "loss": 1.0242,
+      "step": 1316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719388134765994e-05,
+      "loss": 1.0292,
+      "step": 1317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717122188683685e-05,
+      "loss": 0.4027,
+      "step": 1318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714854377057743e-05,
+      "loss": 1.0701,
+      "step": 1319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712584700373506e-05,
+      "loss": 0.3899,
+      "step": 1320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710313159116703e-05,
+      "loss": 1.1005,
+      "step": 1321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708039753773466e-05,
+      "loss": 1.0707,
+      "step": 1322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705764484830324e-05,
+      "loss": 1.0102,
+      "step": 1323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703487352774202e-05,
+      "loss": 1.0392,
+      "step": 1324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701208358092433e-05,
+      "loss": 1.0501,
+      "step": 1325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698927501272738e-05,
+      "loss": 0.9518,
+      "step": 1326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696644782803242e-05,
+      "loss": 1.0342,
+      "step": 1327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869436020317247e-05,
+      "loss": 1.019,
+      "step": 1328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869207376286934e-05,
+      "loss": 0.9993,
+      "step": 1329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689785462383174e-05,
+      "loss": 1.0344,
+      "step": 1330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687495302203686e-05,
+      "loss": 1.081,
+      "step": 1331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685203282820994e-05,
+      "loss": 1.0285,
+      "step": 1332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868290940472561e-05,
+      "loss": 1.0475,
+      "step": 1333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868061366840844e-05,
+      "loss": 0.9883,
+      "step": 1334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678316074360803e-05,
+      "loss": 0.9972,
+      "step": 1335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86760166230744e-05,
+      "loss": 1.0265,
+      "step": 1336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673715315041326e-05,
+      "loss": 1.0638,
+      "step": 1337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671412150754092e-05,
+      "loss": 1.0017,
+      "step": 1338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669107130705596e-05,
+      "loss": 1.0416,
+      "step": 1339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866680025538913e-05,
+      "loss": 0.939,
+      "step": 1340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664491525298388e-05,
+      "loss": 0.9635,
+      "step": 1341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662180940927462e-05,
+      "loss": 1.0193,
+      "step": 1342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865986850277083e-05,
+      "loss": 0.9742,
+      "step": 1343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865755421132338e-05,
+      "loss": 1.0407,
+      "step": 1344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655238067080393e-05,
+      "loss": 0.9807,
+      "step": 1345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865292007053754e-05,
+      "loss": 0.9633,
+      "step": 1346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650600222190895e-05,
+      "loss": 1.027,
+      "step": 1347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8648278522536935e-05,
+      "loss": 1.0419,
+      "step": 1348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864595497207251e-05,
+      "loss": 1.048,
+      "step": 1349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643629571294895e-05,
+      "loss": 1.0424,
+      "step": 1350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864130232070174e-05,
+      "loss": 1.0457,
+      "step": 1351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86389732207911e-05,
+      "loss": 1.1029,
+      "step": 1352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863664227206142e-05,
+      "loss": 1.0231,
+      "step": 1353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634309475011547e-05,
+      "loss": 1.0793,
+      "step": 1354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863197483014072e-05,
+      "loss": 1.0013,
+      "step": 1355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629638337948578e-05,
+      "loss": 0.9838,
+      "step": 1356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627299998935144e-05,
+      "loss": 1.0065,
+      "step": 1357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862495981360085e-05,
+      "loss": 1.0765,
+      "step": 1358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622617782446516e-05,
+      "loss": 1.0205,
+      "step": 1359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620273905973357e-05,
+      "loss": 1.0555,
+      "step": 1360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617928184682985e-05,
+      "loss": 1.0268,
+      "step": 1361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615580619077405e-05,
+      "loss": 1.0526,
+      "step": 1362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861323120965902e-05,
+      "loss": 0.4008,
+      "step": 1363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610879956930618e-05,
+      "loss": 1.0625,
+      "step": 1364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608526861395395e-05,
+      "loss": 1.036,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860617192355693e-05,
+      "loss": 1.0033,
+      "step": 1366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860381514391921e-05,
+      "loss": 1.0388,
+      "step": 1367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.86014565229866e-05,
+      "loss": 1.0641,
+      "step": 1368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599096061263866e-05,
+      "loss": 1.0036,
+      "step": 1369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596733759256168e-05,
+      "loss": 1.012,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594369617469066e-05,
+      "loss": 1.0252,
+      "step": 1371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592003636408504e-05,
+      "loss": 1.052,
+      "step": 1372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589635816580823e-05,
+      "loss": 0.3895,
+      "step": 1373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587266158492757e-05,
+      "loss": 1.0186,
+      "step": 1374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858489466265144e-05,
+      "loss": 1.0036,
+      "step": 1375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582521329564387e-05,
+      "loss": 1.0129,
+      "step": 1376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580146159739515e-05,
+      "loss": 0.9883,
+      "step": 1377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577769153685135e-05,
+      "loss": 0.9966,
+      "step": 1378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575390311909946e-05,
+      "loss": 1.0601,
+      "step": 1379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573009634923044e-05,
+      "loss": 1.0521,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570627123233907e-05,
+      "loss": 0.3848,
+      "step": 1381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568242777352427e-05,
+      "loss": 1.0288,
+      "step": 1382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565856597788865e-05,
+      "loss": 1.0609,
+      "step": 1383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856346858505389e-05,
+      "loss": 0.3207,
+      "step": 1384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856107873965856e-05,
+      "loss": 1.0205,
+      "step": 1385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855868706211432e-05,
+      "loss": 1.0473,
+      "step": 1386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556293552933013e-05,
+      "loss": 0.9872,
+      "step": 1387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553898212626872e-05,
+      "loss": 0.9894,
+      "step": 1388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855150104170852e-05,
+      "loss": 1.068,
+      "step": 1389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549102040690972e-05,
+      "loss": 0.9923,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854670121008764e-05,
+      "loss": 0.9767,
+      "step": 1391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544298550412322e-05,
+      "loss": 1.0287,
+      "step": 1392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854189406217921e-05,
+      "loss": 1.0453,
+      "step": 1393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539487745902887e-05,
+      "loss": 0.985,
+      "step": 1394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537079602098322e-05,
+      "loss": 0.9727,
+      "step": 1395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534669631280885e-05,
+      "loss": 1.0466,
+      "step": 1396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853225783396633e-05,
+      "loss": 1.0323,
+      "step": 1397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529844210670803e-05,
+      "loss": 1.0314,
+      "step": 1398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852742876191084e-05,
+      "loss": 1.0455,
+      "step": 1399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525011488203373e-05,
+      "loss": 1.0394,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522592390065715e-05,
+      "loss": 1.0136,
+      "step": 1401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520171468015583e-05,
+      "loss": 1.0269,
+      "step": 1402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517748722571066e-05,
+      "loss": 1.0022,
+      "step": 1403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515324154250664e-05,
+      "loss": 0.3491,
+      "step": 1404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851289776357325e-05,
+      "loss": 0.9933,
+      "step": 1405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510469551058097e-05,
+      "loss": 1.0634,
+      "step": 1406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508039517224863e-05,
+      "loss": 0.9913,
+      "step": 1407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505607662593597e-05,
+      "loss": 1.0098,
+      "step": 1408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850317398768474e-05,
+      "loss": 1.0104,
+      "step": 1409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850073849301912e-05,
+      "loss": 0.9307,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498301179117952e-05,
+      "loss": 1.0206,
+      "step": 1411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495862046502847e-05,
+      "loss": 1.022,
+      "step": 1412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493421095695803e-05,
+      "loss": 1.0266,
+      "step": 1413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84909783272192e-05,
+      "loss": 1.0233,
+      "step": 1414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848853374159582e-05,
+      "loss": 1.0879,
+      "step": 1415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848608733934882e-05,
+      "loss": 1.0479,
+      "step": 1416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483639121001754e-05,
+      "loss": 1.0402,
+      "step": 1417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481189087078567e-05,
+      "loss": 1.0503,
+      "step": 1418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478737238103585e-05,
+      "loss": 1.045,
+      "step": 1419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476283574601525e-05,
+      "loss": 1.0275,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473828097097493e-05,
+      "loss": 1.0154,
+      "step": 1421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847137080611699e-05,
+      "loss": 1.0519,
+      "step": 1422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468911702185894e-05,
+      "loss": 0.9924,
+      "step": 1423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466450785830473e-05,
+      "loss": 0.9724,
+      "step": 1424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463988057577392e-05,
+      "loss": 0.9688,
+      "step": 1425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461523517953694e-05,
+      "loss": 0.982,
+      "step": 1426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459057167486813e-05,
+      "loss": 1.0343,
+      "step": 1427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845658900670457e-05,
+      "loss": 1.0546,
+      "step": 1428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845411903613517e-05,
+      "loss": 1.0501,
+      "step": 1429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8451647256307216e-05,
+      "loss": 0.9837,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844917366774969e-05,
+      "loss": 1.0023,
+      "step": 1431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844669827099196e-05,
+      "loss": 1.017,
+      "step": 1432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444221066563785e-05,
+      "loss": 0.3834,
+      "step": 1433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844174205499531e-05,
+      "loss": 1.0722,
+      "step": 1434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8439261236817064e-05,
+      "loss": 1.0496,
+      "step": 1435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436778612559966e-05,
+      "loss": 1.0065,
+      "step": 1436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843429418275532e-05,
+      "loss": 0.4001,
+      "step": 1437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843180794793482e-05,
+      "loss": 1.0042,
+      "step": 1438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429319908630537e-05,
+      "loss": 1.0228,
+      "step": 1439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426830065374938e-05,
+      "loss": 1.0355,
+      "step": 1440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842433841870087e-05,
+      "loss": 1.0455,
+      "step": 1441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842184496914157e-05,
+      "loss": 0.9872,
+      "step": 1442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419349717230656e-05,
+      "loss": 0.919,
+      "step": 1443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841685266350214e-05,
+      "loss": 0.9859,
+      "step": 1444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841435380849041e-05,
+      "loss": 1.0725,
+      "step": 1445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411853152730243e-05,
+      "loss": 0.9848,
+      "step": 1446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409350696756804e-05,
+      "loss": 1.0355,
+      "step": 1447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406846441105642e-05,
+      "loss": 1.0111,
+      "step": 1448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404340386312688e-05,
+      "loss": 0.9601,
+      "step": 1449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401832532914267e-05,
+      "loss": 0.9924,
+      "step": 1450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399322881447075e-05,
+      "loss": 0.9622,
+      "step": 1451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396811432448202e-05,
+      "loss": 1.0209,
+      "step": 1452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394298186455125e-05,
+      "loss": 1.022,
+      "step": 1453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391783144005693e-05,
+      "loss": 0.9833,
+      "step": 1454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838926630563816e-05,
+      "loss": 1.0408,
+      "step": 1455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838674767189114e-05,
+      "loss": 1.0263,
+      "step": 1456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384227243303658e-05,
+      "loss": 1.0383,
+      "step": 1457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381705020415094e-05,
+      "loss": 1.0137,
+      "step": 1458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379181003765234e-05,
+      "loss": 1.0428,
+      "step": 1459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837665519389424e-05,
+      "loss": 0.9167,
+      "step": 1460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837412759134266e-05,
+      "loss": 1.0344,
+      "step": 1461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837159819665142e-05,
+      "loss": 1.0212,
+      "step": 1462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369067010361835e-05,
+      "loss": 0.997,
+      "step": 1463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366534033015604e-05,
+      "loss": 1.0647,
+      "step": 1464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363999265154807e-05,
+      "loss": 0.9349,
+      "step": 1465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361462707321903e-05,
+      "loss": 1.0066,
+      "step": 1466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358924360059744e-05,
+      "loss": 1.0081,
+      "step": 1467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356384223911556e-05,
+      "loss": 1.0321,
+      "step": 1468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835384229942095e-05,
+      "loss": 1.0605,
+      "step": 1469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351298587131925e-05,
+      "loss": 0.9925,
+      "step": 1470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834875308758886e-05,
+      "loss": 1.0003,
+      "step": 1471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346205801336507e-05,
+      "loss": 0.9931,
+      "step": 1472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834365672892002e-05,
+      "loss": 0.9605,
+      "step": 1473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834110587088491e-05,
+      "loss": 1.0205,
+      "step": 1474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338553227777096e-05,
+      "loss": 0.968,
+      "step": 1475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833599880014286e-05,
+      "loss": 1.0294,
+      "step": 1476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333442588528874e-05,
+      "loss": 0.9758,
+      "step": 1477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330884593482197e-05,
+      "loss": 1.0456,
+      "step": 1478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832832481555025e-05,
+      "loss": 0.9808,
+      "step": 1479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325763255280865e-05,
+      "loss": 1.0275,
+      "step": 1480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323199913222228e-05,
+      "loss": 0.9849,
+      "step": 1481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320634789922915e-05,
+      "loss": 0.9998,
+      "step": 1482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83180678859319e-05,
+      "loss": 0.9631,
+      "step": 1483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315499201798507e-05,
+      "loss": 1.0362,
+      "step": 1484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831292873807247e-05,
+      "loss": 1.0665,
+      "step": 1485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310356495303888e-05,
+      "loss": 0.9327,
+      "step": 1486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830778247404324e-05,
+      "loss": 1.0241,
+      "step": 1487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305206674841397e-05,
+      "loss": 1.0189,
+      "step": 1488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302629098249604e-05,
+      "loss": 0.9321,
+      "step": 1489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300049744819474e-05,
+      "loss": 1.0325,
+      "step": 1490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297468615103025e-05,
+      "loss": 0.9631,
+      "step": 1491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294885709652642e-05,
+      "loss": 0.9796,
+      "step": 1492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829230102902108e-05,
+      "loss": 1.0539,
+      "step": 1493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828971457376149e-05,
+      "loss": 1.0409,
+      "step": 1494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287126344427402e-05,
+      "loss": 1.0292,
+      "step": 1495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284536341572713e-05,
+      "loss": 0.3655,
+      "step": 1496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281944565751712e-05,
+      "loss": 0.9561,
+      "step": 1497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279351017519063e-05,
+      "loss": 0.9865,
+      "step": 1498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276755697429803e-05,
+      "loss": 1.0119,
+      "step": 1499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827415860603936e-05,
+      "loss": 1.0047,
+      "step": 1500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271559743903532e-05,
+      "loss": 1.0312,
+      "step": 1501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82689591115785e-05,
+      "loss": 0.3642,
+      "step": 1502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266356709620827e-05,
+      "loss": 1.0254,
+      "step": 1503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263752538587443e-05,
+      "loss": 0.9336,
+      "step": 1504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826114659903567e-05,
+      "loss": 1.0872,
+      "step": 1505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258538891523202e-05,
+      "loss": 1.0298,
+      "step": 1506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8255929416608112e-05,
+      "loss": 1.0075,
+      "step": 1507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825331817484885e-05,
+      "loss": 1.0949,
+      "step": 1508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825070516680425e-05,
+      "loss": 1.0219,
+      "step": 1509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824809039303351e-05,
+      "loss": 0.997,
+      "step": 1510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245473854096226e-05,
+      "loss": 1.0451,
+      "step": 1511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242855550552355e-05,
+      "loss": 0.9915,
+      "step": 1512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824023548296224e-05,
+      "loss": 1.1133,
+      "step": 1513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82376136518866e-05,
+      "loss": 1.0459,
+      "step": 1514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234990057886526e-05,
+      "loss": 1.0083,
+      "step": 1515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232364701523502e-05,
+      "loss": 0.3401,
+      "step": 1516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229737583359367e-05,
+      "loss": 1.0421,
+      "step": 1517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227108703956352e-05,
+      "loss": 0.98,
+      "step": 1518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822447806387706e-05,
+      "loss": 0.9869,
+      "step": 1519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221845663684475e-05,
+      "loss": 1.0299,
+      "step": 1520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821921150394195e-05,
+      "loss": 1.0097,
+      "step": 1521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8216575585213226e-05,
+      "loss": 0.9764,
+      "step": 1522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821393790806241e-05,
+      "loss": 1.011,
+      "step": 1523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211298473053987e-05,
+      "loss": 0.9827,
+      "step": 1524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820865728075282e-05,
+      "loss": 1.0118,
+      "step": 1525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206014331724153e-05,
+      "loss": 0.9862,
+      "step": 1526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82033696265336e-05,
+      "loss": 1.0202,
+      "step": 1527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200723165747146e-05,
+      "loss": 1.0546,
+      "step": 1528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198074949931164e-05,
+      "loss": 1.0531,
+      "step": 1529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195424979652395e-05,
+      "loss": 1.0111,
+      "step": 1530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192773255477957e-05,
+      "loss": 1.0256,
+      "step": 1531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819011977797534e-05,
+      "loss": 0.9626,
+      "step": 1532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818746454771242e-05,
+      "loss": 0.9709,
+      "step": 1533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184807565257427e-05,
+      "loss": 1.0372,
+      "step": 1534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182148831178992e-05,
+      "loss": 0.3707,
+      "step": 1535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179488346046103e-05,
+      "loss": 0.9643,
+      "step": 1536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817682611042813e-05,
+      "loss": 0.9957,
+      "step": 1537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174162124894814e-05,
+      "loss": 1.0314,
+      "step": 1538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171496390016276e-05,
+      "loss": 1.004,
+      "step": 1539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168828906363002e-05,
+      "loss": 1.0003,
+      "step": 1540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816615967450586e-05,
+      "loss": 0.9344,
+      "step": 1541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816348869501609e-05,
+      "loss": 1.0113,
+      "step": 1542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816081596846531e-05,
+      "loss": 1.025,
+      "step": 1543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158141495425502e-05,
+      "loss": 1.0407,
+      "step": 1544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815546527646903e-05,
+      "loss": 1.0116,
+      "step": 1545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815278731216863e-05,
+      "loss": 1.0124,
+      "step": 1546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150107603097406e-05,
+      "loss": 1.0245,
+      "step": 1547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814742614982885e-05,
+      "loss": 1.0032,
+      "step": 1548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144742952936807e-05,
+      "loss": 1.1273,
+      "step": 1549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814205801299551e-05,
+      "loss": 1.0329,
+      "step": 1550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813937133057956e-05,
+      "loss": 0.345,
+      "step": 1551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136682906263935e-05,
+      "loss": 1.0372,
+      "step": 1552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133992740623977e-05,
+      "loss": 1.0072,
+      "step": 1553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813130083423541e-05,
+      "loss": 0.9843,
+      "step": 1554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812860718767433e-05,
+      "loss": 0.3831,
+      "step": 1555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812591180151719e-05,
+      "loss": 0.3963,
+      "step": 1556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123214676340834e-05,
+      "loss": 0.3504,
+      "step": 1557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120515812722476e-05,
+      "loss": 1.0074,
+      "step": 1558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117815211239692e-05,
+      "loss": 1.0294,
+      "step": 1559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115112872470436e-05,
+      "loss": 0.9946,
+      "step": 1560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811240879699303e-05,
+      "loss": 1.027,
+      "step": 1561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109702985386176e-05,
+      "loss": 1.0037,
+      "step": 1562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106995438228942e-05,
+      "loss": 1.023,
+      "step": 1563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104286156100766e-05,
+      "loss": 0.9871,
+      "step": 1564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101575139581456e-05,
+      "loss": 1.0403,
+      "step": 1565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098862389251202e-05,
+      "loss": 0.9845,
+      "step": 1566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096147905690554e-05,
+      "loss": 0.9637,
+      "step": 1567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093431689480435e-05,
+      "loss": 0.9986,
+      "step": 1568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090713741202135e-05,
+      "loss": 1.0108,
+      "step": 1569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808799406143733e-05,
+      "loss": 0.9841,
+      "step": 1570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808527265076805e-05,
+      "loss": 1.0197,
+      "step": 1571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082549509776702e-05,
+      "loss": 0.9428,
+      "step": 1572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079824639046065e-05,
+      "loss": 1.0158,
+      "step": 1573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077098039159283e-05,
+      "loss": 0.9122,
+      "step": 1574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074369710699877e-05,
+      "loss": 1.0528,
+      "step": 1575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071639654251733e-05,
+      "loss": 1.0264,
+      "step": 1576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068907870399108e-05,
+      "loss": 1.0288,
+      "step": 1577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066174359726628e-05,
+      "loss": 1.067,
+      "step": 1578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806343912281929e-05,
+      "loss": 1.0544,
+      "step": 1579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060702160262456e-05,
+      "loss": 0.9961,
+      "step": 1580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057963472641865e-05,
+      "loss": 1.0093,
+      "step": 1581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055223060543624e-05,
+      "loss": 1.007,
+      "step": 1582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052480924554198e-05,
+      "loss": 0.9388,
+      "step": 1583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804973706526043e-05,
+      "loss": 0.9454,
+      "step": 1584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804699148324954e-05,
+      "loss": 0.9887,
+      "step": 1585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044244179109104e-05,
+      "loss": 1.0315,
+      "step": 1586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041495153427062e-05,
+      "loss": 0.9801,
+      "step": 1587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803874440679174e-05,
+      "loss": 1.0206,
+      "step": 1588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803599193979182e-05,
+      "loss": 0.3504,
+      "step": 1589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033237753016352e-05,
+      "loss": 1.0182,
+      "step": 1590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803048184705476e-05,
+      "loss": 0.9846,
+      "step": 1591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027724222496835e-05,
+      "loss": 1.048,
+      "step": 1592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024964879932726e-05,
+      "loss": 0.9735,
+      "step": 1593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802220381995297e-05,
+      "loss": 0.9442,
+      "step": 1594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019441043148446e-05,
+      "loss": 0.9855,
+      "step": 1595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801667655011042e-05,
+      "loss": 0.9999,
+      "step": 1596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801391034143052e-05,
+      "loss": 1.0545,
+      "step": 1597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011142417700735e-05,
+      "loss": 0.9767,
+      "step": 1598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008372779513432e-05,
+      "loss": 1.007,
+      "step": 1599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005601427461335e-05,
+      "loss": 1.034,
+      "step": 1600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002828362137535e-05,
+      "loss": 0.9979,
+      "step": 1601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000053584135505e-05,
+      "loss": 1.0564,
+      "step": 1602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997277094049058e-05,
+      "loss": 0.9735,
+      "step": 1603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994498892472402e-05,
+      "loss": 0.9994,
+      "step": 1604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991718980000086e-05,
+      "loss": 0.9952,
+      "step": 1605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988937357227043e-05,
+      "loss": 1.0018,
+      "step": 1606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986154024748566e-05,
+      "loss": 1.0272,
+      "step": 1607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983368983160307e-05,
+      "loss": 1.1105,
+      "step": 1608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980582233058296e-05,
+      "loss": 0.9661,
+      "step": 1609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977793775038925e-05,
+      "loss": 1.0656,
+      "step": 1610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975003609698945e-05,
+      "loss": 0.9594,
+      "step": 1611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972211737635478e-05,
+      "loss": 0.9953,
+      "step": 1612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969418159446014e-05,
+      "loss": 1.0807,
+      "step": 1613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966622875728396e-05,
+      "loss": 1.0062,
+      "step": 1614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963825887080842e-05,
+      "loss": 1.0566,
+      "step": 1615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796102719410194e-05,
+      "loss": 1.0283,
+      "step": 1616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795822679739063e-05,
+      "loss": 1.0114,
+      "step": 1617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955424697546228e-05,
+      "loss": 0.9793,
+      "step": 1618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79526208951684e-05,
+      "loss": 1.0792,
+      "step": 1619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794981539085719e-05,
+      "loss": 0.3493,
+      "step": 1620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947008185213005e-05,
+      "loss": 1.0228,
+      "step": 1621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944199278836606e-05,
+      "loss": 0.9954,
+      "step": 1622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794138867232913e-05,
+      "loss": 1.0813,
+      "step": 1623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793857636629207e-05,
+      "loss": 1.0189,
+      "step": 1624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935762361327286e-05,
+      "loss": 1.0075,
+      "step": 1625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932946658036998e-05,
+      "loss": 1.062,
+      "step": 1626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930129257023795e-05,
+      "loss": 1.0525,
+      "step": 1627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7927310158890626e-05,
+      "loss": 0.3502,
+      "step": 1628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924489364240804e-05,
+      "loss": 1.0217,
+      "step": 1629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921666873678007e-05,
+      "loss": 1.0565,
+      "step": 1630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918842687806264e-05,
+      "loss": 1.0043,
+      "step": 1631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791601680722999e-05,
+      "loss": 1.0168,
+      "step": 1632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791318923255394e-05,
+      "loss": 0.3679,
+      "step": 1633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910359964383245e-05,
+      "loss": 0.9938,
+      "step": 1634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907529003323388e-05,
+      "loss": 1.0299,
+      "step": 1635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790469634998023e-05,
+      "loss": 1.0879,
+      "step": 1636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790186200495998e-05,
+      "loss": 0.9393,
+      "step": 1637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899025968869212e-05,
+      "loss": 1.0367,
+      "step": 1638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896188242314863e-05,
+      "loss": 1.0467,
+      "step": 1639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893348825904234e-05,
+      "loss": 0.9787,
+      "step": 1640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789050772024499e-05,
+      "loss": 1.0149,
+      "step": 1641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788766492594515e-05,
+      "loss": 1.0642,
+      "step": 1642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884820443613096e-05,
+      "loss": 1.0217,
+      "step": 1643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881974273857575e-05,
+      "loss": 0.3914,
+      "step": 1644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879126417287692e-05,
+      "loss": 1.0089,
+      "step": 1645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876276874512917e-05,
+      "loss": 1.0178,
+      "step": 1646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873425646143077e-05,
+      "loss": 1.0408,
+      "step": 1647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787057273278836e-05,
+      "loss": 1.0507,
+      "step": 1648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867718135059316e-05,
+      "loss": 0.9361,
+      "step": 1649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7864861853566857e-05,
+      "loss": 0.9422,
+      "step": 1650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862003888922248e-05,
+      "loss": 1.0257,
+      "step": 1651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785914424173712e-05,
+      "loss": 1.0426,
+      "step": 1652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856282912623473e-05,
+      "loss": 1.0599,
+      "step": 1653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785341990219365e-05,
+      "loss": 1.0386,
+      "step": 1654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785055521106036e-05,
+      "loss": 1.0065,
+      "step": 1655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847688839836673e-05,
+      "loss": 0.9067,
+      "step": 1656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844820789136026e-05,
+      "loss": 1.0262,
+      "step": 1657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78419510595722e-05,
+      "loss": 0.9196,
+      "step": 1658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839079651759346e-05,
+      "loss": 1.0627,
+      "step": 1659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783620656631197e-05,
+      "loss": 0.9804,
+      "step": 1660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833331803844945e-05,
+      "loss": 0.9572,
+      "step": 1661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830455364973485e-05,
+      "loss": 1.0159,
+      "step": 1662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827577250313183e-05,
+      "loss": 1.0195,
+      "step": 1663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782469746047998e-05,
+      "loss": 0.9744,
+      "step": 1664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821815996090177e-05,
+      "loss": 1.0529,
+      "step": 1665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781893285776043e-05,
+      "loss": 0.9297,
+      "step": 1666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816048046107766e-05,
+      "loss": 0.9619,
+      "step": 1667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781316156174955e-05,
+      "loss": 0.9535,
+      "step": 1668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781027340530352e-05,
+      "loss": 1.0271,
+      "step": 1669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807383577387773e-05,
+      "loss": 0.9542,
+      "step": 1670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804492078620755e-05,
+      "loss": 1.0166,
+      "step": 1671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780159890962127e-05,
+      "loss": 1.0539,
+      "step": 1672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798704071008487e-05,
+      "loss": 0.9368,
+      "step": 1673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779580756340192e-05,
+      "loss": 0.9675,
+      "step": 1674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779290938742146e-05,
+      "loss": 1.0525,
+      "step": 1675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790009543687334e-05,
+      "loss": 1.0089,
+      "step": 1676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778710803282014e-05,
+      "loss": 0.9952,
+      "step": 1677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784204855440826e-05,
+      "loss": 0.3953,
+      "step": 1678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781300012170697e-05,
+      "loss": 1.0133,
+      "step": 1679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778393503631418e-05,
+      "loss": 1.043,
+      "step": 1680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775485330445007e-05,
+      "loss": 0.9949,
+      "step": 1681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777257549323384e-05,
+      "loss": 1.015,
+      "step": 1682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769663992620647e-05,
+      "loss": 0.9698,
+      "step": 1683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766750829228522e-05,
+      "loss": 0.9965,
+      "step": 1684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77638360036809e-05,
+      "loss": 0.3567,
+      "step": 1685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760919516601586e-05,
+      "loss": 1.0267,
+      "step": 1686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758001368614733e-05,
+      "loss": 0.9438,
+      "step": 1687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775508156034485e-05,
+      "loss": 1.0401,
+      "step": 1688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752160092416807e-05,
+      "loss": 0.9545,
+      "step": 1689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774923696545582e-05,
+      "loss": 1.0482,
+      "step": 1690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746312180087464e-05,
+      "loss": 0.9883,
+      "step": 1691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743385736937674e-05,
+      "loss": 0.9516,
+      "step": 1692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740457636632734e-05,
+      "loss": 0.9774,
+      "step": 1693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737527879799287e-05,
+      "loss": 1.0369,
+      "step": 1694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773459646706432e-05,
+      "loss": 1.0045,
+      "step": 1695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773166339905519e-05,
+      "loss": 1.0589,
+      "step": 1696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772872867639959e-05,
+      "loss": 0.9697,
+      "step": 1697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725792299725594e-05,
+      "loss": 1.0161,
+      "step": 1698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77228542696616e-05,
+      "loss": 1.0171,
+      "step": 1699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7719914586836374e-05,
+      "loss": 0.9886,
+      "step": 1700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771697325187904e-05,
+      "loss": 0.366,
+      "step": 1701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714030265419067e-05,
+      "loss": 1.0582,
+      "step": 1702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711085628086287e-05,
+      "loss": 0.9472,
+      "step": 1703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770813934051087e-05,
+      "loss": 1.0234,
+      "step": 1704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705191403323354e-05,
+      "loss": 0.9642,
+      "step": 1705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702241817154626e-05,
+      "loss": 1.0032,
+      "step": 1706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769929058263592e-05,
+      "loss": 1.0106,
+      "step": 1707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696337700398828e-05,
+      "loss": 0.9735,
+      "step": 1708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693383171075294e-05,
+      "loss": 0.9695,
+      "step": 1709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690426995297615e-05,
+      "loss": 1.0403,
+      "step": 1710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687469173698443e-05,
+      "loss": 1.0397,
+      "step": 1711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768450970691077e-05,
+      "loss": 1.0099,
+      "step": 1712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681548595567955e-05,
+      "loss": 1.0366,
+      "step": 1713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678585840303707e-05,
+      "loss": 0.9189,
+      "step": 1714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675621441752075e-05,
+      "loss": 0.985,
+      "step": 1715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672655400547466e-05,
+      "loss": 1.094,
+      "step": 1716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669687717324646e-05,
+      "loss": 0.9771,
+      "step": 1717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666718392718726e-05,
+      "loss": 1.0612,
+      "step": 1718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663747427365165e-05,
+      "loss": 1.0313,
+      "step": 1719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660774821899776e-05,
+      "loss": 1.0248,
+      "step": 1720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657800576958727e-05,
+      "loss": 0.9886,
+      "step": 1721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7654824693178535e-05,
+      "loss": 1.0216,
+      "step": 1722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651847171196063e-05,
+      "loss": 0.9452,
+      "step": 1723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648868011648525e-05,
+      "loss": 0.3605,
+      "step": 1724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645887215173495e-05,
+      "loss": 0.9873,
+      "step": 1725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764290478240888e-05,
+      "loss": 0.965,
+      "step": 1726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639920713992964e-05,
+      "loss": 0.988,
+      "step": 1727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636935010564357e-05,
+      "loss": 1.0224,
+      "step": 1728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7633947672762023e-05,
+      "loss": 0.9895,
+      "step": 1729
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630958701225278e-05,
+      "loss": 1.0199,
+      "step": 1730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627968096593798e-05,
+      "loss": 1.0178,
+      "step": 1731
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624975859507594e-05,
+      "loss": 1.0067,
+      "step": 1732
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621981990607034e-05,
+      "loss": 1.0506,
+      "step": 1733
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618986490532833e-05,
+      "loss": 0.9995,
+      "step": 1734
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761598935992605e-05,
+      "loss": 1.029,
+      "step": 1735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761299059942811e-05,
+      "loss": 1.0291,
+      "step": 1736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760999020968076e-05,
+      "loss": 1.0885,
+      "step": 1737
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606988191326125e-05,
+      "loss": 1.0521,
+      "step": 1738
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760398454500666e-05,
+      "loss": 1.0296,
+      "step": 1739
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7600979271365167e-05,
+      "loss": 0.9929,
+      "step": 1740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759797237104481e-05,
+      "loss": 1.0674,
+      "step": 1741
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594963844689084e-05,
+      "loss": 0.3928,
+      "step": 1742
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591953692941847e-05,
+      "loss": 1.0008,
+      "step": 1743
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588941916447296e-05,
+      "loss": 1.0484,
+      "step": 1744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585928515849984e-05,
+      "loss": 1.02,
+      "step": 1745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582913491794803e-05,
+      "loss": 1.013,
+      "step": 1746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579896844926993e-05,
+      "loss": 0.3295,
+      "step": 1747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7576878575892145e-05,
+      "loss": 1.057,
+      "step": 1748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75738586853362e-05,
+      "loss": 1.0033,
+      "step": 1749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757083717390544e-05,
+      "loss": 1.0074,
+      "step": 1750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567814042246496e-05,
+      "loss": 1.0189,
+      "step": 1751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564789291006343e-05,
+      "loss": 0.9966,
+      "step": 1752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756176292083231e-05,
+      "loss": 1.0633,
+      "step": 1753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558734932372067e-05,
+      "loss": 0.9344,
+      "step": 1754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555705326273624e-05,
+      "loss": 1.0499,
+      "step": 1755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755267410318535e-05,
+      "loss": 1.0328,
+      "step": 1756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7549641263755957e-05,
+      "loss": 1.1131,
+      "step": 1757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75466068086345e-05,
+      "loss": 1.0039,
+      "step": 1758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543570738470375e-05,
+      "loss": 1.0273,
+      "step": 1759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754053305391333e-05,
+      "loss": 1.0349,
+      "step": 1760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753749375561346e-05,
+      "loss": 1.0701,
+      "step": 1761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534452844221203e-05,
+      "loss": 1.0086,
+      "step": 1762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531410320387336e-05,
+      "loss": 1.0305,
+      "step": 1763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528366184762992e-05,
+      "loss": 0.955,
+      "step": 1764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525320437999644e-05,
+      "loss": 1.0631,
+      "step": 1765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522273080749107e-05,
+      "loss": 1.0055,
+      "step": 1766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519224113663543e-05,
+      "loss": 1.0049,
+      "step": 1767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516173537395462e-05,
+      "loss": 0.9962,
+      "step": 1768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751312135259771e-05,
+      "loss": 0.9883,
+      "step": 1769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751006755992349e-05,
+      "loss": 0.9068,
+      "step": 1770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507012160026336e-05,
+      "loss": 0.9357,
+      "step": 1771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750395515356013e-05,
+      "loss": 0.9702,
+      "step": 1772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75008965411791e-05,
+      "loss": 1.0267,
+      "step": 1773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497836323537826e-05,
+      "loss": 1.0799,
+      "step": 1774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749477450129121e-05,
+      "loss": 1.025,
+      "step": 1775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491711075094513e-05,
+      "loss": 0.3821,
+      "step": 1776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488646045603342e-05,
+      "loss": 1.0162,
+      "step": 1777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485579413473638e-05,
+      "loss": 0.9906,
+      "step": 1778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482511179361684e-05,
+      "loss": 0.9713,
+      "step": 1779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479441343924114e-05,
+      "loss": 0.9919,
+      "step": 1780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476369907817905e-05,
+      "loss": 1.0252,
+      "step": 1781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473296871700363e-05,
+      "loss": 1.0339,
+      "step": 1782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470222236229152e-05,
+      "loss": 0.9945,
+      "step": 1783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746714600206227e-05,
+      "loss": 0.9887,
+      "step": 1784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746406816985806e-05,
+      "loss": 0.9822,
+      "step": 1785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7460988740275207e-05,
+      "loss": 1.0182,
+      "step": 1786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457907713972738e-05,
+      "loss": 0.9933,
+      "step": 1787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454825091610014e-05,
+      "loss": 0.9858,
+      "step": 1788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745174087384675e-05,
+      "loss": 1.009,
+      "step": 1789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448655061342998e-05,
+      "loss": 0.9984,
+      "step": 1790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445567654759146e-05,
+      "loss": 0.9465,
+      "step": 1791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442478654755927e-05,
+      "loss": 1.0031,
+      "step": 1792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743938806199442e-05,
+      "loss": 1.0292,
+      "step": 1793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436295877136035e-05,
+      "loss": 0.9922,
+      "step": 1794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743320210084253e-05,
+      "loss": 1.0064,
+      "step": 1795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430106733776003e-05,
+      "loss": 1.0391,
+      "step": 1796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742700977659889e-05,
+      "loss": 1.0608,
+      "step": 1797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423911229973963e-05,
+      "loss": 0.9888,
+      "step": 1798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420811094564345e-05,
+      "loss": 0.9828,
+      "step": 1799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741770937103349e-05,
+      "loss": 1.0365,
+      "step": 1800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414606060045197e-05,
+      "loss": 1.0563,
+      "step": 1801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411501162263605e-05,
+      "loss": 0.9065,
+      "step": 1802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408394678353184e-05,
+      "loss": 0.9223,
+      "step": 1803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405286608978755e-05,
+      "loss": 1.0001,
+      "step": 1804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402176954805474e-05,
+      "loss": 1.0576,
+      "step": 1805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739906571649883e-05,
+      "loss": 1.0264,
+      "step": 1806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395952894724663e-05,
+      "loss": 1.0103,
+      "step": 1807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392838490149138e-05,
+      "loss": 0.9806,
+      "step": 1808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389722503438773e-05,
+      "loss": 1.0315,
+      "step": 1809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386604935260412e-05,
+      "loss": 0.9666,
+      "step": 1810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738348578628125e-05,
+      "loss": 1.004,
+      "step": 1811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380365057168803e-05,
+      "loss": 1.0279,
+      "step": 1812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377242748590943e-05,
+      "loss": 1.0269,
+      "step": 1813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374118861215873e-05,
+      "loss": 0.979,
+      "step": 1814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370993395712127e-05,
+      "loss": 1.021,
+      "step": 1815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736786635274859e-05,
+      "loss": 0.9577,
+      "step": 1816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364737732994473e-05,
+      "loss": 0.9622,
+      "step": 1817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361607537119332e-05,
+      "loss": 0.9573,
+      "step": 1818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735847576579306e-05,
+      "loss": 1.015,
+      "step": 1819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355342419685877e-05,
+      "loss": 1.1377,
+      "step": 1820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352207499468355e-05,
+      "loss": 1.0454,
+      "step": 1821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734907100581139e-05,
+      "loss": 1.0277,
+      "step": 1822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345932939386217e-05,
+      "loss": 0.9542,
+      "step": 1823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342793300864426e-05,
+      "loss": 1.0168,
+      "step": 1824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339652090917912e-05,
+      "loss": 0.9758,
+      "step": 1825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336509310218933e-05,
+      "loss": 0.9885,
+      "step": 1826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333364959440068e-05,
+      "loss": 0.954,
+      "step": 1827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733021903925424e-05,
+      "loss": 1.049,
+      "step": 1828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7327071550334697e-05,
+      "loss": 1.0433,
+      "step": 1829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323922493355037e-05,
+      "loss": 1.0077,
+      "step": 1830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320771868989186e-05,
+      "loss": 1.0253,
+      "step": 1831
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317619677911405e-05,
+      "loss": 1.0656,
+      "step": 1832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314465920796296e-05,
+      "loss": 1.0523,
+      "step": 1833
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311310598318787e-05,
+      "loss": 1.0189,
+      "step": 1834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308153711154144e-05,
+      "loss": 1.0084,
+      "step": 1835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304995259977973e-05,
+      "loss": 1.016,
+      "step": 1836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7301835245466214e-05,
+      "loss": 0.9859,
+      "step": 1837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298673668295135e-05,
+      "loss": 1.0411,
+      "step": 1838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7295510529141345e-05,
+      "loss": 1.0134,
+      "step": 1839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729234582868178e-05,
+      "loss": 1.0347,
+      "step": 1840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7289179567593715e-05,
+      "loss": 1.0309,
+      "step": 1841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728601174655477e-05,
+      "loss": 1.0544,
+      "step": 1842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728284236624287e-05,
+      "loss": 0.9925,
+      "step": 1843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279671427336305e-05,
+      "loss": 0.9873,
+      "step": 1844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276498930513676e-05,
+      "loss": 0.9346,
+      "step": 1845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273324876453934e-05,
+      "loss": 0.9885,
+      "step": 1846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727014926583635e-05,
+      "loss": 1.0008,
+      "step": 1847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266972099340538e-05,
+      "loss": 0.4138,
+      "step": 1848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263793377646438e-05,
+      "loss": 0.9662,
+      "step": 1849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260613101434324e-05,
+      "loss": 0.95,
+      "step": 1850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257431271384803e-05,
+      "loss": 0.9709,
+      "step": 1851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725424788817882e-05,
+      "loss": 1.0731,
+      "step": 1852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725106295249765e-05,
+      "loss": 1.0699,
+      "step": 1853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724787646502289e-05,
+      "loss": 0.9404,
+      "step": 1854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244688426436486e-05,
+      "loss": 1.065,
+      "step": 1855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72414988374207e-05,
+      "loss": 0.9919,
+      "step": 1856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238307698658135e-05,
+      "loss": 0.9961,
+      "step": 1857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235115010831727e-05,
+      "loss": 1.0434,
+      "step": 1858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723192077462474e-05,
+      "loss": 0.9813,
+      "step": 1859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228724990720766e-05,
+      "loss": 1.0143,
+      "step": 1860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225527659803734e-05,
+      "loss": 0.9541,
+      "step": 1861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222328782557903e-05,
+      "loss": 0.9969,
+      "step": 1862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721912835966786e-05,
+      "loss": 0.9505,
+      "step": 1863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215926391818526e-05,
+      "loss": 1.0462,
+      "step": 1864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212722879695153e-05,
+      "loss": 1.0444,
+      "step": 1865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720951782398332e-05,
+      "loss": 1.0131,
+      "step": 1866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206311225368933e-05,
+      "loss": 1.048,
+      "step": 1867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203103084538245e-05,
+      "loss": 0.9962,
+      "step": 1868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199893402177815e-05,
+      "loss": 0.9986,
+      "step": 1869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719668217897455e-05,
+      "loss": 1.0145,
+      "step": 1870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719346941561568e-05,
+      "loss": 0.9625,
+      "step": 1871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190255112788773e-05,
+      "loss": 1.0237,
+      "step": 1872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187039271181708e-05,
+      "loss": 1.0357,
+      "step": 1873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718382189148271e-05,
+      "loss": 1.018,
+      "step": 1874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180602974380324e-05,
+      "loss": 0.99,
+      "step": 1875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177382520563434e-05,
+      "loss": 0.9688,
+      "step": 1876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174160530721242e-05,
+      "loss": 0.9864,
+      "step": 1877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717093700554328e-05,
+      "loss": 0.9836,
+      "step": 1878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167711945719423e-05,
+      "loss": 1.0389,
+      "step": 1879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716448535193985e-05,
+      "loss": 1.0555,
+      "step": 1880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716125722489509e-05,
+      "loss": 0.9934,
+      "step": 1881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715802756527599e-05,
+      "loss": 1.0653,
+      "step": 1882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154796373773725e-05,
+      "loss": 0.9978,
+      "step": 1883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71515636510798e-05,
+      "loss": 0.9761,
+      "step": 1884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148329397886042e-05,
+      "loss": 0.9959,
+      "step": 1885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145093614884625e-05,
+      "loss": 1.0103,
+      "step": 1886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714185630276802e-05,
+      "loss": 1.065,
+      "step": 1887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713861746222905e-05,
+      "loss": 0.9998,
+      "step": 1888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135377093960856e-05,
+      "loss": 0.9623,
+      "step": 1889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132135198656904e-05,
+      "loss": 1.002,
+      "step": 1890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128891777010993e-05,
+      "loss": 0.8999,
+      "step": 1891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712564682971724e-05,
+      "loss": 0.9752,
+      "step": 1892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122400357470093e-05,
+      "loss": 0.9437,
+      "step": 1893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119152360964333e-05,
+      "loss": 1.0251,
+      "step": 1894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7115902840895057e-05,
+      "loss": 0.9366,
+      "step": 1895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112651797957687e-05,
+      "loss": 0.9937,
+      "step": 1896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710939923284798e-05,
+      "loss": 1.0116,
+      "step": 1897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106145146262023e-05,
+      "loss": 0.997,
+      "step": 1898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102889538896207e-05,
+      "loss": 1.0152,
+      "step": 1899
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709963241144727e-05,
+      "loss": 0.9922,
+      "step": 1900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709637376461226e-05,
+      "loss": 1.007,
+      "step": 1901
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7093113599088565e-05,
+      "loss": 0.97,
+      "step": 1902
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7089851915573884e-05,
+      "loss": 0.9679,
+      "step": 1903
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708658871476625e-05,
+      "loss": 0.9885,
+      "step": 1904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083323997364015e-05,
+      "loss": 0.9854,
+      "step": 1905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080057764065865e-05,
+      "loss": 1.0575,
+      "step": 1906
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7076790015570798e-05,
+      "loss": 0.9674,
+      "step": 1907
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073520752578138e-05,
+      "loss": 1.0527,
+      "step": 1908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7070249975787548e-05,
+      "loss": 1.0206,
+      "step": 1909
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066977685898995e-05,
+      "loss": 1.034,
+      "step": 1910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063703883612782e-05,
+      "loss": 1.014,
+      "step": 1911
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060428569629532e-05,
+      "loss": 1.0071,
+      "step": 1912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057151744650197e-05,
+      "loss": 0.9589,
+      "step": 1913
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053873409376037e-05,
+      "loss": 1.04,
+      "step": 1914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705059356450866e-05,
+      "loss": 1.0347,
+      "step": 1915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047312210749964e-05,
+      "loss": 0.9889,
+      "step": 1916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044029348802205e-05,
+      "loss": 0.9441,
+      "step": 1917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040744979367942e-05,
+      "loss": 1.0808,
+      "step": 1918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703745910315005e-05,
+      "loss": 0.9806,
+      "step": 1919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703417172085175e-05,
+      "loss": 0.972,
+      "step": 1920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7030882833176563e-05,
+      "loss": 1.0174,
+      "step": 1921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027592440828344e-05,
+      "loss": 1.0201,
+      "step": 1922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024300544511267e-05,
+      "loss": 1.0368,
+      "step": 1923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021007144929825e-05,
+      "loss": 0.9499,
+      "step": 1924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701771224278884e-05,
+      "loss": 1.0826,
+      "step": 1925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701441583879345e-05,
+      "loss": 0.9751,
+      "step": 1926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011117933649114e-05,
+      "loss": 0.9432,
+      "step": 1927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007818528061616e-05,
+      "loss": 1.0363,
+      "step": 1928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004517622737054e-05,
+      "loss": 0.966,
+      "step": 1929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001215218381855e-05,
+      "loss": 1.0311,
+      "step": 1930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6997911315702765e-05,
+      "loss": 1.0487,
+      "step": 1931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994605915406848e-05,
+      "loss": 1.0253,
+      "step": 1932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699129901820149e-05,
+      "loss": 1.0059,
+      "step": 1933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987990624794395e-05,
+      "loss": 0.9882,
+      "step": 1934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984680735893594e-05,
+      "loss": 1.0189,
+      "step": 1935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698136935220743e-05,
+      "loss": 0.943,
+      "step": 1936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978056474444572e-05,
+      "loss": 0.9864,
+      "step": 1937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6974742103314e-05,
+      "loss": 1.002,
+      "step": 1938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971426239525026e-05,
+      "loss": 1.0295,
+      "step": 1939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6968108883787272e-05,
+      "loss": 1.0552,
+      "step": 1940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964790036810686e-05,
+      "loss": 0.9205,
+      "step": 1941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961469699305524e-05,
+      "loss": 0.9807,
+      "step": 1942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695814787198238e-05,
+      "loss": 0.92,
+      "step": 1943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954824555552137e-05,
+      "loss": 0.9917,
+      "step": 1944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951499750726036e-05,
+      "loss": 1.017,
+      "step": 1945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948173458215603e-05,
+      "loss": 0.998,
+      "step": 1946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69448456787327e-05,
+      "loss": 1.0079,
+      "step": 1947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941516412989498e-05,
+      "loss": 1.0051,
+      "step": 1948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693818566169849e-05,
+      "loss": 0.9487,
+      "step": 1949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934853425572495e-05,
+      "loss": 0.9807,
+      "step": 1950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931519705324634e-05,
+      "loss": 1.0554,
+      "step": 1951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928184501668356e-05,
+      "loss": 1.0213,
+      "step": 1952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924847815317424e-05,
+      "loss": 0.9793,
+      "step": 1953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6921509646985922e-05,
+      "loss": 1.0019,
+      "step": 1954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6918169997388246e-05,
+      "loss": 1.0161,
+      "step": 1955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691482886723911e-05,
+      "loss": 0.9908,
+      "step": 1956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691148625725355e-05,
+      "loss": 1.0197,
+      "step": 1957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908142168146916e-05,
+      "loss": 1.0068,
+      "step": 1958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904796600634868e-05,
+      "loss": 0.9954,
+      "step": 1959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901449555433393e-05,
+      "loss": 1.0073,
+      "step": 1960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898101033258785e-05,
+      "loss": 0.966,
+      "step": 1961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894751034827663e-05,
+      "loss": 1.0085,
+      "step": 1962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689139956085695e-05,
+      "loss": 0.9704,
+      "step": 1963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68880466120639e-05,
+      "loss": 1.0218,
+      "step": 1964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688469218916607e-05,
+      "loss": 0.9982,
+      "step": 1965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881336292881342e-05,
+      "loss": 0.3667,
+      "step": 1966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877978923927902e-05,
+      "loss": 1.0121,
+      "step": 1967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874620083024263e-05,
+      "loss": 0.9815,
+      "step": 1968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871259770889242e-05,
+      "loss": 0.928,
+      "step": 1969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867897988241983e-05,
+      "loss": 1.0209,
+      "step": 1970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686453473580193e-05,
+      "loss": 1.0293,
+      "step": 1971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861170014288855e-05,
+      "loss": 0.9579,
+      "step": 1972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685780382442284e-05,
+      "loss": 0.9443,
+      "step": 1973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854436166924276e-05,
+      "loss": 0.9421,
+      "step": 1974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851067042513876e-05,
+      "loss": 1.0233,
+      "step": 1975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847696451912664e-05,
+      "loss": 1.0209,
+      "step": 1976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844324395841976e-05,
+      "loss": 1.0052,
+      "step": 1977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684095087502346e-05,
+      "loss": 1.0256,
+      "step": 1978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6837575890179082e-05,
+      "loss": 0.9907,
+      "step": 1979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6834199442031123e-05,
+      "loss": 1.0241,
+      "step": 1980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830821531302167e-05,
+      "loss": 0.9575,
+      "step": 1981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827442158715125e-05,
+      "loss": 0.3983,
+      "step": 1982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682406132499321e-05,
+      "loss": 1.0175,
+      "step": 1983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820679030859947e-05,
+      "loss": 1.0109,
+      "step": 1984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817295277039183e-05,
+      "loss": 1.0127,
+      "step": 1985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681391006425507e-05,
+      "loss": 1.0049,
+      "step": 1986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810523393232082e-05,
+      "loss": 1.0087,
+      "step": 1987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807135264694983e-05,
+      "loss": 1.0002,
+      "step": 1988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803745679368873e-05,
+      "loss": 1.0071,
+      "step": 1989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800354637979158e-05,
+      "loss": 0.9907,
+      "step": 1990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796962141251544e-05,
+      "loss": 0.9551,
+      "step": 1991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793568189912055e-05,
+      "loss": 0.9579,
+      "step": 1992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790172784687034e-05,
+      "loss": 1.0108,
+      "step": 1993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786775926303124e-05,
+      "loss": 1.0206,
+      "step": 1994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678337761548729e-05,
+      "loss": 0.9537,
+      "step": 1995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6779977852966795e-05,
+      "loss": 0.9843,
+      "step": 1996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776576639469217e-05,
+      "loss": 0.9626,
+      "step": 1997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773173975722453e-05,
+      "loss": 1.0178,
+      "step": 1998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769769862454702e-05,
+      "loss": 1.0012,
+      "step": 1999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766364300394476e-05,
+      "loss": 1.0154,
+      "step": 2000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762957290270592e-05,
+      "loss": 0.9633,
+      "step": 2001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6759548832812188e-05,
+      "loss": 0.954,
+      "step": 2002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756138928748704e-05,
+      "loss": 1.0495,
+      "step": 2003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752727578809885e-05,
+      "loss": 0.981,
+      "step": 2004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749314783725794e-05,
+      "loss": 1.0503,
+      "step": 2005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745900544226802e-05,
+      "loss": 1.0866,
+      "step": 2006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742484861043584e-05,
+      "loss": 0.9717,
+      "step": 2007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673906773490713e-05,
+      "loss": 1.0103,
+      "step": 2008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735649166548738e-05,
+      "loss": 1.0201,
+      "step": 2009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732229156700006e-05,
+      "loss": 0.966,
+      "step": 2010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728807706092852e-05,
+      "loss": 0.9742,
+      "step": 2011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725384815459496e-05,
+      "loss": 0.9335,
+      "step": 2012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6721960485532473e-05,
+      "loss": 0.3612,
+      "step": 2013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718534717044614e-05,
+      "loss": 0.9846,
+      "step": 2014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671510751072907e-05,
+      "loss": 1.0298,
+      "step": 2015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671167886731929e-05,
+      "loss": 0.9852,
+      "step": 2016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708248787549037e-05,
+      "loss": 1.0491,
+      "step": 2017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704817272152377e-05,
+      "loss": 0.9934,
+      "step": 2018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701384321863692e-05,
+      "loss": 0.9415,
+      "step": 2019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697949937417657e-05,
+      "loss": 0.9835,
+      "step": 2020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694514119549267e-05,
+      "loss": 0.9598,
+      "step": 2021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691076868993812e-05,
+      "loss": 1.0687,
+      "step": 2022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687638186486904e-05,
+      "loss": 0.9651,
+      "step": 2023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6684198072764446e-05,
+      "loss": 0.9304,
+      "step": 2024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680756528562655e-05,
+      "loss": 0.9987,
+      "step": 2025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677313554618054e-05,
+      "loss": 0.9871,
+      "step": 2026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667386915166747e-05,
+      "loss": 1.0348,
+      "step": 2027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6670423320448036e-05,
+      "loss": 1.0448,
+      "step": 2028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666697606169719e-05,
+      "loss": 0.9378,
+      "step": 2029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663527376152683e-05,
+      "loss": 0.986,
+      "step": 2030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660077264552557e-05,
+      "loss": 1.0001,
+      "step": 2031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656625727635174e-05,
+      "loss": 1.0194,
+      "step": 2032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653172766139192e-05,
+      "loss": 1.0402,
+      "step": 2033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649718380803574e-05,
+      "loss": 0.9549,
+      "step": 2034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646262572367592e-05,
+      "loss": 1.0179,
+      "step": 2035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664280534157082e-05,
+      "loss": 0.9772,
+      "step": 2036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663934668915314e-05,
+      "loss": 0.9966,
+      "step": 2037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663588661585473e-05,
+      "loss": 0.9948,
+      "step": 2038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6632425122416077e-05,
+      "loss": 1.0455,
+      "step": 2039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662896220957798e-05,
+      "loss": 1.0813,
+      "step": 2040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625497878081523e-05,
+      "loss": 1.0239,
+      "step": 2041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662203212866812e-05,
+      "loss": 0.9884,
+      "step": 2042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6618564962079454e-05,
+      "loss": 0.956,
+      "step": 2043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661509637905754e-05,
+      "loss": 1.0705,
+      "step": 2044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611626380344688e-05,
+      "loss": 1.0581,
+      "step": 2045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660815496668351e-05,
+      "loss": 1.0299,
+      "step": 2046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604682138816914e-05,
+      "loss": 1.0256,
+      "step": 2047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660120789748812e-05,
+      "loss": 0.9653,
+      "step": 2048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597732243440647e-05,
+      "loss": 1.0111,
+      "step": 2049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659425517741832e-05,
+      "loss": 1.0005,
+      "step": 2050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659077670016526e-05,
+      "loss": 0.9592,
+      "step": 2051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6587296812425894e-05,
+      "loss": 0.9575,
+      "step": 2052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583815514944942e-05,
+      "loss": 0.9944,
+      "step": 2053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580332808467444e-05,
+      "loss": 1.0408,
+      "step": 2054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657684869373873e-05,
+      "loss": 0.9561,
+      "step": 2055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573363171504424e-05,
+      "loss": 0.9473,
+      "step": 2056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656987624251047e-05,
+      "loss": 0.9781,
+      "step": 2057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566387907503092e-05,
+      "loss": 0.9995,
+      "step": 2058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6562898167228832e-05,
+      "loss": 0.9686,
+      "step": 2059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559407022434527e-05,
+      "loss": 1.0454,
+      "step": 2060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655591447386731e-05,
+      "loss": 1.0494,
+      "step": 2061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6552420522274622e-05,
+      "loss": 0.9398,
+      "step": 2062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548925168404202e-05,
+      "loss": 0.9399,
+      "step": 2063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654542841300408e-05,
+      "loss": 0.9876,
+      "step": 2064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65419302568226e-05,
+      "loss": 0.9869,
+      "step": 2065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6538430700608394e-05,
+      "loss": 0.9946,
+      "step": 2066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534929745110404e-05,
+      "loss": 0.987,
+      "step": 2067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6531427391077862e-05,
+      "loss": 0.9444,
+      "step": 2068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652792363926031e-05,
+      "loss": 1.0142,
+      "step": 2069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524418490407577e-05,
+      "loss": 0.9735,
+      "step": 2070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65209119452698e-05,
+      "loss": 1.0503,
+      "step": 2071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517404004597407e-05,
+      "loss": 1.0309,
+      "step": 2072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513894669141136e-05,
+      "loss": 0.9714,
+      "step": 2073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510383939652013e-05,
+      "loss": 1.0265,
+      "step": 2074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506871816881366e-05,
+      "loss": 0.9575,
+      "step": 2075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503358301580823e-05,
+      "loss": 0.3641,
+      "step": 2076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499843394502307e-05,
+      "loss": 1.0312,
+      "step": 2077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649632709639804e-05,
+      "loss": 1.0015,
+      "step": 2078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649280940802054e-05,
+      "loss": 1.035,
+      "step": 2079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489290330122632e-05,
+      "loss": 0.9714,
+      "step": 2080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648576986345742e-05,
+      "loss": 0.9454,
+      "step": 2081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482248008778325e-05,
+      "loss": 0.9645,
+      "step": 2082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647872476683905e-05,
+      "loss": 1.0091,
+      "step": 2083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475200138393604e-05,
+      "loss": 0.9879,
+      "step": 2084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647167412419629e-05,
+      "loss": 1.0303,
+      "step": 2085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468146725001705e-05,
+      "loss": 0.9758,
+      "step": 2086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6464617941564746e-05,
+      "loss": 0.8957,
+      "step": 2087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461087774640604e-05,
+      "loss": 0.3642,
+      "step": 2088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457556224984767e-05,
+      "loss": 1.0163,
+      "step": 2089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454023293353024e-05,
+      "loss": 0.9959,
+      "step": 2090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450488980501445e-05,
+      "loss": 1.0873,
+      "step": 2091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644695328718641e-05,
+      "loss": 1.0249,
+      "step": 2092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443416214164593e-05,
+      "loss": 1.0224,
+      "step": 2093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439877762192954e-05,
+      "loss": 0.4082,
+      "step": 2094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436337932028755e-05,
+      "loss": 0.9966,
+      "step": 2095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432796724429555e-05,
+      "loss": 0.9762,
+      "step": 2096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429254140153205e-05,
+      "loss": 1.0706,
+      "step": 2097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425710179957847e-05,
+      "loss": 1.0109,
+      "step": 2098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642216484460192e-05,
+      "loss": 0.9923,
+      "step": 2099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641861813484416e-05,
+      "loss": 0.9603,
+      "step": 2100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415070051443597e-05,
+      "loss": 0.9114,
+      "step": 2101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411520595159554e-05,
+      "loss": 1.001,
+      "step": 2102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407969766751644e-05,
+      "loss": 1.0182,
+      "step": 2103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404417566979777e-05,
+      "loss": 0.9725,
+      "step": 2104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400863996604156e-05,
+      "loss": 0.9617,
+      "step": 2105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397309056385277e-05,
+      "loss": 0.9941,
+      "step": 2106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6393752747083934e-05,
+      "loss": 0.9038,
+      "step": 2107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390195069461202e-05,
+      "loss": 0.9972,
+      "step": 2108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386636024278463e-05,
+      "loss": 0.9983,
+      "step": 2109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383075612297384e-05,
+      "loss": 1.0288,
+      "step": 2110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379513834279925e-05,
+      "loss": 1.0417,
+      "step": 2111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375950690988337e-05,
+      "loss": 0.9669,
+      "step": 2112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372386183185164e-05,
+      "loss": 0.3788,
+      "step": 2113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636882031163325e-05,
+      "loss": 1.0063,
+      "step": 2114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365253077095714e-05,
+      "loss": 1.0092,
+      "step": 2115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636168448033599e-05,
+      "loss": 1.0187,
+      "step": 2116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635811452211778e-05,
+      "loss": 1.0445,
+      "step": 2117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354543203205086e-05,
+      "loss": 1.0099,
+      "step": 2118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635097052436221e-05,
+      "loss": 0.9683,
+      "step": 2119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6347396486353736e-05,
+      "loss": 0.9648,
+      "step": 2120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634382108994454e-05,
+      "loss": 1.0292,
+      "step": 2121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634024433589979e-05,
+      "loss": 0.9712,
+      "step": 2122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633666622498494e-05,
+      "loss": 0.9927,
+      "step": 2123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333086757965745e-05,
+      "loss": 1.0748,
+      "step": 2124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329505935608237e-05,
+      "loss": 1.0415,
+      "step": 2125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632592375867875e-05,
+      "loss": 0.996,
+      "step": 2126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322340227943895e-05,
+      "loss": 0.9677,
+      "step": 2127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318755344170592e-05,
+      "loss": 1.0198,
+      "step": 2128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6315169108126027e-05,
+      "loss": 1.0137,
+      "step": 2129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311581520577697e-05,
+      "loss": 0.9576,
+      "step": 2130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307992582293374e-05,
+      "loss": 1.0056,
+      "step": 2131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630440229404112e-05,
+      "loss": 1.0175,
+      "step": 2132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300810656589294e-05,
+      "loss": 0.9829,
+      "step": 2133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629721767070654e-05,
+      "loss": 1.06,
+      "step": 2134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6293623337161785e-05,
+      "loss": 0.9733,
+      "step": 2135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6290027656724253e-05,
+      "loss": 0.9765,
+      "step": 2136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286430630163455e-05,
+      "loss": 0.9836,
+      "step": 2137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628283225824918e-05,
+      "loss": 0.9815,
+      "step": 2138
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279232541751515e-05,
+      "loss": 0.9797,
+      "step": 2139
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275631481440837e-05,
+      "loss": 1.0196,
+      "step": 2140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62720290780878e-05,
+      "loss": 0.9835,
+      "step": 2141
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268425332463352e-05,
+      "loss": 0.9912,
+      "step": 2142
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626482024533873e-05,
+      "loss": 0.9724,
+      "step": 2143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261213817485458e-05,
+      "loss": 0.9911,
+      "step": 2144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6257606049675336e-05,
+      "loss": 0.9424,
+      "step": 2145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6253996942680466e-05,
+      "loss": 0.9954,
+      "step": 2146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250386497273227e-05,
+      "loss": 0.9979,
+      "step": 2147
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246774714226286e-05,
+      "loss": 0.9437,
+      "step": 2148
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62431615943126e-05,
+      "loss": 1.0048,
+      "step": 2149
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6239547138305408e-05,
+      "loss": 1.0327,
+      "step": 2150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623593134697824e-05,
+      "loss": 1.0187,
+      "step": 2151
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623231422110491e-05,
+      "loss": 0.9788,
+      "step": 2152
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228695761459505e-05,
+      "loss": 1.0209,
+      "step": 2153
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225075968816417e-05,
+      "loss": 0.9729,
+      "step": 2154
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221454843950314e-05,
+      "loss": 1.0375,
+      "step": 2155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217832387636148e-05,
+      "loss": 1.0446,
+      "step": 2156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621420860064916e-05,
+      "loss": 1.0072,
+      "step": 2157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210583483764867e-05,
+      "loss": 1.0044,
+      "step": 2158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6206957037759084e-05,
+      "loss": 0.9839,
+      "step": 2159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203329263407904e-05,
+      "loss": 0.9959,
+      "step": 2160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199700161487695e-05,
+      "loss": 0.937,
+      "step": 2161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619606973277513e-05,
+      "loss": 0.9866,
+      "step": 2162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6192437978047144e-05,
+      "loss": 0.9908,
+      "step": 2163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618880489808097e-05,
+      "loss": 0.9907,
+      "step": 2164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185170493654122e-05,
+      "loss": 0.9679,
+      "step": 2165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618153476554439e-05,
+      "loss": 0.9521,
+      "step": 2166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177897714529857e-05,
+      "loss": 1.0417,
+      "step": 2167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174259341388884e-05,
+      "loss": 1.0121,
+      "step": 2168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170619646900122e-05,
+      "loss": 0.3726,
+      "step": 2169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166978631842485e-05,
+      "loss": 0.9627,
+      "step": 2170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163336296995196e-05,
+      "loss": 0.9246,
+      "step": 2171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159692643137743e-05,
+      "loss": 1.0031,
+      "step": 2172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61560476710499e-05,
+      "loss": 0.9921,
+      "step": 2173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152401381511728e-05,
+      "loss": 0.9846,
+      "step": 2174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148753775303566e-05,
+      "loss": 0.3463,
+      "step": 2175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614510485320603e-05,
+      "loss": 0.9637,
+      "step": 2176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141454616000024e-05,
+      "loss": 0.9631,
+      "step": 2177
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137803064466737e-05,
+      "loss": 0.9747,
+      "step": 2178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613415019938763e-05,
+      "loss": 1.0253,
+      "step": 2179
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613049602154445e-05,
+      "loss": 1.108,
+      "step": 2180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126840531719223e-05,
+      "loss": 1.0848,
+      "step": 2181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612318373069426e-05,
+      "loss": 0.9973,
+      "step": 2182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119525619252153e-05,
+      "loss": 0.9911,
+      "step": 2183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115866198175765e-05,
+      "loss": 0.3456,
+      "step": 2184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6112205468248245e-05,
+      "loss": 0.9832,
+      "step": 2185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6108543430253026e-05,
+      "loss": 1.0331,
+      "step": 2186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104880084973818e-05,
+      "loss": 0.9635,
+      "step": 2187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101215433194606e-05,
+      "loss": 0.936,
+      "step": 2188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097549475699663e-05,
+      "loss": 1.0175,
+      "step": 2189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093882213273537e-05,
+      "loss": 1.0392,
+      "step": 2190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090213646701056e-05,
+      "loss": 0.9991,
+      "step": 2191
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086543776767323e-05,
+      "loss": 1.0265,
+      "step": 2192
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6082872604257724e-05,
+      "loss": 1.0098,
+      "step": 2193
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607920012995793e-05,
+      "loss": 1.0115,
+      "step": 2194
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075526354653877e-05,
+      "loss": 1.0465,
+      "step": 2195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071851279131787e-05,
+      "loss": 0.958,
+      "step": 2196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068174904178166e-05,
+      "loss": 1.025,
+      "step": 2197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064497230579787e-05,
+      "loss": 0.9372,
+      "step": 2198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60608182591237e-05,
+      "loss": 0.3668,
+      "step": 2199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605713799059725e-05,
+      "loss": 1.0459,
+      "step": 2200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053456425788037e-05,
+      "loss": 0.9827,
+      "step": 2201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604977356548396e-05,
+      "loss": 0.9951,
+      "step": 2202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6046089410473178e-05,
+      "loss": 1.0043,
+      "step": 2203
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042403961544133e-05,
+      "loss": 0.9691,
+      "step": 2204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6038717219485548e-05,
+      "loss": 0.9604,
+      "step": 2205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603502918508642e-05,
+      "loss": 0.9499,
+      "step": 2206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031339859136018e-05,
+      "loss": 0.9332,
+      "step": 2207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027649242423895e-05,
+      "loss": 0.9338,
+      "step": 2208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6023957335739876e-05,
+      "loss": 1.069,
+      "step": 2209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020264139874062e-05,
+      "loss": 0.9936,
+      "step": 2210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601656965561683e-05,
+      "loss": 1.0296,
+      "step": 2211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012873883758832e-05,
+      "loss": 0.966,
+      "step": 2212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009176825091e-05,
+      "loss": 0.3621,
+      "step": 2213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005478480404536e-05,
+      "loss": 0.9955,
+      "step": 2214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600177885049092e-05,
+      "loss": 1.0696,
+      "step": 2215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998077936141913e-05,
+      "loss": 0.996,
+      "step": 2216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5994375738149534e-05,
+      "loss": 0.9359,
+      "step": 2217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599067225730609e-05,
+      "loss": 0.9872,
+      "step": 2218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986967494404164e-05,
+      "loss": 0.9927,
+      "step": 2219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983261450236606e-05,
+      "loss": 0.9651,
+      "step": 2220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5979554125596544e-05,
+      "loss": 0.9824,
+      "step": 2221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5975845521277378e-05,
+      "loss": 0.9381,
+      "step": 2222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972135638072784e-05,
+      "loss": 0.9926,
+      "step": 2223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5968424476776713e-05,
+      "loss": 0.9496,
+      "step": 2224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596471203818338e-05,
+      "loss": 0.3836,
+      "step": 2225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596099832308729e-05,
+      "loss": 1.0242,
+      "step": 2226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5957283332283206e-05,
+      "loss": 1.0514,
+      "step": 2227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953567066566174e-05,
+      "loss": 1.0129,
+      "step": 2228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949849526731506e-05,
+      "loss": 0.3467,
+      "step": 2229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5946130713574787e-05,
+      "loss": 1.034,
+      "step": 2230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594241062789188e-05,
+      "loss": 0.9801,
+      "step": 2231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938689270478918e-05,
+      "loss": 0.9227,
+      "step": 2232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593496664213231e-05,
+      "loss": 1.0142,
+      "step": 2233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593124274364872e-05,
+      "loss": 0.9125,
+      "step": 2234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927517575825103e-05,
+      "loss": 0.9801,
+      "step": 2235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5923791139458683e-05,
+      "loss": 0.9684,
+      "step": 2236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592006343534694e-05,
+      "loss": 0.9771,
+      "step": 2237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591633446428765e-05,
+      "loss": 1.0122,
+      "step": 2238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912604227078842e-05,
+      "loss": 1.0243,
+      "step": 2239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908872724518816e-05,
+      "loss": 0.348,
+      "step": 2240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905139957406156e-05,
+      "loss": 1.0058,
+      "step": 2241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59014059265397e-05,
+      "loss": 1.0093,
+      "step": 2242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897670632718565e-05,
+      "loss": 0.9706,
+      "step": 2243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5893934076742144e-05,
+      "loss": 1.0105,
+      "step": 2244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589019625941009e-05,
+      "loss": 0.9839,
+      "step": 2245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588645718152233e-05,
+      "loss": 1.0455,
+      "step": 2246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588271684387906e-05,
+      "loss": 0.9898,
+      "step": 2247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5878975247280755e-05,
+      "loss": 0.9572,
+      "step": 2248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587523239252814e-05,
+      "loss": 0.9145,
+      "step": 2249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5871488280422223e-05,
+      "loss": 0.9882,
+      "step": 2250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586774291176428e-05,
+      "loss": 0.9941,
+      "step": 2251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5863996287355858e-05,
+      "loss": 1.0096,
+      "step": 2252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860248407998758e-05,
+      "loss": 0.9437,
+      "step": 2253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5856499274495064e-05,
+      "loss": 0.9501,
+      "step": 2254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852748887647133e-05,
+      "loss": 0.9544,
+      "step": 2255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848997248257573e-05,
+      "loss": 0.9911,
+      "step": 2256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584524435712927e-05,
+      "loss": 1.0011,
+      "step": 2257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841490215065382e-05,
+      "loss": 1.011,
+      "step": 2258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5837734822869325e-05,
+      "loss": 0.9376,
+      "step": 2259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833978181344788e-05,
+      "loss": 0.9666,
+      "step": 2260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583022029129573e-05,
+      "loss": 0.9684,
+      "step": 2261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582646115352637e-05,
+      "loss": 1.0338,
+      "step": 2262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822700768841198e-05,
+      "loss": 0.3946,
+      "step": 2263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581893913804497e-05,
+      "loss": 0.9947,
+      "step": 2264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5815176261942715e-05,
+      "loss": 0.9782,
+      "step": 2265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811412141339714e-05,
+      "loss": 1.0185,
+      "step": 2266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807646777041527e-05,
+      "loss": 1.007,
+      "step": 2267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580388016985398e-05,
+      "loss": 0.9808,
+      "step": 2268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800112320583157e-05,
+      "loss": 0.9716,
+      "step": 2269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796343230035413e-05,
+      "loss": 0.9776,
+      "step": 2270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579257289901737e-05,
+      "loss": 0.9663,
+      "step": 2271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5788801328335906e-05,
+      "loss": 0.9283,
+      "step": 2272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785028518798182e-05,
+      "loss": 0.9996,
+      "step": 2273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781254471211606e-05,
+      "loss": 1.0219,
+      "step": 2274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577747918638386e-05,
+      "loss": 0.9766,
+      "step": 2275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5773702665122894e-05,
+      "loss": 0.9958,
+      "step": 2276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576992490823691e-05,
+      "loss": 1.0471,
+      "step": 2277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576614591653439e-05,
+      "loss": 1.0083,
+      "step": 2278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762365690824067e-05,
+      "loss": 1.0392,
+      "step": 2279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5758584231914946e-05,
+      "loss": 0.9729,
+      "step": 2280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754801540616295e-05,
+      "loss": 0.9927,
+      "step": 2281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751017617737644e-05,
+      "loss": 1.0607,
+      "step": 2282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747232464088787e-05,
+      "loss": 1.0159,
+      "step": 2283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574344608047978e-05,
+      "loss": 1.0099,
+      "step": 2284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573965846772094e-05,
+      "loss": 1.0459,
+      "step": 2285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735869626622863e-05,
+      "loss": 0.9816,
+      "step": 2286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732079557996382e-05,
+      "loss": 0.9563,
+      "step": 2287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572828826265261e-05,
+      "loss": 1.0574,
+      "step": 2288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724495741402926e-05,
+      "loss": 0.8951,
+      "step": 2289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720701995058954e-05,
+      "loss": 0.9757,
+      "step": 2290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57169070244326e-05,
+      "loss": 1.0119,
+      "step": 2291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571311083033601e-05,
+      "loss": 0.9539,
+      "step": 2292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570931341358162e-05,
+      "loss": 0.9502,
+      "step": 2293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705514774982094e-05,
+      "loss": 0.9861,
+      "step": 2294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570171491535039e-05,
+      "loss": 0.9806,
+      "step": 2295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697913835499706e-05,
+      "loss": 1.0229,
+      "step": 2296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5694111536243507e-05,
+      "loss": 0.951,
+      "step": 2297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5690308018395526e-05,
+      "loss": 1.0277,
+      "step": 2298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686503282769747e-05,
+      "loss": 0.916,
+      "step": 2299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568269733018041e-05,
+      "loss": 0.98,
+      "step": 2300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5678890161442036e-05,
+      "loss": 1.0103,
+      "step": 2301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675081777369386e-05,
+      "loss": 0.9673,
+      "step": 2302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567127217877749e-05,
+      "loss": 0.9568,
+      "step": 2303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566746136648164e-05,
+      "loss": 1.0232,
+      "step": 2304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663649341297385e-05,
+      "loss": 1.0096,
+      "step": 2305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659836104040532e-05,
+      "loss": 1.0081,
+      "step": 2306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5656021655527144e-05,
+      "loss": 1.027,
+      "step": 2307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5652205996573554e-05,
+      "loss": 0.9219,
+      "step": 2308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648389127996342e-05,
+      "loss": 1.0829,
+      "step": 2309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564457105061236e-05,
+      "loss": 0.9339,
+      "step": 2310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640751765238703e-05,
+      "loss": 1.0286,
+      "step": 2311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636931272692738e-05,
+      "loss": 0.9869,
+      "step": 2312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5633109573792088e-05,
+      "loss": 1.0172,
+      "step": 2313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5629286669354625e-05,
+      "loss": 0.3525,
+      "step": 2314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625462560198488e-05,
+      "loss": 1.032,
+      "step": 2315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562163724714207e-05,
+      "loss": 1.1125,
+      "step": 2316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617810731004025e-05,
+      "loss": 1.0204,
+      "step": 2317
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561398301260326e-05,
+      "loss": 0.9637,
+      "step": 2318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561015409275895e-05,
+      "loss": 1.0131,
+      "step": 2319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5606323972290505e-05,
+      "loss": 1.0193,
+      "step": 2320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602492652017614e-05,
+      "loss": 0.9498,
+      "step": 2321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5598660132760218e-05,
+      "loss": 0.956,
+      "step": 2322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55948264153385e-05,
+      "loss": 0.9469,
+      "step": 2323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559099150057292e-05,
+      "loss": 0.9684,
+      "step": 2324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558715538928418e-05,
+      "loss": 0.9449,
+      "step": 2325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583318082293245e-05,
+      "loss": 0.9984,
+      "step": 2326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5579479580421333e-05,
+      "loss": 0.9624,
+      "step": 2327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575639884489917e-05,
+      "loss": 0.9653,
+      "step": 2328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571798995320732e-05,
+      "loss": 0.9584,
+      "step": 2329
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5567956913735755e-05,
+      "loss": 0.9999,
+      "step": 2330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564113640557233e-05,
+      "loss": 1.0039,
+      "step": 2331
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556026917660766e-05,
+      "loss": 0.9651,
+      "step": 2332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556423522709783e-05,
+      "loss": 1.0028,
+      "step": 2333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555257667968661e-05,
+      "loss": 0.947,
+      "step": 2334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5548728648361406e-05,
+      "loss": 1.0063,
+      "step": 2335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544879429557672e-05,
+      "loss": 0.9752,
+      "step": 2336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5541029024099187e-05,
+      "loss": 0.9985,
+      "step": 2337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553717743280997e-05,
+      "loss": 1.0007,
+      "step": 2338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533324656514293e-05,
+      "loss": 0.3862,
+      "step": 2339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552947069603669e-05,
+      "loss": 0.9889,
+      "step": 2340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552561555220194e-05,
+      "loss": 1.001,
+      "step": 2341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521759225835083e-05,
+      "loss": 0.3343,
+      "step": 2342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55179017177614e-05,
+      "loss": 1.0159,
+      "step": 2343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514043028806444e-05,
+      "loss": 1.0107,
+      "step": 2344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5510183159796004e-05,
+      "loss": 0.9864,
+      "step": 2345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5506322111556124e-05,
+      "loss": 0.9927,
+      "step": 2346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502459884913105e-05,
+      "loss": 1.0346,
+      "step": 2347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498596480693505e-05,
+      "loss": 0.9263,
+      "step": 2348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549473189972412e-05,
+      "loss": 1.0388,
+      "step": 2349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549086614283201e-05,
+      "loss": 0.9891,
+      "step": 2350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548699921084448e-05,
+      "loss": 1.0254,
+      "step": 2351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548313110458909e-05,
+      "loss": 0.9537,
+      "step": 2352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5479261824893653e-05,
+      "loss": 0.9929,
+      "step": 2353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475391372586225e-05,
+      "loss": 0.3615,
+      "step": 2354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5471519748495115e-05,
+      "loss": 1.0632,
+      "step": 2355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467646953448895e-05,
+      "loss": 0.9392,
+      "step": 2356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546377298827637e-05,
+      "loss": 1.0305,
+      "step": 2357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459897853806615e-05,
+      "loss": 0.9531,
+      "step": 2358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456021550868934e-05,
+      "loss": 1.0057,
+      "step": 2359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5452144080292893e-05,
+      "loss": 0.9797,
+      "step": 2360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448265442908307e-05,
+      "loss": 1.0405,
+      "step": 2361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5444385639545244e-05,
+      "loss": 0.9651,
+      "step": 2362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5440504671034005e-05,
+      "loss": 0.9822,
+      "step": 2363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543662253820517e-05,
+      "loss": 0.3981,
+      "step": 2364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432739241889535e-05,
+      "loss": 1.0071,
+      "step": 2365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428854782918176e-05,
+      "loss": 0.9691,
+      "step": 2366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424969162122385e-05,
+      "loss": 0.9805,
+      "step": 2367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421082380333736e-05,
+      "loss": 0.9561,
+      "step": 2368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417194438384025e-05,
+      "loss": 1.0434,
+      "step": 2369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5413305337105317e-05,
+      "loss": 1.0059,
+      "step": 2370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5409415077329906e-05,
+      "loss": 1.0048,
+      "step": 2371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5405523659890348e-05,
+      "loss": 0.9916,
+      "step": 2372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540163108561944e-05,
+      "loss": 1.0153,
+      "step": 2373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5397737355350228e-05,
+      "loss": 0.9825,
+      "step": 2374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393842469916003e-05,
+      "loss": 1.0262,
+      "step": 2375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389946430150313e-05,
+      "loss": 0.9345,
+      "step": 2376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386049236886945e-05,
+      "loss": 0.9779,
+      "step": 2377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538215089095993e-05,
+      "loss": 0.889,
+      "step": 2378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537825139320355e-05,
+      "loss": 1.0368,
+      "step": 2379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374350744452334e-05,
+      "loss": 0.3782,
+      "step": 2380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370448945541055e-05,
+      "loss": 1.0105,
+      "step": 2381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536654599730474e-05,
+      "loss": 0.3524,
+      "step": 2382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362641900578646e-05,
+      "loss": 1.0184,
+      "step": 2383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358736656198292e-05,
+      "loss": 0.9874,
+      "step": 2384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354830264999434e-05,
+      "loss": 0.9778,
+      "step": 2385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535092272781807e-05,
+      "loss": 0.9845,
+      "step": 2386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534701404549046e-05,
+      "loss": 1.0059,
+      "step": 2387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534310421885309e-05,
+      "loss": 1.02,
+      "step": 2388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53391932487427e-05,
+      "loss": 0.9885,
+      "step": 2389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5335281135996273e-05,
+      "loss": 0.9891,
+      "step": 2390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331367881451037e-05,
+      "loss": 0.9788,
+      "step": 2391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532745348594446e-05,
+      "loss": 1.0008,
+      "step": 2392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5323537950314268e-05,
+      "loss": 0.9823,
+      "step": 2393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531962127539841e-05,
+      "loss": 1.0048,
+      "step": 2394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53157034620351e-05,
+      "loss": 0.9887,
+      "step": 2395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5311784511062785e-05,
+      "loss": 0.9918,
+      "step": 2396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5307864423320146e-05,
+      "loss": 0.9463,
+      "step": 2397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530394319964613e-05,
+      "loss": 0.9655,
+      "step": 2398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5300020840879904e-05,
+      "loss": 0.9944,
+      "step": 2399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5296097347860896e-05,
+      "loss": 0.9673,
+      "step": 2400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292172721428765e-05,
+      "loss": 0.969,
+      "step": 2401
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5288246962423422e-05,
+      "loss": 0.9282,
+      "step": 2402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528432007168501e-05,
+      "loss": 0.9159,
+      "step": 2403
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5280392050053926e-05,
+      "loss": 0.9507,
+      "step": 2404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5276462898370792e-05,
+      "loss": 0.364,
+      "step": 2405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527253261747649e-05,
+      "loss": 1.0042,
+      "step": 2406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268601208212134e-05,
+      "loss": 0.9052,
+      "step": 2407
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526466867141908e-05,
+      "loss": 0.3946,
+      "step": 2408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260735007938928e-05,
+      "loss": 0.9904,
+      "step": 2409
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256800218613523e-05,
+      "loss": 1.0182,
+      "step": 2410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252864304284937e-05,
+      "loss": 1.0018,
+      "step": 2411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248927265795496e-05,
+      "loss": 0.986,
+      "step": 2412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244989103987764e-05,
+      "loss": 0.9686,
+      "step": 2413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241049819704536e-05,
+      "loss": 1.0149,
+      "step": 2414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237109413788864e-05,
+      "loss": 1.028,
+      "step": 2415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5233167887084024e-05,
+      "loss": 1.0143,
+      "step": 2416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229225240433541e-05,
+      "loss": 1.0348,
+      "step": 2417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225281474681181e-05,
+      "loss": 1.0264,
+      "step": 2418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221336590670941e-05,
+      "loss": 0.9824,
+      "step": 2419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217390589247062e-05,
+      "loss": 0.9982,
+      "step": 2420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5213443471254027e-05,
+      "loss": 0.9856,
+      "step": 2421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209495237536552e-05,
+      "loss": 0.9695,
+      "step": 2422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.52055458889396e-05,
+      "loss": 1.0251,
+      "step": 2423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201595426308364e-05,
+      "loss": 1.0384,
+      "step": 2424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519764385048828e-05,
+      "loss": 0.9559,
+      "step": 2425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519369116232502e-05,
+      "loss": 1.0382,
+      "step": 2426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189737362664496e-05,
+      "loss": 1.0243,
+      "step": 2427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185782452352857e-05,
+      "loss": 0.9976,
+      "step": 2428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181826432236494e-05,
+      "loss": 1.0085,
+      "step": 2429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177869303162026e-05,
+      "loss": 0.9471,
+      "step": 2430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173911065976315e-05,
+      "loss": 0.9381,
+      "step": 2431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516995172152646e-05,
+      "loss": 1.0066,
+      "step": 2432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51659912706598e-05,
+      "loss": 0.9326,
+      "step": 2433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162029714223906e-05,
+      "loss": 0.9239,
+      "step": 2434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158067053066588e-05,
+      "loss": 1.0652,
+      "step": 2435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515410328803589e-05,
+      "loss": 0.8968,
+      "step": 2436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150138419980092e-05,
+      "loss": 0.9705,
+      "step": 2437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146172449747716e-05,
+      "loss": 0.9902,
+      "step": 2438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142205378187515e-05,
+      "loss": 0.9937,
+      "step": 2439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138237206148476e-05,
+      "loss": 1.0082,
+      "step": 2440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5134267934479826e-05,
+      "loss": 1.0,
+      "step": 2441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5130297564031025e-05,
+      "loss": 0.9835,
+      "step": 2442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126326095651767e-05,
+      "loss": 1.0169,
+      "step": 2443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122353530191985e-05,
+      "loss": 1.0434,
+      "step": 2444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118379868501844e-05,
+      "loss": 1.0215,
+      "step": 2445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114405111431742e-05,
+      "loss": 0.9658,
+      "step": 2446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110429259832313e-05,
+      "loss": 0.9531,
+      "step": 2447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106452314554427e-05,
+      "loss": 1.0121,
+      "step": 2448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102474276449188e-05,
+      "loss": 0.9696,
+      "step": 2449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098495146367928e-05,
+      "loss": 0.9736,
+      "step": 2450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094514925162222e-05,
+      "loss": 0.989,
+      "step": 2451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509053361368387e-05,
+      "loss": 0.9946,
+      "step": 2452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508655121278491e-05,
+      "loss": 0.4596,
+      "step": 2453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082567723317614e-05,
+      "loss": 0.9266,
+      "step": 2454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507858314613448e-05,
+      "loss": 1.0975,
+      "step": 2455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074597482088251e-05,
+      "loss": 0.9847,
+      "step": 2456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070610732031892e-05,
+      "loss": 0.9927,
+      "step": 2457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066622896818604e-05,
+      "loss": 0.9697,
+      "step": 2458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062633977301814e-05,
+      "loss": 0.9786,
+      "step": 2459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5058643974335199e-05,
+      "loss": 1.0129,
+      "step": 2460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054652888772646e-05,
+      "loss": 1.0058,
+      "step": 2461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050660721468289e-05,
+      "loss": 0.3633,
+      "step": 2462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5046667473276485e-05,
+      "loss": 1.0031,
+      "step": 2463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042673145051831e-05,
+      "loss": 1.0099,
+      "step": 2464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038677737649146e-05,
+      "loss": 1.0112,
+      "step": 2465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503468125192348e-05,
+      "loss": 0.9976,
+      "step": 2466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030683688730123e-05,
+      "loss": 1.0302,
+      "step": 2467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502668504892459e-05,
+      "loss": 1.0232,
+      "step": 2468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022685333362618e-05,
+      "loss": 0.9876,
+      "step": 2469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018684542900194e-05,
+      "loss": 0.9883,
+      "step": 2470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014682678393517e-05,
+      "loss": 0.3771,
+      "step": 2471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010679740699026e-05,
+      "loss": 0.9705,
+      "step": 2472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5006675730673382e-05,
+      "loss": 0.9735,
+      "step": 2473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002670649173483e-05,
+      "loss": 0.9632,
+      "step": 2474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499866449705645e-05,
+      "loss": 0.933,
+      "step": 2475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499465727517964e-05,
+      "loss": 0.405,
+      "step": 2476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499064898440063e-05,
+      "loss": 1.0473,
+      "step": 2477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986639625577237e-05,
+      "loss": 0.9529,
+      "step": 2478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982629199567497e-05,
+      "loss": 0.9603,
+      "step": 2479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4978617707229674e-05,
+      "loss": 0.9895,
+      "step": 2480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497460514942227e-05,
+      "loss": 0.4054,
+      "step": 2481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970591527004005e-05,
+      "loss": 1.0227,
+      "step": 2482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966576840833836e-05,
+      "loss": 0.9602,
+      "step": 2483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962561091770936e-05,
+      "loss": 1.0351,
+      "step": 2484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958544280674719e-05,
+      "loss": 0.9508,
+      "step": 2485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.495452640840481e-05,
+      "loss": 0.9763,
+      "step": 2486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.495050747582108e-05,
+      "loss": 0.9343,
+      "step": 2487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946487483783608e-05,
+      "loss": 0.9677,
+      "step": 2488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942466433152713e-05,
+      "loss": 0.9801,
+      "step": 2489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938444324788939e-05,
+      "loss": 0.9685,
+      "step": 2490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934421159553051e-05,
+      "loss": 1.0261,
+      "step": 2491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930396938306049e-05,
+      "loss": 0.9566,
+      "step": 2492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926371661909142e-05,
+      "loss": 0.9169,
+      "step": 2493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922345331223785e-05,
+      "loss": 0.9911,
+      "step": 2494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918317947111642e-05,
+      "loss": 0.9397,
+      "step": 2495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491428951043462e-05,
+      "loss": 0.4191,
+      "step": 2496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491026002205483e-05,
+      "loss": 0.3368,
+      "step": 2497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4906229482834627e-05,
+      "loss": 0.9857,
+      "step": 2498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4902197893636581e-05,
+      "loss": 1.0302,
+      "step": 2499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489816525532349e-05,
+      "loss": 0.9993,
+      "step": 2500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489413156875837e-05,
+      "loss": 0.9554,
+      "step": 2501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890096834804472e-05,
+      "loss": 0.9618,
+      "step": 2502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886061054325264e-05,
+      "loss": 0.9565,
+      "step": 2503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882024228184442e-05,
+      "loss": 1.0115,
+      "step": 2504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487798635724592e-05,
+      "loss": 0.962,
+      "step": 2505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873947442373846e-05,
+      "loss": 0.967,
+      "step": 2506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869907484432578e-05,
+      "loss": 0.9923,
+      "step": 2507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865866484286701e-05,
+      "loss": 0.9499,
+      "step": 2508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861824442801032e-05,
+      "loss": 0.3814,
+      "step": 2509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857781360840604e-05,
+      "loss": 0.9878,
+      "step": 2510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853737239270674e-05,
+      "loss": 0.3635,
+      "step": 2511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484969207895672e-05,
+      "loss": 1.0524,
+      "step": 2512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845645880764439e-05,
+      "loss": 1.0396,
+      "step": 2513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484159864555976e-05,
+      "loss": 0.9328,
+      "step": 2514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837550374208827e-05,
+      "loss": 0.9982,
+      "step": 2515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833501067578005e-05,
+      "loss": 0.9967,
+      "step": 2516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4829450726533885e-05,
+      "loss": 0.9897,
+      "step": 2517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825399351943275e-05,
+      "loss": 0.9828,
+      "step": 2518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821346944673207e-05,
+      "loss": 0.9848,
+      "step": 2519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817293505590933e-05,
+      "loss": 0.9963,
+      "step": 2520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813239035563924e-05,
+      "loss": 0.9997,
+      "step": 2521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809183535459876e-05,
+      "loss": 0.3512,
+      "step": 2522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805127006146702e-05,
+      "loss": 0.9403,
+      "step": 2523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801069448492538e-05,
+      "loss": 0.3474,
+      "step": 2524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797010863365735e-05,
+      "loss": 0.9678,
+      "step": 2525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4792951251634872e-05,
+      "loss": 0.9438,
+      "step": 2526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478889061416874e-05,
+      "loss": 0.9368,
+      "step": 2527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478482895183635e-05,
+      "loss": 0.3902,
+      "step": 2528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780766265506939e-05,
+      "loss": 0.9213,
+      "step": 2529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776702556049956e-05,
+      "loss": 0.8948,
+      "step": 2530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772637824335072e-05,
+      "loss": 0.9764,
+      "step": 2531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768572071232179e-05,
+      "loss": 1.0312,
+      "step": 2532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764505297611387e-05,
+      "loss": 1.0338,
+      "step": 2533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760437504343017e-05,
+      "loss": 1.0006,
+      "step": 2534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756368692297613e-05,
+      "loss": 0.9455,
+      "step": 2535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4752298862345947e-05,
+      "loss": 1.0193,
+      "step": 2536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474822801535899e-05,
+      "loss": 0.9588,
+      "step": 2537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744156152207944e-05,
+      "loss": 1.0022,
+      "step": 2538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474008327376423e-05,
+      "loss": 0.9844,
+      "step": 2539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736009380899472e-05,
+      "loss": 0.8913,
+      "step": 2540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4731934474485526e-05,
+      "loss": 0.9179,
+      "step": 2541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727858555394457e-05,
+      "loss": 0.9759,
+      "step": 2542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472378162449855e-05,
+      "loss": 0.941,
+      "step": 2543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719703682670306e-05,
+      "loss": 0.9777,
+      "step": 2544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4715624730782438e-05,
+      "loss": 0.9753,
+      "step": 2545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711544769707886e-05,
+      "loss": 1.0143,
+      "step": 2546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470746380031979e-05,
+      "loss": 1.0,
+      "step": 2547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703381823491523e-05,
+      "loss": 0.9962,
+      "step": 2548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699298840096658e-05,
+      "loss": 0.9442,
+      "step": 2549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4695214851008994e-05,
+      "loss": 0.9959,
+      "step": 2550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4691129857102544e-05,
+      "loss": 0.949,
+      "step": 2551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468704385925153e-05,
+      "loss": 0.9596,
+      "step": 2552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.46829568583304e-05,
+      "loss": 0.9862,
+      "step": 2553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.46788688552138e-05,
+      "loss": 0.9745,
+      "step": 2554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674779850776602e-05,
+      "loss": 1.0026,
+      "step": 2555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4670689845893895e-05,
+      "loss": 0.987,
+      "step": 2556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666598841440974e-05,
+      "loss": 1.0118,
+      "step": 2557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662506838293351e-05,
+      "loss": 0.9627,
+      "step": 2558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658413837326753e-05,
+      "loss": 0.9952,
+      "step": 2559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465431983941712e-05,
+      "loss": 0.9873,
+      "step": 2560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650224845440606e-05,
+      "loss": 1.021,
+      "step": 2561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646128856273574e-05,
+      "loss": 1.0252,
+      "step": 2562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642031872792604e-05,
+      "loss": 0.3741,
+      "step": 2563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637933895874488e-05,
+      "loss": 1.0383,
+      "step": 2564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463383492639623e-05,
+      "loss": 0.9588,
+      "step": 2565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629734965235048e-05,
+      "loss": 1.0155,
+      "step": 2566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462563401326837e-05,
+      "loss": 0.3411,
+      "step": 2567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621532071373838e-05,
+      "loss": 0.976,
+      "step": 2568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617429140429304e-05,
+      "loss": 0.9873,
+      "step": 2569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4613325221312836e-05,
+      "loss": 1.0105,
+      "step": 2570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4609220314902706e-05,
+      "loss": 0.9408,
+      "step": 2571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46051144220774e-05,
+      "loss": 0.9963,
+      "step": 2572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601007543715625e-05,
+      "loss": 0.9789,
+      "step": 2573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459689968069628e-05,
+      "loss": 1.016,
+      "step": 2574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459279083389849e-05,
+      "loss": 0.9677,
+      "step": 2575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4588681004201586e-05,
+      "loss": 0.9924,
+      "step": 2576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458457019248511e-05,
+      "loss": 0.9117,
+      "step": 2577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458045839962881e-05,
+      "loss": 1.0302,
+      "step": 2578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576345626512649e-05,
+      "loss": 0.9434,
+      "step": 2579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45722318740168e-05,
+      "loss": 0.9345,
+      "step": 2580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568117143021639e-05,
+      "loss": 0.9587,
+      "step": 2581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456400143440776e-05,
+      "loss": 0.9458,
+      "step": 2582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559884749055959e-05,
+      "loss": 1.0341,
+      "step": 2583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4555767087847246e-05,
+      "loss": 0.9823,
+      "step": 2584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551648451662836e-05,
+      "loss": 0.9328,
+      "step": 2585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547528841384158e-05,
+      "loss": 0.9457,
+      "step": 2586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543408257892844e-05,
+      "loss": 0.9387,
+      "step": 2587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453928670207074e-05,
+      "loss": 1.0007,
+      "step": 2588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453516417479989e-05,
+      "loss": 0.9974,
+      "step": 2589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531040676962558e-05,
+      "loss": 0.9467,
+      "step": 2590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526916209441207e-05,
+      "loss": 0.9878,
+      "step": 2591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522790773118511e-05,
+      "loss": 1.037,
+      "step": 2592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518664368877353e-05,
+      "loss": 1.0041,
+      "step": 2593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514536997600819e-05,
+      "loss": 1.0217,
+      "step": 2594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510408660172207e-05,
+      "loss": 0.987,
+      "step": 2595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506279357475015e-05,
+      "loss": 1.0313,
+      "step": 2596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502149090392953e-05,
+      "loss": 0.9634,
+      "step": 2597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498017859809933e-05,
+      "loss": 0.9456,
+      "step": 2598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493885666610084e-05,
+      "loss": 0.9228,
+      "step": 2599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489752511677728e-05,
+      "loss": 0.9597,
+      "step": 2600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485618395897396e-05,
+      "loss": 0.9133,
+      "step": 2601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481483320153828e-05,
+      "loss": 0.9734,
+      "step": 2602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447734728533197e-05,
+      "loss": 0.9448,
+      "step": 2603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473210292316968e-05,
+      "loss": 1.0005,
+      "step": 2604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469072341994177e-05,
+      "loss": 1.0084,
+      "step": 2605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446493343524916e-05,
+      "loss": 0.9356,
+      "step": 2606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460793572967675e-05,
+      "loss": 1.0159,
+      "step": 2607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456652756035693e-05,
+      "loss": 0.9798,
+      "step": 2608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452510985339387e-05,
+      "loss": 0.9493,
+      "step": 2609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448368261765134e-05,
+      "loss": 1.0281,
+      "step": 2610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444224586199514e-05,
+      "loss": 0.9751,
+      "step": 2611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4440079959529307e-05,
+      "loss": 0.9891,
+      "step": 2612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435934382641508e-05,
+      "loss": 0.9805,
+      "step": 2613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431787856423304e-05,
+      "loss": 0.9144,
+      "step": 2614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427640381762093e-05,
+      "loss": 0.934,
+      "step": 2615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442349195954547e-05,
+      "loss": 0.9604,
+      "step": 2616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419342590661232e-05,
+      "loss": 0.9658,
+      "step": 2617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415192275997387e-05,
+      "loss": 0.9816,
+      "step": 2618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411041016442138e-05,
+      "loss": 1.0139,
+      "step": 2619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406888812883892e-05,
+      "loss": 1.0098,
+      "step": 2620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440273566621126e-05,
+      "loss": 1.0011,
+      "step": 2621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4398581577313053e-05,
+      "loss": 1.049,
+      "step": 2622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394426547078282e-05,
+      "loss": 0.9681,
+      "step": 2623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439027057639616e-05,
+      "loss": 1.0333,
+      "step": 2624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438611366615611e-05,
+      "loss": 1.0225,
+      "step": 2625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4381955817247737e-05,
+      "loss": 0.9421,
+      "step": 2626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377797030560872e-05,
+      "loss": 0.9427,
+      "step": 2627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373637306985523e-05,
+      "loss": 0.9671,
+      "step": 2628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369476647411915e-05,
+      "loss": 0.999,
+      "step": 2629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365315052730461e-05,
+      "loss": 1.0293,
+      "step": 2630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361152523831788e-05,
+      "loss": 1.0324,
+      "step": 2631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4356989061606707e-05,
+      "loss": 0.9797,
+      "step": 2632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352824666946243e-05,
+      "loss": 0.9646,
+      "step": 2633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434865934074161e-05,
+      "loss": 0.9411,
+      "step": 2634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344493083884233e-05,
+      "loss": 1.0403,
+      "step": 2635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340325897265722e-05,
+      "loss": 1.0108,
+      "step": 2636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433615778177789e-05,
+      "loss": 0.9605,
+      "step": 2637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4331988738312762e-05,
+      "loss": 1.0018,
+      "step": 2638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327818767762542e-05,
+      "loss": 1.0276,
+      "step": 2639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4323647871019648e-05,
+      "loss": 0.9286,
+      "step": 2640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319476048976687e-05,
+      "loss": 0.9732,
+      "step": 2641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315303302526469e-05,
+      "loss": 0.9635,
+      "step": 2642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311129632561994e-05,
+      "loss": 0.3696,
+      "step": 2643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4306955039976473e-05,
+      "loss": 1.0311,
+      "step": 2644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302779525663305e-05,
+      "loss": 1.009,
+      "step": 2645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298603090516083e-05,
+      "loss": 0.9436,
+      "step": 2646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294425735428608e-05,
+      "loss": 0.9458,
+      "step": 2647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290247461294868e-05,
+      "loss": 0.9817,
+      "step": 2648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286068269009058e-05,
+      "loss": 0.9729,
+      "step": 2649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281888159465555e-05,
+      "loss": 0.9311,
+      "step": 2650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277707133558945e-05,
+      "loss": 0.9581,
+      "step": 2651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273525192184005e-05,
+      "loss": 1.0041,
+      "step": 2652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269342336235709e-05,
+      "loss": 1.0234,
+      "step": 2653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265158566609228e-05,
+      "loss": 0.9357,
+      "step": 2654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4260973884199922e-05,
+      "loss": 0.941,
+      "step": 2655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4256788289903356e-05,
+      "loss": 0.9549,
+      "step": 2656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4252601784615283e-05,
+      "loss": 0.9621,
+      "step": 2657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248414369231656e-05,
+      "loss": 0.9564,
+      "step": 2658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244226044648615e-05,
+      "loss": 0.9845,
+      "step": 2659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240036811762503e-05,
+      "loss": 0.9439,
+      "step": 2660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235846671469852e-05,
+      "loss": 0.3493,
+      "step": 2661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231655624667398e-05,
+      "loss": 0.9592,
+      "step": 2662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227463672252052e-05,
+      "loss": 1.0082,
+      "step": 2663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4223270815120935e-05,
+      "loss": 0.9981,
+      "step": 2664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219077054171358e-05,
+      "loss": 1.0054,
+      "step": 2665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214882390300821e-05,
+      "loss": 1.0065,
+      "step": 2666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210686824407023e-05,
+      "loss": 0.9407,
+      "step": 2667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206490357387852e-05,
+      "loss": 1.0025,
+      "step": 2668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202292990141392e-05,
+      "loss": 0.9563,
+      "step": 2669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198094723565919e-05,
+      "loss": 1.0285,
+      "step": 2670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193895558559893e-05,
+      "loss": 0.9249,
+      "step": 2671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189695496021981e-05,
+      "loss": 1.0,
+      "step": 2672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418549453685103e-05,
+      "loss": 0.993,
+      "step": 2673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181292681946088e-05,
+      "loss": 0.3577,
+      "step": 2674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4177089932206384e-05,
+      "loss": 0.9787,
+      "step": 2675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172886288531356e-05,
+      "loss": 0.9068,
+      "step": 2676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416868175182061e-05,
+      "loss": 0.9804,
+      "step": 2677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4164476322973958e-05,
+      "loss": 0.9859,
+      "step": 2678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160270002891405e-05,
+      "loss": 1.0189,
+      "step": 2679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156062792473137e-05,
+      "loss": 0.9921,
+      "step": 2680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415185469261954e-05,
+      "loss": 0.905,
+      "step": 2681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147645704231181e-05,
+      "loss": 0.9566,
+      "step": 2682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143435828208827e-05,
+      "loss": 0.9883,
+      "step": 2683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139225065453426e-05,
+      "loss": 1.0181,
+      "step": 2684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135013416866116e-05,
+      "loss": 0.9099,
+      "step": 2685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130800883348235e-05,
+      "loss": 1.0105,
+      "step": 2686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126587465801303e-05,
+      "loss": 0.9848,
+      "step": 2687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122373165127025e-05,
+      "loss": 0.9603,
+      "step": 2688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118157982227306e-05,
+      "loss": 0.9725,
+      "step": 2689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4113941918004228e-05,
+      "loss": 0.9705,
+      "step": 2690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410972497336007e-05,
+      "loss": 0.9296,
+      "step": 2691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105507149197298e-05,
+      "loss": 1.0261,
+      "step": 2692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410128844641856e-05,
+      "loss": 1.0006,
+      "step": 2693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097068865926702e-05,
+      "loss": 0.3698,
+      "step": 2694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092848408624748e-05,
+      "loss": 0.9721,
+      "step": 2695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4088627075415922e-05,
+      "loss": 0.9936,
+      "step": 2696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4084404867203618e-05,
+      "loss": 1.0116,
+      "step": 2697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080181784891433e-05,
+      "loss": 0.9626,
+      "step": 2698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075957829383145e-05,
+      "loss": 0.9576,
+      "step": 2699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4071733001582715e-05,
+      "loss": 0.9652,
+      "step": 2700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067507302394299e-05,
+      "loss": 0.9872,
+      "step": 2701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063280732722232e-05,
+      "loss": 1.0187,
+      "step": 2702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4059053293471045e-05,
+      "loss": 1.026,
+      "step": 2703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054824985545442e-05,
+      "loss": 0.907,
+      "step": 2704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050595809850319e-05,
+      "loss": 0.9713,
+      "step": 2705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404636576729076e-05,
+      "loss": 0.9804,
+      "step": 2706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042134858772031e-05,
+      "loss": 0.9867,
+      "step": 2707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403790308519959e-05,
+      "loss": 0.9395,
+      "step": 2708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403367044747907e-05,
+      "loss": 0.3685,
+      "step": 2709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029436946516301e-05,
+      "loss": 0.9487,
+      "step": 2710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4025202583217281e-05,
+      "loss": 0.9614,
+      "step": 2711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020967358488209e-05,
+      "loss": 1.0472,
+      "step": 2712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401673127323546e-05,
+      "loss": 0.8765,
+      "step": 2713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012494328365594e-05,
+      "loss": 0.9838,
+      "step": 2714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008256524785355e-05,
+      "loss": 1.0081,
+      "step": 2715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004017863401678e-05,
+      "loss": 0.968,
+      "step": 2716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999778345121668e-05,
+      "loss": 0.9616,
+      "step": 2717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995537970852625e-05,
+      "loss": 1.0067,
+      "step": 2718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399129674150202e-05,
+      "loss": 1.0401,
+      "step": 2719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987054657977526e-05,
+      "loss": 0.9177,
+      "step": 2720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398281172118698e-05,
+      "loss": 1.0069,
+      "step": 2721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397856793203841e-05,
+      "loss": 1.0473,
+      "step": 2722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974323291440031e-05,
+      "loss": 1.0301,
+      "step": 2723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970077800300228e-05,
+      "loss": 1.0433,
+      "step": 2724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3965831459527577e-05,
+      "loss": 1.032,
+      "step": 2725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961584270030834e-05,
+      "loss": 0.9576,
+      "step": 2726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957336232718937e-05,
+      "loss": 1.0029,
+      "step": 2727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953087348501004e-05,
+      "loss": 1.0073,
+      "step": 2728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948837618286335e-05,
+      "loss": 0.9675,
+      "step": 2729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944587042984413e-05,
+      "loss": 0.9859,
+      "step": 2730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940335623504896e-05,
+      "loss": 0.3817,
+      "step": 2731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936083360757628e-05,
+      "loss": 1.0703,
+      "step": 2732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931830255652634e-05,
+      "loss": 1.0357,
+      "step": 2733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927576309100113e-05,
+      "loss": 0.9783,
+      "step": 2734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923321522010454e-05,
+      "loss": 0.98,
+      "step": 2735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919065895294214e-05,
+      "loss": 0.9528,
+      "step": 2736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914809429862144e-05,
+      "loss": 0.9897,
+      "step": 2737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910552126625158e-05,
+      "loss": 1.0011,
+      "step": 2738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390629398649436e-05,
+      "loss": 0.9723,
+      "step": 2739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902035010381037e-05,
+      "loss": 0.9731,
+      "step": 2740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389777519919664e-05,
+      "loss": 0.3819,
+      "step": 2741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893514553852811e-05,
+      "loss": 0.9543,
+      "step": 2742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3889253075261368e-05,
+      "loss": 1.0057,
+      "step": 2743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884990764334308e-05,
+      "loss": 0.9552,
+      "step": 2744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880727621983797e-05,
+      "loss": 1.0174,
+      "step": 2745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876463649122195e-05,
+      "loss": 0.9974,
+      "step": 2746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872198846662027e-05,
+      "loss": 0.9025,
+      "step": 2747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3867933215516e-05,
+      "loss": 1.0028,
+      "step": 2748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863666756596996e-05,
+      "loss": 1.0392,
+      "step": 2749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3859399470818082e-05,
+      "loss": 0.9427,
+      "step": 2750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855131359092488e-05,
+      "loss": 1.0146,
+      "step": 2751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850862422333635e-05,
+      "loss": 0.9997,
+      "step": 2752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846592661455112e-05,
+      "loss": 0.9541,
+      "step": 2753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3842322077370686e-05,
+      "loss": 1.0092,
+      "step": 2754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838050670994304e-05,
+      "loss": 0.9914,
+      "step": 2755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833778443240082e-05,
+      "loss": 0.9631,
+      "step": 2756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382950539502232e-05,
+      "loss": 0.9606,
+      "step": 2757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3825231527255485e-05,
+      "loss": 0.9875,
+      "step": 2758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820956840854225e-05,
+      "loss": 0.9809,
+      "step": 2759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816681336733365e-05,
+      "loss": 0.9691,
+      "step": 2760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812405015807895e-05,
+      "loss": 0.9697,
+      "step": 2761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808127878992994e-05,
+      "loss": 0.9675,
+      "step": 2762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3803849927204003e-05,
+      "loss": 1.0184,
+      "step": 2763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799571161356449e-05,
+      "loss": 1.0066,
+      "step": 2764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795291582366019e-05,
+      "loss": 0.9069,
+      "step": 2765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3791011191148586e-05,
+      "loss": 1.0062,
+      "step": 2766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786729988620194e-05,
+      "loss": 0.9056,
+      "step": 2767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3782447975697058e-05,
+      "loss": 0.9938,
+      "step": 2768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778165153295566e-05,
+      "loss": 0.9794,
+      "step": 2769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773881522332283e-05,
+      "loss": 1.0234,
+      "step": 2770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769597083723945e-05,
+      "loss": 0.9498,
+      "step": 2771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765311838387463e-05,
+      "loss": 1.0219,
+      "step": 2772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761025787239912e-05,
+      "loss": 1.0446,
+      "step": 2773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756738931198553e-05,
+      "loss": 0.9894,
+      "step": 2774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752451271180808e-05,
+      "loss": 0.9872,
+      "step": 2775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374816280810428e-05,
+      "loss": 0.9385,
+      "step": 2776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3743873542886735e-05,
+      "loss": 0.9853,
+      "step": 2777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739583476446114e-05,
+      "loss": 1.0308,
+      "step": 2778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3735292609700535e-05,
+      "loss": 0.9792,
+      "step": 2779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3731000943568279e-05,
+      "loss": 0.9861,
+      "step": 2780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726708478967801e-05,
+      "loss": 1.0445,
+      "step": 2781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722415216817728e-05,
+      "loss": 0.9981,
+      "step": 2782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3718121158036862e-05,
+      "loss": 0.9654,
+      "step": 2783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3713826303544167e-05,
+      "loss": 0.9691,
+      "step": 2784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370953065425878e-05,
+      "loss": 0.9217,
+      "step": 2785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3705234211100011e-05,
+      "loss": 0.9311,
+      "step": 2786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700936974987336e-05,
+      "loss": 1.0561,
+      "step": 2787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696638946840404e-05,
+      "loss": 1.0204,
+      "step": 2788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692340127579034e-05,
+      "loss": 0.9729,
+      "step": 2789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688040518123208e-05,
+      "loss": 0.9365,
+      "step": 2790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683740119393087e-05,
+      "loss": 1.0388,
+      "step": 2791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679438932308993e-05,
+      "loss": 1.0049,
+      "step": 2792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675136957791417e-05,
+      "loss": 0.9848,
+      "step": 2793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670834196761025e-05,
+      "loss": 0.9131,
+      "step": 2794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666530650138644e-05,
+      "loss": 0.9045,
+      "step": 2795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366222631884527e-05,
+      "loss": 0.9846,
+      "step": 2796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657921203802076e-05,
+      "loss": 0.9611,
+      "step": 2797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3653615305930393e-05,
+      "loss": 1.0007,
+      "step": 2798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364930862615172e-05,
+      "loss": 1.0165,
+      "step": 2799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364500116538773e-05,
+      "loss": 0.9896,
+      "step": 2800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3640692924560253e-05,
+      "loss": 0.986,
+      "step": 2801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3636383904591295e-05,
+      "loss": 0.9247,
+      "step": 2802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632074106403025e-05,
+      "loss": 0.3665,
+      "step": 2803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627763530917784e-05,
+      "loss": 1.0087,
+      "step": 2804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623452179058067e-05,
+      "loss": 0.9617,
+      "step": 2805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3619140051746546e-05,
+      "loss": 0.9936,
+      "step": 2806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614827149906056e-05,
+      "loss": 0.9611,
+      "step": 2807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610513474459598e-05,
+      "loss": 1.0037,
+      "step": 2808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606199026330335e-05,
+      "loss": 0.9291,
+      "step": 2809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36018838064416e-05,
+      "loss": 1.0008,
+      "step": 2810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597567815716892e-05,
+      "loss": 0.9537,
+      "step": 2811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593251055079866e-05,
+      "loss": 0.9698,
+      "step": 2812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588933525454353e-05,
+      "loss": 1.0726,
+      "step": 2813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3584615227764342e-05,
+      "loss": 0.9905,
+      "step": 2814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580296162933987e-05,
+      "loss": 0.9603,
+      "step": 2815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357597633188761e-05,
+      "loss": 0.9822,
+      "step": 2816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571655735549694e-05,
+      "loss": 1.0171,
+      "step": 2817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567334374844884e-05,
+      "loss": 0.34,
+      "step": 2818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563012250697993e-05,
+      "loss": 1.018,
+      "step": 2819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558689364033994e-05,
+      "loss": 1.0103,
+      "step": 2820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554365715778022e-05,
+      "loss": 0.9714,
+      "step": 2821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3550041306855377e-05,
+      "loss": 0.947,
+      "step": 2822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545716138191526e-05,
+      "loss": 0.3842,
+      "step": 2823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541390210712093e-05,
+      "loss": 0.9347,
+      "step": 2824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537063525342865e-05,
+      "loss": 0.9991,
+      "step": 2825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532736083009795e-05,
+      "loss": 1.0074,
+      "step": 2826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3528407884638987e-05,
+      "loss": 0.9613,
+      "step": 2827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524078931156724e-05,
+      "loss": 0.9364,
+      "step": 2828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519749223489436e-05,
+      "loss": 1.0121,
+      "step": 2829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515418762563724e-05,
+      "loss": 0.965,
+      "step": 2830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511087549306341e-05,
+      "loss": 0.975,
+      "step": 2831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506755584644212e-05,
+      "loss": 0.9651,
+      "step": 2832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502422869504413e-05,
+      "loss": 0.9037,
+      "step": 2833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498089404814185e-05,
+      "loss": 1.0214,
+      "step": 2834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349375519150093e-05,
+      "loss": 0.9722,
+      "step": 2835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3489420230492206e-05,
+      "loss": 0.9473,
+      "step": 2836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485084522715741e-05,
+      "loss": 1.0075,
+      "step": 2837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480748069099412e-05,
+      "loss": 1.0314,
+      "step": 2838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476410870571259e-05,
+      "loss": 1.0034,
+      "step": 2839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3472072928059481e-05,
+      "loss": 1.013,
+      "step": 2840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467734242492439e-05,
+      "loss": 0.945,
+      "step": 2841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463394814798648e-05,
+      "loss": 0.989,
+      "step": 2842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459054645906795e-05,
+      "loss": 0.9734,
+      "step": 2843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454713736745704e-05,
+      "loss": 0.9874,
+      "step": 2844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450372088244374e-05,
+      "loss": 0.9377,
+      "step": 2845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446029701331961e-05,
+      "loss": 0.9549,
+      "step": 2846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3441686576937771e-05,
+      "loss": 0.969,
+      "step": 2847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437342715991271e-05,
+      "loss": 0.9465,
+      "step": 2848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432998119422089e-05,
+      "loss": 0.9853,
+      "step": 2849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3428652788160008e-05,
+      "loss": 0.9278,
+      "step": 2850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3424306723134967e-05,
+      "loss": 0.9325,
+      "step": 2851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419959925277068e-05,
+      "loss": 0.9842,
+      "step": 2852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415612395516565e-05,
+      "loss": 1.0111,
+      "step": 2853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411264134783863e-05,
+      "loss": 0.9165,
+      "step": 2854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406915144009531e-05,
+      "loss": 1.0302,
+      "step": 2855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3402565424124298e-05,
+      "loss": 0.9953,
+      "step": 2856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398214976059038e-05,
+      "loss": 0.9392,
+      "step": 2857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393863800744792e-05,
+      "loss": 0.9553,
+      "step": 2858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389511899112746e-05,
+      "loss": 0.9061,
+      "step": 2859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338515927209425e-05,
+      "loss": 0.4031,
+      "step": 2860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33808059206208e-05,
+      "loss": 0.9385,
+      "step": 2861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376451845624063e-05,
+      "loss": 1.0095,
+      "step": 2862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3372097048035844e-05,
+      "loss": 0.9582,
+      "step": 2863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336774152878811e-05,
+      "loss": 0.9713,
+      "step": 2864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363385288812982e-05,
+      "loss": 0.9043,
+      "step": 2865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359028329042739e-05,
+      "loss": 0.9635,
+      "step": 2866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354670650409802e-05,
+      "loss": 0.9472,
+      "step": 2867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350312253846763e-05,
+      "loss": 0.9593,
+      "step": 2868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334595314028635e-05,
+      "loss": 0.9571,
+      "step": 2869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341593310661458e-05,
+      "loss": 1.0023,
+      "step": 2870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3337232765905132e-05,
+      "loss": 0.9908,
+      "step": 2871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332871506950564e-05,
+      "loss": 1.011,
+      "step": 2872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332850953473111e-05,
+      "loss": 1.006,
+      "step": 2873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332414685018026e-05,
+      "loss": 0.9315,
+      "step": 2874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319783454231678e-05,
+      "loss": 0.9414,
+      "step": 2875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315419347819167e-05,
+      "loss": 0.9738,
+      "step": 2876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311054531876688e-05,
+      "loss": 0.9555,
+      "step": 2877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306689007338347e-05,
+      "loss": 0.974,
+      "step": 2878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330232277513841e-05,
+      "loss": 0.9241,
+      "step": 2879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297955836211292e-05,
+      "loss": 1.024,
+      "step": 2880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293588191491554e-05,
+      "loss": 0.996,
+      "step": 2881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289219841913909e-05,
+      "loss": 1.0206,
+      "step": 2882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3284850788413231e-05,
+      "loss": 0.9457,
+      "step": 2883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280481031924532e-05,
+      "loss": 0.9401,
+      "step": 2884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327611057338298e-05,
+      "loss": 0.9764,
+      "step": 2885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271739413723899e-05,
+      "loss": 0.9913,
+      "step": 2886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267367553882751e-05,
+      "loss": 0.355,
+      "step": 2887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262994994795153e-05,
+      "loss": 0.9742,
+      "step": 2888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258621737396878e-05,
+      "loss": 0.9803,
+      "step": 2889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254247782623835e-05,
+      "loss": 0.9595,
+      "step": 2890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249873131412096e-05,
+      "loss": 0.925,
+      "step": 2891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3245497784697879e-05,
+      "loss": 1.0096,
+      "step": 2892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241121743417545e-05,
+      "loss": 0.9987,
+      "step": 2893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32367450085076e-05,
+      "loss": 0.969,
+      "step": 2894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232367580904714e-05,
+      "loss": 0.3753,
+      "step": 2895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227989461545696e-05,
+      "loss": 1.0072,
+      "step": 2896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223610651367498e-05,
+      "loss": 0.9648,
+      "step": 2897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321923115130723e-05,
+      "loss": 1.0068,
+      "step": 2898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214850962302144e-05,
+      "loss": 1.0521,
+      "step": 2899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321047008528964e-05,
+      "loss": 1.0028,
+      "step": 2900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3206088521207265e-05,
+      "loss": 0.9756,
+      "step": 2901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3201706270992714e-05,
+      "loss": 0.9702,
+      "step": 2902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197323335583826e-05,
+      "loss": 0.9334,
+      "step": 2903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3192939715918595e-05,
+      "loss": 0.9383,
+      "step": 2904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3188555412935151e-05,
+      "loss": 0.9755,
+      "step": 2905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3184170427571774e-05,
+      "loss": 0.3781,
+      "step": 2906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3179784760766899e-05,
+      "loss": 0.9684,
+      "step": 2907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175398413459086e-05,
+      "loss": 0.9302,
+      "step": 2908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3171011386587062e-05,
+      "loss": 0.9252,
+      "step": 2909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3166623681089689e-05,
+      "loss": 0.9758,
+      "step": 2910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162235297905977e-05,
+      "loss": 0.9634,
+      "step": 2911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157846237975076e-05,
+      "loss": 0.9992,
+      "step": 2912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315345650223629e-05,
+      "loss": 0.3523,
+      "step": 2913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314906609162906e-05,
+      "loss": 0.9389,
+      "step": 2914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144675007092972e-05,
+      "loss": 1.0401,
+      "step": 2915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140283249567762e-05,
+      "loss": 1.0166,
+      "step": 2916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135890819993303e-05,
+      "loss": 1.0339,
+      "step": 2917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131497719309616e-05,
+      "loss": 0.9579,
+      "step": 2918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312710394845686e-05,
+      "loss": 0.9055,
+      "step": 2919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3122709508375355e-05,
+      "loss": 0.9478,
+      "step": 2920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311831440000554e-05,
+      "loss": 0.9808,
+      "step": 2921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311391862428801e-05,
+      "loss": 0.9758,
+      "step": 2922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109522182163503e-05,
+      "loss": 1.0067,
+      "step": 2923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105125074572897e-05,
+      "loss": 0.9663,
+      "step": 2924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100727302457212e-05,
+      "loss": 1.0098,
+      "step": 2925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096328866757616e-05,
+      "loss": 0.9345,
+      "step": 2926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309192976841541e-05,
+      "loss": 0.9816,
+      "step": 2927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308753000837204e-05,
+      "loss": 1.0091,
+      "step": 2928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30831295875691e-05,
+      "loss": 0.9093,
+      "step": 2929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078728506948313e-05,
+      "loss": 0.985,
+      "step": 2930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074326767451555e-05,
+      "loss": 0.9594,
+      "step": 2931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306992437002084e-05,
+      "loss": 1.03,
+      "step": 2932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065521315598314e-05,
+      "loss": 0.9624,
+      "step": 2933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061117605126283e-05,
+      "loss": 1.0136,
+      "step": 2934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056713239547172e-05,
+      "loss": 0.9779,
+      "step": 2935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052308219803553e-05,
+      "loss": 0.9363,
+      "step": 2936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047902546838146e-05,
+      "loss": 0.3845,
+      "step": 2937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043496221593808e-05,
+      "loss": 0.9973,
+      "step": 2938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3039089245013523e-05,
+      "loss": 1.0136,
+      "step": 2939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034681618040433e-05,
+      "loss": 1.006,
+      "step": 2940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030273341617809e-05,
+      "loss": 0.9554,
+      "step": 2941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025864416689058e-05,
+      "loss": 0.9842,
+      "step": 2942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021454844197731e-05,
+      "loss": 0.9572,
+      "step": 2943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017044625087521e-05,
+      "loss": 0.9724,
+      "step": 2944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012633760302252e-05,
+      "loss": 0.3873,
+      "step": 2945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008222250785888e-05,
+      "loss": 0.9766,
+      "step": 2946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003810097482536e-05,
+      "loss": 0.9822,
+      "step": 2947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999397301336435e-05,
+      "loss": 0.9504,
+      "step": 2948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2994983863291962e-05,
+      "loss": 0.9211,
+      "step": 2949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990569784293636e-05,
+      "loss": 0.9435,
+      "step": 2950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2986155065286108e-05,
+      "loss": 0.9357,
+      "step": 2951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981739707214169e-05,
+      "loss": 0.9407,
+      "step": 2952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977323711022747e-05,
+      "loss": 0.9841,
+      "step": 2953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972907077656904e-05,
+      "loss": 0.9394,
+      "step": 2954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296848980806184e-05,
+      "loss": 0.9922,
+      "step": 2955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964071903182891e-05,
+      "loss": 1.0,
+      "step": 2956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959653363965528e-05,
+      "loss": 0.9664,
+      "step": 2957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2955234191355358e-05,
+      "loss": 0.9657,
+      "step": 2958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295081438629813e-05,
+      "loss": 0.9596,
+      "step": 2959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946393949739715e-05,
+      "loss": 0.958,
+      "step": 2960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941972882626133e-05,
+      "loss": 0.9463,
+      "step": 2961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2937551185903527e-05,
+      "loss": 0.9865,
+      "step": 2962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933128860518184e-05,
+      "loss": 0.9241,
+      "step": 2963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928705907416524e-05,
+      "loss": 0.9445,
+      "step": 2964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924282327545095e-05,
+      "loss": 0.9784,
+      "step": 2965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919858121850583e-05,
+      "loss": 0.9803,
+      "step": 2966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915433291279811e-05,
+      "loss": 0.9367,
+      "step": 2967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911007836779733e-05,
+      "loss": 0.9447,
+      "step": 2968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906581759297438e-05,
+      "loss": 0.9902,
+      "step": 2969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2902155059780142e-05,
+      "loss": 0.9831,
+      "step": 2970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897727739175203e-05,
+      "loss": 0.3553,
+      "step": 2971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2893299798430108e-05,
+      "loss": 0.9924,
+      "step": 2972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888871238492472e-05,
+      "loss": 0.9542,
+      "step": 2973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2884442060310053e-05,
+      "loss": 1.001,
+      "step": 2974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2880012264830738e-05,
+      "loss": 0.9958,
+      "step": 2975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2875581853002534e-05,
+      "loss": 0.9265,
+      "step": 2976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871150825773596e-05,
+      "loss": 0.9664,
+      "step": 2977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2866719184092198e-05,
+      "loss": 0.9931,
+      "step": 2978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286228692890676e-05,
+      "loss": 0.9953,
+      "step": 2979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2857854061165821e-05,
+      "loss": 0.9413,
+      "step": 2980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285342058181806e-05,
+      "loss": 1.0079,
+      "step": 2981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2848986491812274e-05,
+      "loss": 1.0377,
+      "step": 2982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844551792097402e-05,
+      "loss": 1.0332,
+      "step": 2983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284011648362251e-05,
+      "loss": 0.9274,
+      "step": 2984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2835680567336797e-05,
+      "loss": 0.962,
+      "step": 2985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831244044189586e-05,
+      "loss": 0.9683,
+      "step": 2986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2826806915130334e-05,
+      "loss": 0.9789,
+      "step": 2987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822369181108632e-05,
+      "loss": 0.9175,
+      "step": 2988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817930843074188e-05,
+      "loss": 0.9605,
+      "step": 2989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813491901976851e-05,
+      "loss": 0.999,
+      "step": 2990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2809052358766595e-05,
+      "loss": 0.9271,
+      "step": 2991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804612214393521e-05,
+      "loss": 0.9205,
+      "step": 2992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280017146980786e-05,
+      "loss": 0.993,
+      "step": 2993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795730125959973e-05,
+      "loss": 0.9524,
+      "step": 2994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791288183800352e-05,
+      "loss": 0.9971,
+      "step": 2995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786845644279607e-05,
+      "loss": 0.9628,
+      "step": 2996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782402508348484e-05,
+      "loss": 0.8959,
+      "step": 2997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777958776957854e-05,
+      "loss": 0.3682,
+      "step": 2998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277351445105872e-05,
+      "loss": 0.946,
+      "step": 2999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27690695316022e-05,
+      "loss": 0.9436,
+      "step": 3000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764624019539558e-05,
+      "loss": 0.9343,
+      "step": 3001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2760177915822166e-05,
+      "loss": 0.9348,
+      "step": 3002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755731221401534e-05,
+      "loss": 0.9559,
+      "step": 3003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751283937229296e-05,
+      "loss": 0.9543,
+      "step": 3004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746836064257207e-05,
+      "loss": 1.0147,
+      "step": 3005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2742387603437154e-05,
+      "loss": 0.9759,
+      "step": 3006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273793855572115e-05,
+      "loss": 1.0338,
+      "step": 3007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733488922061335e-05,
+      "loss": 0.9086,
+      "step": 3008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2729038703409963e-05,
+      "loss": 0.9873,
+      "step": 3009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2724587900719428e-05,
+      "loss": 0.9306,
+      "step": 3010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720136514942239e-05,
+      "loss": 1.0358,
+      "step": 3011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715684547031035e-05,
+      "loss": 0.989,
+      "step": 3012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711231997938576e-05,
+      "loss": 0.9338,
+      "step": 3013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2706778868617749e-05,
+      "loss": 1.0374,
+      "step": 3014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702325160021566e-05,
+      "loss": 1.0034,
+      "step": 3015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269787087310316e-05,
+      "loss": 1.0098,
+      "step": 3016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2693416008815787e-05,
+      "loss": 1.0293,
+      "step": 3017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688960568112827e-05,
+      "loss": 0.8997,
+      "step": 3018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684504551947791e-05,
+      "loss": 0.9831,
+      "step": 3019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680047961274306e-05,
+      "loss": 0.9381,
+      "step": 3020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675590797046118e-05,
+      "loss": 1.0438,
+      "step": 3021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671133060217109e-05,
+      "loss": 0.9147,
+      "step": 3022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666674751741265e-05,
+      "loss": 0.9815,
+      "step": 3023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662215872572712e-05,
+      "loss": 0.9821,
+      "step": 3024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657756423665688e-05,
+      "loss": 0.9531,
+      "step": 3025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653296405974555e-05,
+      "loss": 0.96,
+      "step": 3026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648835820453803e-05,
+      "loss": 0.3622,
+      "step": 3027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264437466805803e-05,
+      "loss": 0.99,
+      "step": 3028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263991294974197e-05,
+      "loss": 0.9858,
+      "step": 3029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2635450666460469e-05,
+      "loss": 0.9232,
+      "step": 3030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630987819168495e-05,
+      "loss": 0.9982,
+      "step": 3031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2626524408821137e-05,
+      "loss": 0.9652,
+      "step": 3032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2622060436373608e-05,
+      "loss": 0.9438,
+      "step": 3033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261759590278124e-05,
+      "loss": 0.3816,
+      "step": 3034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261313080899948e-05,
+      "loss": 1.0206,
+      "step": 3035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2608665155983903e-05,
+      "loss": 0.9737,
+      "step": 3036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604198944690196e-05,
+      "loss": 0.9836,
+      "step": 3037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259973217607417e-05,
+      "loss": 0.972,
+      "step": 3038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595264851091757e-05,
+      "loss": 1.0,
+      "step": 3039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590796970698999e-05,
+      "loss": 0.9727,
+      "step": 3040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258632853585207e-05,
+      "loss": 1.0022,
+      "step": 3041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581859547507254e-05,
+      "loss": 0.9009,
+      "step": 3042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257739000662095e-05,
+      "loss": 0.9774,
+      "step": 3043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2572919914149688e-05,
+      "loss": 0.9675,
+      "step": 3044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568449271050103e-05,
+      "loss": 1.0254,
+      "step": 3045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2563978078278955e-05,
+      "loss": 0.9602,
+      "step": 3046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559506336793121e-05,
+      "loss": 0.9791,
+      "step": 3047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555034047549594e-05,
+      "loss": 1.0219,
+      "step": 3048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550561211505485e-05,
+      "loss": 0.9554,
+      "step": 3049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546087829618024e-05,
+      "loss": 0.9541,
+      "step": 3050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541613902844547e-05,
+      "loss": 0.9914,
+      "step": 3051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537139432142524e-05,
+      "loss": 0.9686,
+      "step": 3052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2532664418469529e-05,
+      "loss": 0.9775,
+      "step": 3053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528188862783254e-05,
+      "loss": 0.9673,
+      "step": 3054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523712766041513e-05,
+      "loss": 0.9738,
+      "step": 3055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2519236129202232e-05,
+      "loss": 0.9709,
+      "step": 3056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251475895322345e-05,
+      "loss": 1.03,
+      "step": 3057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510281239063317e-05,
+      "loss": 0.9194,
+      "step": 3058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505802987680114e-05,
+      "loss": 1.0009,
+      "step": 3059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501324200032223e-05,
+      "loss": 0.973,
+      "step": 3060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2496844877078149e-05,
+      "loss": 1.0552,
+      "step": 3061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492365019776502e-05,
+      "loss": 0.9699,
+      "step": 3062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248788462908602e-05,
+      "loss": 1.031,
+      "step": 3063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483403705965542e-05,
+      "loss": 1.0009,
+      "step": 3064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2478922251374026e-05,
+      "loss": 0.9874,
+      "step": 3065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474440266270546e-05,
+      "loss": 0.9637,
+      "step": 3066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246995775161429e-05,
+      "loss": 0.9939,
+      "step": 3067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465474708364553e-05,
+      "loss": 0.9581,
+      "step": 3068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246099113748075e-05,
+      "loss": 1.0019,
+      "step": 3069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456507039922404e-05,
+      "loss": 0.9549,
+      "step": 3070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452022416649153e-05,
+      "loss": 0.9601,
+      "step": 3071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447537268620748e-05,
+      "loss": 0.9825,
+      "step": 3072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443051596797052e-05,
+      "loss": 1.0447,
+      "step": 3073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438565402138039e-05,
+      "loss": 0.9961,
+      "step": 3074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434078685603794e-05,
+      "loss": 0.9356,
+      "step": 3075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429591448154518e-05,
+      "loss": 0.9785,
+      "step": 3076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425103690750522e-05,
+      "loss": 1.0103,
+      "step": 3077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420615414352221e-05,
+      "loss": 0.9388,
+      "step": 3078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416126619920155e-05,
+      "loss": 0.9657,
+      "step": 3079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411637308414958e-05,
+      "loss": 0.9883,
+      "step": 3080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240714748079739e-05,
+      "loss": 0.9924,
+      "step": 3081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402657138028313e-05,
+      "loss": 0.9767,
+      "step": 3082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398166281068701e-05,
+      "loss": 0.9452,
+      "step": 3083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393674910879643e-05,
+      "loss": 0.9755,
+      "step": 3084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389183028422324e-05,
+      "loss": 1.0235,
+      "step": 3085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384690634658055e-05,
+      "loss": 0.3898,
+      "step": 3086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380197730548247e-05,
+      "loss": 0.9733,
+      "step": 3087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2375704317054422e-05,
+      "loss": 0.9523,
+      "step": 3088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371210395138212e-05,
+      "loss": 0.9417,
+      "step": 3089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366715965761358e-05,
+      "loss": 0.9572,
+      "step": 3090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362221029885708e-05,
+      "loss": 0.9615,
+      "step": 3091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235772558847322e-05,
+      "loss": 0.3643,
+      "step": 3092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353229642485958e-05,
+      "loss": 0.9353,
+      "step": 3093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348733192886095e-05,
+      "loss": 0.961,
+      "step": 3094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344236240635912e-05,
+      "loss": 1.0009,
+      "step": 3095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339738786697801e-05,
+      "loss": 1.0039,
+      "step": 3096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335240832034258e-05,
+      "loss": 0.9488,
+      "step": 3097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2330742377607884e-05,
+      "loss": 0.9608,
+      "step": 3098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326243424381392e-05,
+      "loss": 0.9933,
+      "step": 3099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321743973317594e-05,
+      "loss": 0.9455,
+      "step": 3100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2317244025379418e-05,
+      "loss": 0.9663,
+      "step": 3101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312743581529891e-05,
+      "loss": 0.95,
+      "step": 3102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2308242642732151e-05,
+      "loss": 0.9438,
+      "step": 3103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2303741209949443e-05,
+      "loss": 1.0009,
+      "step": 3104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299239284145106e-05,
+      "loss": 0.9371,
+      "step": 3105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2294736866282599e-05,
+      "loss": 0.9182,
+      "step": 3106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2290233957325482e-05,
+      "loss": 0.3864,
+      "step": 3107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2285730558237415e-05,
+      "loss": 1.0374,
+      "step": 3108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2281226669982166e-05,
+      "loss": 1.0407,
+      "step": 3109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276722293523614e-05,
+      "loss": 0.9258,
+      "step": 3110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227221742982573e-05,
+      "loss": 1.0055,
+      "step": 3111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2267712079852599e-05,
+      "loss": 1.0026,
+      "step": 3112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263206244568406e-05,
+      "loss": 0.928,
+      "step": 3113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258699924937441e-05,
+      "loss": 0.9689,
+      "step": 3114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254193121924099e-05,
+      "loss": 0.9701,
+      "step": 3115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2249685836492872e-05,
+      "loss": 0.9909,
+      "step": 3116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245178069608367e-05,
+      "loss": 0.9596,
+      "step": 3117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2240669822235285e-05,
+      "loss": 0.9596,
+      "step": 3118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223616109533843e-05,
+      "loss": 1.0246,
+      "step": 3119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231651889882712e-05,
+      "loss": 0.9595,
+      "step": 3120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222714220683314e-05,
+      "loss": 0.9021,
+      "step": 3121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222263204715483e-05,
+      "loss": 0.9743,
+      "step": 3122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218121411812997e-05,
+      "loss": 0.97,
+      "step": 3123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213610301772958e-05,
+      "loss": 0.9834,
+      "step": 3124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220909871800013e-05,
+      "loss": 0.9759,
+      "step": 3125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204586661460034e-05,
+      "loss": 0.9429,
+      "step": 3126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200074133118291e-05,
+      "loss": 0.9352,
+      "step": 3127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195561133940626e-05,
+      "loss": 1.0082,
+      "step": 3128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2191047664892858e-05,
+      "loss": 0.3438,
+      "step": 3129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186533726940912e-05,
+      "loss": 0.9812,
+      "step": 3130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182019321050815e-05,
+      "loss": 0.9078,
+      "step": 3131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177504448188686e-05,
+      "loss": 0.9772,
+      "step": 3132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2172989109320752e-05,
+      "loss": 0.9801,
+      "step": 3133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168473305413335e-05,
+      "loss": 0.9787,
+      "step": 3134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163957037432859e-05,
+      "loss": 0.8722,
+      "step": 3135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159440306345847e-05,
+      "loss": 0.9578,
+      "step": 3136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2154923113118916e-05,
+      "loss": 0.9978,
+      "step": 3137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150405458718797e-05,
+      "loss": 0.9662,
+      "step": 3138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145887344112301e-05,
+      "loss": 1.0423,
+      "step": 3139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141368770266346e-05,
+      "loss": 0.9219,
+      "step": 3140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213684973814795e-05,
+      "loss": 0.918,
+      "step": 3141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132330248724226e-05,
+      "loss": 0.9542,
+      "step": 3142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127810302962389e-05,
+      "loss": 0.9794,
+      "step": 3143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123289901829744e-05,
+      "loss": 0.9917,
+      "step": 3144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118769046293703e-05,
+      "loss": 0.9334,
+      "step": 3145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114247737321767e-05,
+      "loss": 0.9907,
+      "step": 3146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109725975881536e-05,
+      "loss": 0.9107,
+      "step": 3147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210520376294071e-05,
+      "loss": 1.0297,
+      "step": 3148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100681099467085e-05,
+      "loss": 0.9364,
+      "step": 3149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096157986428551e-05,
+      "loss": 0.9077,
+      "step": 3150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091634424793096e-05,
+      "loss": 0.924,
+      "step": 3151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087110415528804e-05,
+      "loss": 0.9472,
+      "step": 3152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208258595960385e-05,
+      "loss": 0.8952,
+      "step": 3153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078061057986514e-05,
+      "loss": 0.9598,
+      "step": 3154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073535711645164e-05,
+      "loss": 0.9912,
+      "step": 3155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069009921548264e-05,
+      "loss": 0.9575,
+      "step": 3156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206448368866438e-05,
+      "loss": 0.9599,
+      "step": 3157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059957013962163e-05,
+      "loss": 0.9318,
+      "step": 3158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205542989841036e-05,
+      "loss": 1.0566,
+      "step": 3159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050902342977817e-05,
+      "loss": 0.9308,
+      "step": 3160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046374348633473e-05,
+      "loss": 1.0043,
+      "step": 3161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2041845916346362e-05,
+      "loss": 0.979,
+      "step": 3162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037317047085608e-05,
+      "loss": 0.9642,
+      "step": 3163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032787741820429e-05,
+      "loss": 0.9432,
+      "step": 3164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2028258001520141e-05,
+      "loss": 0.9447,
+      "step": 3165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023727827154146e-05,
+      "loss": 0.9674,
+      "step": 3166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2019197219691942e-05,
+      "loss": 0.3732,
+      "step": 3167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2014666180103125e-05,
+      "loss": 0.9795,
+      "step": 3168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010134709357374e-05,
+      "loss": 0.9526,
+      "step": 3169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200560280842447e-05,
+      "loss": 1.005,
+      "step": 3170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001070478274274e-05,
+      "loss": 0.9662,
+      "step": 3171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1996537719876757e-05,
+      "loss": 0.9538,
+      "step": 3172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199200453420196e-05,
+      "loss": 0.9375,
+      "step": 3173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198747092222003e-05,
+      "loss": 0.9653,
+      "step": 3174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19829368849012e-05,
+      "loss": 0.9432,
+      "step": 3175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1978402423215798e-05,
+      "loss": 1.0164,
+      "step": 3176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973867538134238e-05,
+      "loss": 0.9334,
+      "step": 3177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969332230627029e-05,
+      "loss": 0.9429,
+      "step": 3178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1964796501664769e-05,
+      "loss": 1.0007,
+      "step": 3179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960260352218138e-05,
+      "loss": 0.906,
+      "step": 3180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195572378325792e-05,
+      "loss": 0.9611,
+      "step": 3181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195118679575498e-05,
+      "loss": 0.3884,
+      "step": 3182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946649390680277e-05,
+      "loss": 0.9739,
+      "step": 3183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942111569004852e-05,
+      "loss": 0.9272,
+      "step": 3184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1937573331699847e-05,
+      "loss": 0.9942,
+      "step": 3185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1933034679736482e-05,
+      "loss": 1.0147,
+      "step": 3186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192849561408607e-05,
+      "loss": 0.8834,
+      "step": 3187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923956135720014e-05,
+      "loss": 0.993,
+      "step": 3188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19194162456098e-05,
+      "loss": 0.8903,
+      "step": 3189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914875944727012e-05,
+      "loss": 0.956,
+      "step": 3190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191033523404331e-05,
+      "loss": 0.9231,
+      "step": 3191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905794114530453e-05,
+      "loss": 0.9943,
+      "step": 3192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901252587160274e-05,
+      "loss": 0.9583,
+      "step": 3193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189671065290471e-05,
+      "loss": 1.0194,
+      "step": 3194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892168312735772e-05,
+      "loss": 0.9604,
+      "step": 3195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188762556762556e-05,
+      "loss": 0.9748,
+      "step": 3196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1883082418546268e-05,
+      "loss": 0.9511,
+      "step": 3197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878538866470169e-05,
+      "loss": 0.956,
+      "step": 3198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1873994912369623e-05,
+      "loss": 0.95,
+      "step": 3199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186945055721708e-05,
+      "loss": 0.9686,
+      "step": 3200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864905801985075e-05,
+      "loss": 0.9679,
+      "step": 3201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860360647646221e-05,
+      "loss": 0.9603,
+      "step": 3202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1855815095173227e-05,
+      "loss": 0.99,
+      "step": 3203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851269145538881e-05,
+      "loss": 0.958,
+      "step": 3204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846722799716059e-05,
+      "loss": 0.9871,
+      "step": 3205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184217605867772e-05,
+      "loss": 0.981,
+      "step": 3206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183762892339691e-05,
+      "loss": 0.9581,
+      "step": 3207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1833081394846753e-05,
+      "loss": 0.945,
+      "step": 3208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1828533474000463e-05,
+      "loss": 0.921,
+      "step": 3209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823985161831338e-05,
+      "loss": 0.9943,
+      "step": 3210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181943645931276e-05,
+      "loss": 0.9633,
+      "step": 3211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1814887367418187e-05,
+      "loss": 0.9872,
+      "step": 3212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181033788712117e-05,
+      "loss": 0.9283,
+      "step": 3213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180578801939534e-05,
+      "loss": 0.9884,
+      "step": 3214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1801237765214409e-05,
+      "loss": 0.9837,
+      "step": 3215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1796687125552173e-05,
+      "loss": 0.9902,
+      "step": 3216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179213610138251e-05,
+      "loss": 0.9088,
+      "step": 3217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1787584693679382e-05,
+      "loss": 0.9247,
+      "step": 3218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783032903416832e-05,
+      "loss": 0.9635,
+      "step": 3219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177848073156898e-05,
+      "loss": 1.0069,
+      "step": 3220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773928179110037e-05,
+      "loss": 0.3892,
+      "step": 3221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176937524701429e-05,
+      "loss": 0.9675,
+      "step": 3222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1764821936256105e-05,
+      "loss": 0.9644,
+      "step": 3223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760268247809937e-05,
+      "loss": 1.0003,
+      "step": 3224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755714182650314e-05,
+      "loss": 1.0084,
+      "step": 3225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1751159741751847e-05,
+      "loss": 1.0169,
+      "step": 3226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746604926089229e-05,
+      "loss": 0.9566,
+      "step": 3227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1742049736637232e-05,
+      "loss": 1.0101,
+      "step": 3228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737494174370707e-05,
+      "loss": 0.9522,
+      "step": 3229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732938240264587e-05,
+      "loss": 1.0179,
+      "step": 3230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1728381935293883e-05,
+      "loss": 0.9403,
+      "step": 3231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1723825260433683e-05,
+      "loss": 0.9856,
+      "step": 3232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719268216659165e-05,
+      "loss": 0.9587,
+      "step": 3233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714710804945568e-05,
+      "loss": 0.9784,
+      "step": 3234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710153026268228e-05,
+      "loss": 0.9573,
+      "step": 3235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705594881602544e-05,
+      "loss": 0.972,
+      "step": 3236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701036371924009e-05,
+      "loss": 0.9271,
+      "step": 3237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696477498208177e-05,
+      "loss": 1.0521,
+      "step": 3238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691918261430693e-05,
+      "loss": 0.9846,
+      "step": 3239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1687358662567272e-05,
+      "loss": 0.9601,
+      "step": 3240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682798702593717e-05,
+      "loss": 0.9136,
+      "step": 3241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1678238382485891e-05,
+      "loss": 0.4059,
+      "step": 3242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673677703219752e-05,
+      "loss": 0.9303,
+      "step": 3243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669116665771326e-05,
+      "loss": 0.9172,
+      "step": 3244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664555271116714e-05,
+      "loss": 0.9776,
+      "step": 3245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659993520232098e-05,
+      "loss": 0.9598,
+      "step": 3246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655431414093732e-05,
+      "loss": 0.9385,
+      "step": 3247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1650868953677947e-05,
+      "loss": 0.934,
+      "step": 3248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646306139961156e-05,
+      "loss": 0.9501,
+      "step": 3249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164174297391984e-05,
+      "loss": 0.9089,
+      "step": 3250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1637179456530559e-05,
+      "loss": 0.9152,
+      "step": 3251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632615588769946e-05,
+      "loss": 0.9381,
+      "step": 3252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628051371614715e-05,
+      "loss": 0.9737,
+      "step": 3253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623486806041641e-05,
+      "loss": 0.9436,
+      "step": 3254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161892189302759e-05,
+      "loss": 0.9694,
+      "step": 3255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1614356633549492e-05,
+      "loss": 1.0127,
+      "step": 3256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609791028584356e-05,
+      "loss": 0.9628,
+      "step": 3257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160522507910926e-05,
+      "loss": 0.944,
+      "step": 3258
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1600658786101367e-05,
+      "loss": 0.8843,
+      "step": 3259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596092150537894e-05,
+      "loss": 0.9352,
+      "step": 3260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159152517339615e-05,
+      "loss": 0.955,
+      "step": 3261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586957855653507e-05,
+      "loss": 0.9429,
+      "step": 3262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582390198287415e-05,
+      "loss": 0.9353,
+      "step": 3263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577822202275391e-05,
+      "loss": 0.9326,
+      "step": 3264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573253868595028e-05,
+      "loss": 1.0275,
+      "step": 3265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568685198223993e-05,
+      "loss": 0.9258,
+      "step": 3266
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1564116192140024e-05,
+      "loss": 0.9491,
+      "step": 3267
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155954685132093e-05,
+      "loss": 0.9281,
+      "step": 3268
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554977176744589e-05,
+      "loss": 0.9153,
+      "step": 3269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155040716938895e-05,
+      "loss": 0.9815,
+      "step": 3270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545836830232044e-05,
+      "loss": 0.9146,
+      "step": 3271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541266160251964e-05,
+      "loss": 0.9955,
+      "step": 3272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1536695160426872e-05,
+      "loss": 0.9971,
+      "step": 3273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532123831735002e-05,
+      "loss": 0.9242,
+      "step": 3274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527552175154663e-05,
+      "loss": 0.9646,
+      "step": 3275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152298019166423e-05,
+      "loss": 0.9852,
+      "step": 3276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151840788224215e-05,
+      "loss": 0.9544,
+      "step": 3277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151383524786694e-05,
+      "loss": 0.9133,
+      "step": 3278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1509262289517184e-05,
+      "loss": 0.8843,
+      "step": 3279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504689008171539e-05,
+      "loss": 0.9935,
+      "step": 3280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1500115404808726e-05,
+      "loss": 0.9672,
+      "step": 3281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1495541480407536e-05,
+      "loss": 0.9752,
+      "step": 3282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490967235946833e-05,
+      "loss": 1.0125,
+      "step": 3283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486392672405549e-05,
+      "loss": 0.9382,
+      "step": 3284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1481817790762679e-05,
+      "loss": 0.9104,
+      "step": 3285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477242591997292e-05,
+      "loss": 0.9664,
+      "step": 3286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1472667077088522e-05,
+      "loss": 0.3398,
+      "step": 3287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146809124701557e-05,
+      "loss": 0.9103,
+      "step": 3288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1463515102757703e-05,
+      "loss": 0.9474,
+      "step": 3289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458938645294262e-05,
+      "loss": 0.9329,
+      "step": 3290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454361875604648e-05,
+      "loss": 0.96,
+      "step": 3291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449784794668333e-05,
+      "loss": 0.9695,
+      "step": 3292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445207403464852e-05,
+      "loss": 0.9551,
+      "step": 3293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1440629702973808e-05,
+      "loss": 0.9874,
+      "step": 3294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436051694174878e-05,
+      "loss": 0.9871,
+      "step": 3295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431473378047786e-05,
+      "loss": 0.9389,
+      "step": 3296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142689475557234e-05,
+      "loss": 0.9319,
+      "step": 3297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422315827728407e-05,
+      "loss": 0.9735,
+      "step": 3298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417736595495918e-05,
+      "loss": 0.8981,
+      "step": 3299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413157059854874e-05,
+      "loss": 1.0465,
+      "step": 3300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408577221785331e-05,
+      "loss": 0.9552,
+      "step": 3301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1403997082267419e-05,
+      "loss": 0.9864,
+      "step": 3302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399416642281333e-05,
+      "loss": 0.9544,
+      "step": 3303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394835902807325e-05,
+      "loss": 1.0064,
+      "step": 3304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1390254864825714e-05,
+      "loss": 0.9703,
+      "step": 3305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1385673529316887e-05,
+      "loss": 0.9999,
+      "step": 3306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381091897261292e-05,
+      "loss": 0.9705,
+      "step": 3307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376509969639438e-05,
+      "loss": 0.9666,
+      "step": 3308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371927747431899e-05,
+      "loss": 0.8922,
+      "step": 3309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367345231619313e-05,
+      "loss": 0.928,
+      "step": 3310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136276242318238e-05,
+      "loss": 0.9513,
+      "step": 3311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358179323101863e-05,
+      "loss": 0.9369,
+      "step": 3312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1353595932358588e-05,
+      "loss": 0.9278,
+      "step": 3313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349012251933442e-05,
+      "loss": 0.9559,
+      "step": 3314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1344428282807373e-05,
+      "loss": 0.4111,
+      "step": 3315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1339844025961395e-05,
+      "loss": 0.3802,
+      "step": 3316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335259482376577e-05,
+      "loss": 0.9902,
+      "step": 3317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330674653034058e-05,
+      "loss": 0.361,
+      "step": 3318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1326089538915026e-05,
+      "loss": 1.002,
+      "step": 3319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1321504141000746e-05,
+      "loss": 0.9556,
+      "step": 3320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316918460272527e-05,
+      "loss": 0.9825,
+      "step": 3321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1312332497711753e-05,
+      "loss": 0.9499,
+      "step": 3322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1307746254299858e-05,
+      "loss": 0.9525,
+      "step": 3323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303159731018341e-05,
+      "loss": 0.9464,
+      "step": 3324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298572928848758e-05,
+      "loss": 1.0076,
+      "step": 3325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.129398584877273e-05,
+      "loss": 0.9694,
+      "step": 3326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289398491771932e-05,
+      "loss": 1.0108,
+      "step": 3327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12848108588281e-05,
+      "loss": 1.0344,
+      "step": 3328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280222950923028e-05,
+      "loss": 0.8947,
+      "step": 3329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275634769038573e-05,
+      "loss": 1.0263,
+      "step": 3330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271046314156646e-05,
+      "loss": 0.9795,
+      "step": 3331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126645758725922e-05,
+      "loss": 0.9763,
+      "step": 3332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261868589328322e-05,
+      "loss": 0.9231,
+      "step": 3333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257279321346039e-05,
+      "loss": 0.9635,
+      "step": 3334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252689784294524e-05,
+      "loss": 0.9314,
+      "step": 3335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248099979155966e-05,
+      "loss": 0.9532,
+      "step": 3336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243509906912635e-05,
+      "loss": 0.936,
+      "step": 3337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238919568546846e-05,
+      "loss": 0.9334,
+      "step": 3338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234328965040974e-05,
+      "loss": 0.9882,
+      "step": 3339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229738097377448e-05,
+      "loss": 1.0221,
+      "step": 3340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225146966538758e-05,
+      "loss": 0.9046,
+      "step": 3341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220555573507446e-05,
+      "loss": 1.0162,
+      "step": 3342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215963919266112e-05,
+      "loss": 0.3609,
+      "step": 3343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211372004797413e-05,
+      "loss": 0.3852,
+      "step": 3344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1206779831084059e-05,
+      "loss": 0.9557,
+      "step": 3345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120218739910882e-05,
+      "loss": 0.9331,
+      "step": 3346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1197594709854514e-05,
+      "loss": 0.9465,
+      "step": 3347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193001764304019e-05,
+      "loss": 0.95,
+      "step": 3348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188408563440273e-05,
+      "loss": 0.9713,
+      "step": 3349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118381510824626e-05,
+      "loss": 0.9934,
+      "step": 3350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179221399705016e-05,
+      "loss": 0.943,
+      "step": 3351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1174627438799642e-05,
+      "loss": 0.9835,
+      "step": 3352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170033226513283e-05,
+      "loss": 0.9752,
+      "step": 3353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165438763829149e-05,
+      "loss": 0.9542,
+      "step": 3354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160844051730493e-05,
+      "loss": 1.0013,
+      "step": 3355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156249091200626e-05,
+      "loss": 0.9253,
+      "step": 3356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151653883222911e-05,
+      "loss": 0.9563,
+      "step": 3357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147058428780762e-05,
+      "loss": 0.9968,
+      "step": 3358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114246272885765e-05,
+      "loss": 0.9731,
+      "step": 3359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11378667844371e-05,
+      "loss": 0.956,
+      "step": 3360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133270596502684e-05,
+      "loss": 0.9849,
+      "step": 3361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128674166038027e-05,
+      "loss": 0.399,
+      "step": 3362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124077494026806e-05,
+      "loss": 0.9038,
+      "step": 3363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119480581452752e-05,
+      "loss": 0.9201,
+      "step": 3364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114883429299648e-05,
+      "loss": 0.9547,
+      "step": 3365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110286038551323e-05,
+      "loss": 1.0012,
+      "step": 3366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1105688410191664e-05,
+      "loss": 0.9976,
+      "step": 3367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1101090545204605e-05,
+      "loss": 0.9682,
+      "step": 3368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096492444574129e-05,
+      "loss": 0.9382,
+      "step": 3369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091894109284272e-05,
+      "loss": 0.9811,
+      "step": 3370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1087295540319118e-05,
+      "loss": 0.8785,
+      "step": 3371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1082696738662805e-05,
+      "loss": 0.9869,
+      "step": 3372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107809770529952e-05,
+      "loss": 0.9433,
+      "step": 3373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073498441213495e-05,
+      "loss": 0.3997,
+      "step": 3374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1068898947389017e-05,
+      "loss": 0.9328,
+      "step": 3375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064299224810419e-05,
+      "loss": 0.9108,
+      "step": 3376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059699274462081e-05,
+      "loss": 0.8925,
+      "step": 3377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055099097328437e-05,
+      "loss": 0.9848,
+      "step": 3378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050498694393964e-05,
+      "loss": 1.0139,
+      "step": 3379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045898066643192e-05,
+      "loss": 0.8704,
+      "step": 3380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10412972150607e-05,
+      "loss": 1.0067,
+      "step": 3381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103669614063111e-05,
+      "loss": 0.9155,
+      "step": 3382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1032094844339094e-05,
+      "loss": 1.0141,
+      "step": 3383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027493327169368e-05,
+      "loss": 0.9795,
+      "step": 3384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022891590106702e-05,
+      "loss": 0.9894,
+      "step": 3385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018289634135911e-05,
+      "loss": 0.9927,
+      "step": 3386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013687460241849e-05,
+      "loss": 1.0298,
+      "step": 3387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100908506940943e-05,
+      "loss": 0.9533,
+      "step": 3388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100448246262361e-05,
+      "loss": 0.9459,
+      "step": 3389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999879640869382e-05,
+      "loss": 0.9487,
+      "step": 3390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0995276605131793e-05,
+      "loss": 0.9206,
+      "step": 3391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990673356395935e-05,
+      "loss": 1.0128,
+      "step": 3392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0986069895646946e-05,
+      "loss": 0.9369,
+      "step": 3393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0981466223870012e-05,
+      "loss": 0.9453,
+      "step": 3394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0976862342050353e-05,
+      "loss": 0.9353,
+      "step": 3395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0972258251173252e-05,
+      "loss": 0.9526,
+      "step": 3396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0967653952224016e-05,
+      "loss": 0.9865,
+      "step": 3397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963049446188012e-05,
+      "loss": 0.9902,
+      "step": 3398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958444734050643e-05,
+      "loss": 0.9936,
+      "step": 3399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953839816797368e-05,
+      "loss": 0.9091,
+      "step": 3400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094923469541367e-05,
+      "loss": 0.9745,
+      "step": 3401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944629370885093e-05,
+      "loss": 0.9306,
+      "step": 3402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940023844197221e-05,
+      "loss": 0.8657,
+      "step": 3403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0935418116335674e-05,
+      "loss": 0.9587,
+      "step": 3404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093081218828612e-05,
+      "loss": 1.033,
+      "step": 3405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092620606103427e-05,
+      "loss": 0.9333,
+      "step": 3406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092159973556588e-05,
+      "loss": 0.3781,
+      "step": 3407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916993212866742e-05,
+      "loss": 0.972,
+      "step": 3408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912386493922695e-05,
+      "loss": 0.9317,
+      "step": 3409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0907779579719624e-05,
+      "loss": 0.964,
+      "step": 3410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903172471243442e-05,
+      "loss": 0.9835,
+      "step": 3411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898565169480117e-05,
+      "loss": 0.9729,
+      "step": 3412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0893957675415651e-05,
+      "loss": 0.9201,
+      "step": 3413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0889349990036095e-05,
+      "loss": 0.9532,
+      "step": 3414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088474211432753e-05,
+      "loss": 0.996,
+      "step": 3415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0880134049276085e-05,
+      "loss": 0.9589,
+      "step": 3416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087552579586793e-05,
+      "loss": 0.9778,
+      "step": 3417
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870917355089272e-05,
+      "loss": 1.0036,
+      "step": 3418
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866308727926357e-05,
+      "loss": 0.9463,
+      "step": 3419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0861699915365477e-05,
+      "loss": 0.9449,
+      "step": 3420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857090918392958e-05,
+      "loss": 0.9814,
+      "step": 3421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852481737995168e-05,
+      "loss": 0.3928,
+      "step": 3422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847872375158517e-05,
+      "loss": 0.8922,
+      "step": 3423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843262830869444e-05,
+      "loss": 0.9973,
+      "step": 3424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0838653106114434e-05,
+      "loss": 1.0064,
+      "step": 3425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834043201880014e-05,
+      "loss": 0.9062,
+      "step": 3426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829433119152745e-05,
+      "loss": 0.9708,
+      "step": 3427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0824822858919225e-05,
+      "loss": 0.9294,
+      "step": 3428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820212422166094e-05,
+      "loss": 0.9847,
+      "step": 3429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815601809880027e-05,
+      "loss": 0.8988,
+      "step": 3430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0810991023047735e-05,
+      "loss": 0.9945,
+      "step": 3431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0806380062655968e-05,
+      "loss": 1.0134,
+      "step": 3432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801768929691516e-05,
+      "loss": 0.9439,
+      "step": 3433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797157625141204e-05,
+      "loss": 0.928,
+      "step": 3434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0792546149991889e-05,
+      "loss": 0.9618,
+      "step": 3435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0787934505230476e-05,
+      "loss": 0.9827,
+      "step": 3436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783322691843893e-05,
+      "loss": 0.8832,
+      "step": 3437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0778710710819112e-05,
+      "loss": 0.9799,
+      "step": 3438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774098563143137e-05,
+      "loss": 0.9374,
+      "step": 3439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0769486249803012e-05,
+      "loss": 0.9466,
+      "step": 3440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0764873771785814e-05,
+      "loss": 0.9808,
+      "step": 3441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760261130078652e-05,
+      "loss": 0.9379,
+      "step": 3442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755648325668682e-05,
+      "loss": 0.978,
+      "step": 3443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751035359543075e-05,
+      "loss": 0.9564,
+      "step": 3444
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746422232689057e-05,
+      "loss": 0.9759,
+      "step": 3445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0741808946093872e-05,
+      "loss": 0.8799,
+      "step": 3446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737195500744812e-05,
+      "loss": 0.9617,
+      "step": 3447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0732581897629192e-05,
+      "loss": 0.9783,
+      "step": 3448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0727968137734368e-05,
+      "loss": 0.9404,
+      "step": 3449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723354222047724e-05,
+      "loss": 0.8979,
+      "step": 3450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718740151556682e-05,
+      "loss": 0.9138,
+      "step": 3451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714125927248693e-05,
+      "loss": 0.9453,
+      "step": 3452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0709511550111245e-05,
+      "loss": 0.9328,
+      "step": 3453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704897021131854e-05,
+      "loss": 0.9052,
+      "step": 3454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700282341298074e-05,
+      "loss": 1.0118,
+      "step": 3455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069566751159749e-05,
+      "loss": 0.9294,
+      "step": 3456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691052533017715e-05,
+      "loss": 0.9251,
+      "step": 3457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686437406546397e-05,
+      "loss": 0.9926,
+      "step": 3458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0681822133171216e-05,
+      "loss": 0.9505,
+      "step": 3459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067720671387988e-05,
+      "loss": 0.9615,
+      "step": 3460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0672591149660133e-05,
+      "loss": 0.9347,
+      "step": 3461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667975441499747e-05,
+      "loss": 0.9189,
+      "step": 3462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663359590386526e-05,
+      "loss": 0.9468,
+      "step": 3463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658743597308308e-05,
+      "loss": 0.9443,
+      "step": 3464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065412746325295e-05,
+      "loss": 0.9536,
+      "step": 3465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064951118920835e-05,
+      "loss": 0.9357,
+      "step": 3466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0644894776162435e-05,
+      "loss": 0.9266,
+      "step": 3467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640278225103158e-05,
+      "loss": 0.9395,
+      "step": 3468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635661537018503e-05,
+      "loss": 0.371,
+      "step": 3469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0631044712896484e-05,
+      "loss": 1.008,
+      "step": 3470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626427753725139e-05,
+      "loss": 0.9447,
+      "step": 3471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621810660492548e-05,
+      "loss": 0.917,
+      "step": 3472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617193434186803e-05,
+      "loss": 0.3791,
+      "step": 3473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0612576075796037e-05,
+      "loss": 0.9558,
+      "step": 3474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607958586308402e-05,
+      "loss": 0.9844,
+      "step": 3475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060334096671209e-05,
+      "loss": 0.961,
+      "step": 3476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059872321799531e-05,
+      "loss": 0.9227,
+      "step": 3477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594105341146302e-05,
+      "loss": 0.9344,
+      "step": 3478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589487337153335e-05,
+      "loss": 0.9681,
+      "step": 3479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05848692070047e-05,
+      "loss": 0.9559,
+      "step": 3480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580250951688724e-05,
+      "loss": 0.9254,
+      "step": 3481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575632572193753e-05,
+      "loss": 0.8963,
+      "step": 3482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571014069508163e-05,
+      "loss": 0.9464,
+      "step": 3483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056639544462036e-05,
+      "loss": 0.919,
+      "step": 3484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0561776698518761e-05,
+      "loss": 0.9387,
+      "step": 3485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557157832191829e-05,
+      "loss": 0.9237,
+      "step": 3486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552538846628042e-05,
+      "loss": 0.9441,
+      "step": 3487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547919742815902e-05,
+      "loss": 1.0361,
+      "step": 3488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543300521743941e-05,
+      "loss": 0.9382,
+      "step": 3489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0538681184400717e-05,
+      "loss": 0.9029,
+      "step": 3490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0534061731774805e-05,
+      "loss": 0.9642,
+      "step": 3491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529442164854817e-05,
+      "loss": 0.987,
+      "step": 3492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0524822484629375e-05,
+      "loss": 0.9832,
+      "step": 3493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520202692087138e-05,
+      "loss": 0.9616,
+      "step": 3494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051558278821678e-05,
+      "loss": 0.95,
+      "step": 3495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510962774007005e-05,
+      "loss": 0.9591,
+      "step": 3496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506342650446539e-05,
+      "loss": 0.8927,
+      "step": 3497
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0501722418524127e-05,
+      "loss": 0.9455,
+      "step": 3498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0497102079228545e-05,
+      "loss": 0.9645,
+      "step": 3499
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492481633548586e-05,
+      "loss": 0.9906,
+      "step": 3500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487861082473064e-05,
+      "loss": 0.9839,
+      "step": 3501
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0483240426990824e-05,
+      "loss": 0.9289,
+      "step": 3502
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478619668090725e-05,
+      "loss": 0.9044,
+      "step": 3503
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473998806761653e-05,
+      "loss": 0.3812,
+      "step": 3504
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469377843992516e-05,
+      "loss": 0.9626,
+      "step": 3505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0464756780772241e-05,
+      "loss": 0.9575,
+      "step": 3506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0460135618089775e-05,
+      "loss": 0.9143,
+      "step": 3507
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455514356934094e-05,
+      "loss": 0.9637,
+      "step": 3508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450892998294187e-05,
+      "loss": 1.0003,
+      "step": 3509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446271543159065e-05,
+      "loss": 0.9684,
+      "step": 3510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0441649992517767e-05,
+      "loss": 0.9881,
+      "step": 3511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437028347359342e-05,
+      "loss": 0.9811,
+      "step": 3512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432406608672865e-05,
+      "loss": 0.9485,
+      "step": 3513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427784777447432e-05,
+      "loss": 0.9338,
+      "step": 3514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423162854672158e-05,
+      "loss": 0.9948,
+      "step": 3515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0418540841336171e-05,
+      "loss": 0.998,
+      "step": 3516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041391873842863e-05,
+      "loss": 1.0199,
+      "step": 3517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0409296546938706e-05,
+      "loss": 0.9442,
+      "step": 3518
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0404674267855588e-05,
+      "loss": 0.9821,
+      "step": 3519
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400051902168487e-05,
+      "loss": 0.9834,
+      "step": 3520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395429450866632e-05,
+      "loss": 0.9471,
+      "step": 3521
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390806914939267e-05,
+      "loss": 0.9373,
+      "step": 3522
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0386184295375661e-05,
+      "loss": 1.0353,
+      "step": 3523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381561593165093e-05,
+      "loss": 0.9662,
+      "step": 3524
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0376938809296867e-05,
+      "loss": 0.9208,
+      "step": 3525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372315944760298e-05,
+      "loss": 0.9528,
+      "step": 3526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367693000544723e-05,
+      "loss": 0.9211,
+      "step": 3527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363069977639491e-05,
+      "loss": 0.9078,
+      "step": 3528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358446877033975e-05,
+      "loss": 0.97,
+      "step": 3529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353823699717558e-05,
+      "loss": 0.9228,
+      "step": 3530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0349200446679644e-05,
+      "loss": 0.9445,
+      "step": 3531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0344577118909651e-05,
+      "loss": 0.9246,
+      "step": 3532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339953717397013e-05,
+      "loss": 0.9748,
+      "step": 3533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335330243131176e-05,
+      "loss": 0.988,
+      "step": 3534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330706697101613e-05,
+      "loss": 1.0164,
+      "step": 3535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0326083080297801e-05,
+      "loss": 0.9937,
+      "step": 3536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321459393709237e-05,
+      "loss": 0.9371,
+      "step": 3537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0316835638325432e-05,
+      "loss": 0.9348,
+      "step": 3538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0312211815135914e-05,
+      "loss": 0.9906,
+      "step": 3539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307587925130216e-05,
+      "loss": 1.0562,
+      "step": 3540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.03029639692979e-05,
+      "loss": 0.9707,
+      "step": 3541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298339948628531e-05,
+      "loss": 0.9105,
+      "step": 3542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0293715864111693e-05,
+      "loss": 0.9569,
+      "step": 3543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289091716736983e-05,
+      "loss": 0.862,
+      "step": 3544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028446750749401e-05,
+      "loss": 0.9612,
+      "step": 3545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0279843237372395e-05,
+      "loss": 0.9791,
+      "step": 3546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275218907361777e-05,
+      "loss": 0.9498,
+      "step": 3547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0270594518451804e-05,
+      "loss": 0.9173,
+      "step": 3548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265970071632134e-05,
+      "loss": 0.9511,
+      "step": 3549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261345567892446e-05,
+      "loss": 0.9498,
+      "step": 3550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256721008222424e-05,
+      "loss": 0.963,
+      "step": 3551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252096393611764e-05,
+      "loss": 0.947,
+      "step": 3552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024747172505018e-05,
+      "loss": 0.9353,
+      "step": 3553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242847003527387e-05,
+      "loss": 0.9386,
+      "step": 3554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238222230033122e-05,
+      "loss": 0.8968,
+      "step": 3555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233597405557127e-05,
+      "loss": 0.9479,
+      "step": 3556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228972531089158e-05,
+      "loss": 1.0292,
+      "step": 3557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224347607618981e-05,
+      "loss": 0.9586,
+      "step": 3558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219722636136368e-05,
+      "loss": 0.9084,
+      "step": 3559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021509761763111e-05,
+      "loss": 0.9879,
+      "step": 3560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210472553093e-05,
+      "loss": 0.9697,
+      "step": 3561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205847443511845e-05,
+      "loss": 0.8859,
+      "step": 3562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0201222289877455e-05,
+      "loss": 0.9562,
+      "step": 3563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196597093179665e-05,
+      "loss": 0.3827,
+      "step": 3564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191971854408304e-05,
+      "loss": 0.927,
+      "step": 3565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187346574553216e-05,
+      "loss": 0.9471,
+      "step": 3566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0182721254604252e-05,
+      "loss": 0.946,
+      "step": 3567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0178095895551277e-05,
+      "loss": 0.9011,
+      "step": 3568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0173470498384152e-05,
+      "loss": 0.9935,
+      "step": 3569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016884506409276e-05,
+      "loss": 0.9499,
+      "step": 3570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164219593666984e-05,
+      "loss": 0.9973,
+      "step": 3571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015959408809672e-05,
+      "loss": 1.0072,
+      "step": 3572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154968548371867e-05,
+      "loss": 0.9531,
+      "step": 3573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150342975482327e-05,
+      "loss": 0.352,
+      "step": 3574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145717370418023e-05,
+      "loss": 0.9508,
+      "step": 3575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0141091734168874e-05,
+      "loss": 0.927,
+      "step": 3576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0136466067724805e-05,
+      "loss": 0.959,
+      "step": 3577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131840372075752e-05,
+      "loss": 0.9505,
+      "step": 3578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0127214648211658e-05,
+      "loss": 1.019,
+      "step": 3579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122588897122473e-05,
+      "loss": 0.9676,
+      "step": 3580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0117963119798144e-05,
+      "loss": 0.975,
+      "step": 3581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113337317228631e-05,
+      "loss": 0.9616,
+      "step": 3582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108711490403898e-05,
+      "loss": 0.9945,
+      "step": 3583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0104085640313917e-05,
+      "loss": 0.9893,
+      "step": 3584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099459767948659e-05,
+      "loss": 0.9682,
+      "step": 3585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094833874298102e-05,
+      "loss": 0.9426,
+      "step": 3586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090207960352237e-05,
+      "loss": 0.9232,
+      "step": 3587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085582027101039e-05,
+      "loss": 0.9431,
+      "step": 3588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080956075534512e-05,
+      "loss": 0.9138,
+      "step": 3589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076330106642641e-05,
+      "loss": 0.9836,
+      "step": 3590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071704121415435e-05,
+      "loss": 0.9165,
+      "step": 3591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067078120842893e-05,
+      "loss": 1.0317,
+      "step": 3592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062452105915022e-05,
+      "loss": 0.972,
+      "step": 3593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005782607762183e-05,
+      "loss": 0.9259,
+      "step": 3594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005320003695333e-05,
+      "loss": 0.9543,
+      "step": 3595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048573984899538e-05,
+      "loss": 0.9866,
+      "step": 3596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004394792245047e-05,
+      "loss": 0.9869,
+      "step": 3597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039321850596147e-05,
+      "loss": 0.9253,
+      "step": 3598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0034695770326586e-05,
+      "loss": 0.9461,
+      "step": 3599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030069682631817e-05,
+      "loss": 0.9603,
+      "step": 3600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025443588501863e-05,
+      "loss": 0.9379,
+      "step": 3601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020817488926747e-05,
+      "loss": 0.9225,
+      "step": 3602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0016191384896497e-05,
+      "loss": 0.9728,
+      "step": 3603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011565277401146e-05,
+      "loss": 0.8927,
+      "step": 3604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006939167430716e-05,
+      "loss": 0.9632,
+      "step": 3605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002313055975244e-05,
+      "loss": 0.9982,
+      "step": 3606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997686944024758e-06,
+      "loss": 0.9611,
+      "step": 3607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993060832569285e-06,
+      "loss": 0.9262,
+      "step": 3608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98843472259886e-06,
+      "loss": 0.9979,
+      "step": 3609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983808615103505e-06,
+      "loss": 0.969,
+      "step": 3610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979182511073256e-06,
+      "loss": 0.949,
+      "step": 3611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97455641149814e-06,
+      "loss": 1.0081,
+      "step": 3612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969930317368185e-06,
+      "loss": 0.9343,
+      "step": 3613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965304229673416e-06,
+      "loss": 0.9471,
+      "step": 3614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960678149403858e-06,
+      "loss": 0.9657,
+      "step": 3615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956052077549532e-06,
+      "loss": 0.9731,
+      "step": 3616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951426015100465e-06,
+      "loss": 0.978,
+      "step": 3617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.946799963046671e-06,
+      "loss": 0.9129,
+      "step": 3618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.942173922378172e-06,
+      "loss": 0.9355,
+      "step": 3619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93754789408498e-06,
+      "loss": 0.9337,
+      "step": 3620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93292187915711e-06,
+      "loss": 0.9421,
+      "step": 3621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928295878584568e-06,
+      "loss": 0.9606,
+      "step": 3622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923669893357362e-06,
+      "loss": 0.9348,
+      "step": 3623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919043924465494e-06,
+      "loss": 0.9736,
+      "step": 3624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914417972898966e-06,
+      "loss": 0.9407,
+      "step": 3625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90979203964777e-06,
+      "loss": 0.9657,
+      "step": 3626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9051661257019e-06,
+      "loss": 0.9586,
+      "step": 3627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900540232051348e-06,
+      "loss": 0.9486,
+      "step": 3628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895914359686088e-06,
+      "loss": 0.3466,
+      "step": 3629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891288509596107e-06,
+      "loss": 0.9737,
+      "step": 3630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.886662682771376e-06,
+      "loss": 0.8612,
+      "step": 3631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.882036880201861e-06,
+      "loss": 0.9813,
+      "step": 3632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877411102877532e-06,
+      "loss": 0.9702,
+      "step": 3633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872785351788345e-06,
+      "loss": 0.8806,
+      "step": 3634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868159627924251e-06,
+      "loss": 0.9512,
+      "step": 3635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863533932275197e-06,
+      "loss": 0.3153,
+      "step": 3636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.858908265831128e-06,
+      "loss": 0.9059,
+      "step": 3637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854282629581977e-06,
+      "loss": 0.9739,
+      "step": 3638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84965702451767e-06,
+      "loss": 0.9581,
+      "step": 3639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845031451628136e-06,
+      "loss": 0.8875,
+      "step": 3640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84040591190328e-06,
+      "loss": 0.3633,
+      "step": 3641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835780406333015e-06,
+      "loss": 1.0135,
+      "step": 3642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83115493590724e-06,
+      "loss": 0.894,
+      "step": 3643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.826529501615848e-06,
+      "loss": 0.9738,
+      "step": 3644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.821904104448725e-06,
+      "loss": 0.931,
+      "step": 3645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817278745395746e-06,
+      "loss": 0.952,
+      "step": 3646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812653425446785e-06,
+      "loss": 1.0158,
+      "step": 3647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808028145591697e-06,
+      "loss": 0.9792,
+      "step": 3648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803402906820336e-06,
+      "loss": 0.9555,
+      "step": 3649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798777710122545e-06,
+      "loss": 0.9715,
+      "step": 3650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794152556488158e-06,
+      "loss": 0.9913,
+      "step": 3651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789527446907003e-06,
+      "loss": 0.8882,
+      "step": 3652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784902382368891e-06,
+      "loss": 0.9448,
+      "step": 3653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780277363863635e-06,
+      "loss": 0.9704,
+      "step": 3654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775652392381024e-06,
+      "loss": 1.0025,
+      "step": 3655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771027468910844e-06,
+      "loss": 0.9686,
+      "step": 3656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766402594442875e-06,
+      "loss": 0.946,
+      "step": 3657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761777769966882e-06,
+      "loss": 0.9875,
+      "step": 3658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757152996472616e-06,
+      "loss": 0.938,
+      "step": 3659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752528274949824e-06,
+      "loss": 0.9036,
+      "step": 3660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74790360638824e-06,
+      "loss": 0.3215,
+      "step": 3661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74327899177758e-06,
+      "loss": 0.9833,
+      "step": 3662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738654432107557e-06,
+      "loss": 0.9443,
+      "step": 3663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734029928367868e-06,
+      "loss": 0.9741,
+      "step": 3664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.729405481548199e-06,
+      "loss": 0.944,
+      "step": 3665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724781092638225e-06,
+      "loss": 0.9779,
+      "step": 3666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720156762627607e-06,
+      "loss": 0.9708,
+      "step": 3667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715532492505994e-06,
+      "loss": 0.9541,
+      "step": 3668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71090828326302e-06,
+      "loss": 0.9299,
+      "step": 3669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70628413588831e-06,
+      "loss": 0.9673,
+      "step": 3670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701660051371472e-06,
+      "loss": 1.024,
+      "step": 3671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697036030702104e-06,
+      "loss": 0.9796,
+      "step": 3672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692412074869788e-06,
+      "loss": 0.9776,
+      "step": 3673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.687788184864091e-06,
+      "loss": 0.9308,
+      "step": 3674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.683164361674572e-06,
+      "loss": 0.9777,
+      "step": 3675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678540606290766e-06,
+      "loss": 0.9317,
+      "step": 3676
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6739169197022e-06,
+      "loss": 0.9328,
+      "step": 3677
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669293302898389e-06,
+      "loss": 0.9379,
+      "step": 3678
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664669756868825e-06,
+      "loss": 0.8859,
+      "step": 3679
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660046282602992e-06,
+      "loss": 0.8557,
+      "step": 3680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.655422881090352e-06,
+      "loss": 0.9873,
+      "step": 3681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650799553320358e-06,
+      "loss": 0.8843,
+      "step": 3682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646176300282447e-06,
+      "loss": 0.9865,
+      "step": 3683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64155312296603e-06,
+      "loss": 0.93,
+      "step": 3684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636930022360514e-06,
+      "loss": 0.909,
+      "step": 3685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632306999455283e-06,
+      "loss": 0.987,
+      "step": 3686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627684055239707e-06,
+      "loss": 0.9102,
+      "step": 3687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623061190703137e-06,
+      "loss": 1.0045,
+      "step": 3688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61843840683491e-06,
+      "loss": 0.9593,
+      "step": 3689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613815704624344e-06,
+      "loss": 0.9157,
+      "step": 3690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609193085060738e-06,
+      "loss": 0.9389,
+      "step": 3691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604570549133375e-06,
+      "loss": 1.0146,
+      "step": 3692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599948097831517e-06,
+      "loss": 0.9485,
+      "step": 3693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595325732144417e-06,
+      "loss": 0.9487,
+      "step": 3694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.590703453061296e-06,
+      "loss": 0.9494,
+      "step": 3695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586081261571371e-06,
+      "loss": 0.9451,
+      "step": 3696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581459158663829e-06,
+      "loss": 0.9506,
+      "step": 3697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576837145327844e-06,
+      "loss": 0.9555,
+      "step": 3698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572215222552568e-06,
+      "loss": 1.0077,
+      "step": 3699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567593391327135e-06,
+      "loss": 0.9877,
+      "step": 3700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562971652640662e-06,
+      "loss": 0.8734,
+      "step": 3701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.558350007482235e-06,
+      "loss": 0.9322,
+      "step": 3702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553728456840936e-06,
+      "loss": 0.9327,
+      "step": 3703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549107001705815e-06,
+      "loss": 0.921,
+      "step": 3704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544485643065908e-06,
+      "loss": 0.8858,
+      "step": 3705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539864381910225e-06,
+      "loss": 0.3404,
+      "step": 3706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53524321922776e-06,
+      "loss": 0.9504,
+      "step": 3707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530622156007487e-06,
+      "loss": 0.8995,
+      "step": 3708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526001193238349e-06,
+      "loss": 0.981,
+      "step": 3709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521380331909278e-06,
+      "loss": 1.013,
+      "step": 3710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51675957300918e-06,
+      "loss": 0.9406,
+      "step": 3711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512138917526938e-06,
+      "loss": 1.0148,
+      "step": 3712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507518366451418e-06,
+      "loss": 0.9164,
+      "step": 3713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.502897920771457e-06,
+      "loss": 0.8918,
+      "step": 3714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.498277581475876e-06,
+      "loss": 0.9458,
+      "step": 3715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.493657349553464e-06,
+      "loss": 0.9533,
+      "step": 3716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489037225992998e-06,
+      "loss": 0.9254,
+      "step": 3717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484417211783223e-06,
+      "loss": 0.967,
+      "step": 3718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479797307912865e-06,
+      "loss": 0.9706,
+      "step": 3719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.475177515370626e-06,
+      "loss": 0.9018,
+      "step": 3720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470557835145185e-06,
+      "loss": 0.9291,
+      "step": 3721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465938268225196e-06,
+      "loss": 0.9522,
+      "step": 3722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.461318815599287e-06,
+      "loss": 0.9809,
+      "step": 3723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45669947825606e-06,
+      "loss": 0.9699,
+      "step": 3724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452080257184101e-06,
+      "loss": 0.9252,
+      "step": 3725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447461153371962e-06,
+      "loss": 0.9969,
+      "step": 3726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442842167808173e-06,
+      "loss": 0.901,
+      "step": 3727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43822330148124e-06,
+      "loss": 0.9064,
+      "step": 3728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433604555379646e-06,
+      "loss": 0.9583,
+      "step": 3729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42898593049184e-06,
+      "loss": 0.9568,
+      "step": 3730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42436742780625e-06,
+      "loss": 0.8953,
+      "step": 3731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41974904831128e-06,
+      "loss": 0.9403,
+      "step": 3732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415130792995302e-06,
+      "loss": 0.9511,
+      "step": 3733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41051266284667e-06,
+      "loss": 0.9772,
+      "step": 3734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4058946588537e-06,
+      "loss": 0.9615,
+      "step": 3735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.401276782004692e-06,
+      "loss": 0.9169,
+      "step": 3736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.396659033287913e-06,
+      "loss": 0.9552,
+      "step": 3737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392041413691601e-06,
+      "loss": 0.8895,
+      "step": 3738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.387423924203968e-06,
+      "loss": 0.3883,
+      "step": 3739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.3828065658132e-06,
+      "loss": 0.9671,
+      "step": 3740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378189339507456e-06,
+      "loss": 1.0123,
+      "step": 3741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373572246274863e-06,
+      "loss": 0.983,
+      "step": 3742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36895528710352e-06,
+      "loss": 0.9201,
+      "step": 3743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364338462981502e-06,
+      "loss": 0.9481,
+      "step": 3744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359721774896847e-06,
+      "loss": 1.0109,
+      "step": 3745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.355105223837569e-06,
+      "loss": 0.8828,
+      "step": 3746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350488810791655e-06,
+      "loss": 0.9208,
+      "step": 3747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345872536747055e-06,
+      "loss": 0.9722,
+      "step": 3748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.341256402691699e-06,
+      "loss": 0.9507,
+      "step": 3749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336640409613477e-06,
+      "loss": 0.9743,
+      "step": 3750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33202455850026e-06,
+      "loss": 0.9568,
+      "step": 3751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327408850339872e-06,
+      "loss": 0.9548,
+      "step": 3752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322793286120122e-06,
+      "loss": 0.9959,
+      "step": 3753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318177866828785e-06,
+      "loss": 0.8984,
+      "step": 3754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313562593453605e-06,
+      "loss": 0.946,
+      "step": 3755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308947466982286e-06,
+      "loss": 0.9319,
+      "step": 3756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30433248840251e-06,
+      "loss": 0.9572,
+      "step": 3757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299717658701926e-06,
+      "loss": 0.9601,
+      "step": 3758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295102978868148e-06,
+      "loss": 0.9718,
+      "step": 3759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290488449888757e-06,
+      "loss": 0.9367,
+      "step": 3760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285874072751308e-06,
+      "loss": 0.9628,
+      "step": 3761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281259848443321e-06,
+      "loss": 0.9562,
+      "step": 3762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276645777952277e-06,
+      "loss": 0.921,
+      "step": 3763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272031862265634e-06,
+      "loss": 0.9459,
+      "step": 3764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.267418102370808e-06,
+      "loss": 0.9569,
+      "step": 3765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.262804499255188e-06,
+      "loss": 0.9317,
+      "step": 3766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258191053906126e-06,
+      "loss": 0.9379,
+      "step": 3767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253577767310944e-06,
+      "loss": 0.9982,
+      "step": 3768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248964640456926e-06,
+      "loss": 0.9067,
+      "step": 3769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244351674331322e-06,
+      "loss": 0.9012,
+      "step": 3770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23973886992135e-06,
+      "loss": 0.9577,
+      "step": 3771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23512622821419e-06,
+      "loss": 1.0109,
+      "step": 3772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230513750196991e-06,
+      "loss": 0.9687,
+      "step": 3773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225901436856865e-06,
+      "loss": 0.9449,
+      "step": 3774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221289289180891e-06,
+      "loss": 0.9143,
+      "step": 3775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21667730815611e-06,
+      "loss": 0.9419,
+      "step": 3776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212065494769528e-06,
+      "loss": 0.9218,
+      "step": 3777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207453850008112e-06,
+      "loss": 0.9719,
+      "step": 3778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202842374858799e-06,
+      "loss": 0.9368,
+      "step": 3779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198231070308486e-06,
+      "loss": 0.9684,
+      "step": 3780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193619937344035e-06,
+      "loss": 0.9675,
+      "step": 3781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.189008976952268e-06,
+      "loss": 0.973,
+      "step": 3782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184398190119976e-06,
+      "loss": 0.9998,
+      "step": 3783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179787577833909e-06,
+      "loss": 0.3419,
+      "step": 3784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175177141080778e-06,
+      "loss": 0.9658,
+      "step": 3785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170566880847258e-06,
+      "loss": 0.9573,
+      "step": 3786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165956798119987e-06,
+      "loss": 0.9167,
+      "step": 3787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16134689388557e-06,
+      "loss": 0.9975,
+      "step": 3788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156737169130561e-06,
+      "loss": 0.9753,
+      "step": 3789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152127624841488e-06,
+      "loss": 0.9066,
+      "step": 3790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147518262004834e-06,
+      "loss": 0.9201,
+      "step": 3791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142909081607043e-06,
+      "loss": 0.3619,
+      "step": 3792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.138300084634524e-06,
+      "loss": 0.9509,
+      "step": 3793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133691272073644e-06,
+      "loss": 0.9076,
+      "step": 3794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129082644910731e-06,
+      "loss": 0.3731,
+      "step": 3795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.124474204132072e-06,
+      "loss": 0.9626,
+      "step": 3796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119865950723917e-06,
+      "loss": 0.996,
+      "step": 3797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115257885672476e-06,
+      "loss": 0.9304,
+      "step": 3798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11065000996391e-06,
+      "loss": 0.8998,
+      "step": 3799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106042324584352e-06,
+      "loss": 0.9644,
+      "step": 3800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101434830519888e-06,
+      "loss": 0.9404,
+      "step": 3801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096827528756563e-06,
+      "loss": 0.9329,
+      "step": 3802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092220420280381e-06,
+      "loss": 0.9558,
+      "step": 3803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087613506077307e-06,
+      "loss": 0.9668,
+      "step": 3804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083006787133263e-06,
+      "loss": 0.9762,
+      "step": 3805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078400264434126e-06,
+      "loss": 0.9102,
+      "step": 3806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.073793938965735e-06,
+      "loss": 0.9091,
+      "step": 3807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069187811713885e-06,
+      "loss": 0.9381,
+      "step": 3808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064581883664331e-06,
+      "loss": 1.0038,
+      "step": 3809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059976155802784e-06,
+      "loss": 0.911,
+      "step": 3810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055370629114908e-06,
+      "loss": 0.9257,
+      "step": 3811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050765304586332e-06,
+      "loss": 0.9056,
+      "step": 3812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046160183202634e-06,
+      "loss": 0.9566,
+      "step": 3813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041555265949355e-06,
+      "loss": 0.9243,
+      "step": 3814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03695055381199e-06,
+      "loss": 0.9601,
+      "step": 3815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032346047775984e-06,
+      "loss": 0.958,
+      "step": 3816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027741748826752e-06,
+      "loss": 1.0175,
+      "step": 3817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023137657949647e-06,
+      "loss": 0.9528,
+      "step": 3818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01853377612999e-06,
+      "loss": 0.9701,
+      "step": 3819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013930104353052e-06,
+      "loss": 0.94,
+      "step": 3820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009326643604065e-06,
+      "loss": 0.9397,
+      "step": 3821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.004723394868208e-06,
+      "loss": 1.0202,
+      "step": 3822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00012035913062e-06,
+      "loss": 0.9788,
+      "step": 3823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995517537376393e-06,
+      "loss": 0.8973,
+      "step": 3824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99091493059057e-06,
+      "loss": 0.9672,
+      "step": 3825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986312539758153e-06,
+      "loss": 0.9539,
+      "step": 3826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981710365864094e-06,
+      "loss": 0.9759,
+      "step": 3827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9771084098933e-06,
+      "loss": 0.9459,
+      "step": 3828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972506672830636e-06,
+      "loss": 0.907,
+      "step": 3829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96790515566091e-06,
+      "loss": 1.0017,
+      "step": 3830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963303859368894e-06,
+      "loss": 0.9395,
+      "step": 3831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958702784939303e-06,
+      "loss": 0.9368,
+      "step": 3832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95410193335681e-06,
+      "loss": 0.9204,
+      "step": 3833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949501305606039e-06,
+      "loss": 0.9706,
+      "step": 3834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944900902671566e-06,
+      "loss": 0.926,
+      "step": 3835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94030072553792e-06,
+      "loss": 0.9379,
+      "step": 3836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935700775189584e-06,
+      "loss": 0.9516,
+      "step": 3837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.931101052610986e-06,
+      "loss": 0.9089,
+      "step": 3838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926501558786506e-06,
+      "loss": 0.9501,
+      "step": 3839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921902294700481e-06,
+      "loss": 0.9911,
+      "step": 3840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917303261337196e-06,
+      "loss": 0.9401,
+      "step": 3841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912704459680884e-06,
+      "loss": 0.9781,
+      "step": 3842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908105890715731e-06,
+      "loss": 0.8714,
+      "step": 3843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903507555425873e-06,
+      "loss": 1.0031,
+      "step": 3844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8989094547954e-06,
+      "loss": 0.9002,
+      "step": 3845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.894311589808337e-06,
+      "loss": 0.9316,
+      "step": 3846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889713961448678e-06,
+      "loss": 0.8964,
+      "step": 3847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885116570700354e-06,
+      "loss": 0.9446,
+      "step": 3848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88051941854725e-06,
+      "loss": 0.9137,
+      "step": 3849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875922505973196e-06,
+      "loss": 0.8862,
+      "step": 3850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871325833961976e-06,
+      "loss": 0.9939,
+      "step": 3851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866729403497321e-06,
+      "loss": 0.897,
+      "step": 3852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.862133215562902e-06,
+      "loss": 0.9853,
+      "step": 3853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857537271142352e-06,
+      "loss": 0.9303,
+      "step": 3854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852941571219243e-06,
+      "loss": 0.9479,
+      "step": 3855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848346116777094e-06,
+      "loss": 0.9176,
+      "step": 3856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843750908799379e-06,
+      "loss": 0.9635,
+      "step": 3857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83915594826951e-06,
+      "loss": 0.8873,
+      "step": 3858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834561236170855e-06,
+      "loss": 0.9357,
+      "step": 3859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829966773486722e-06,
+      "loss": 0.8934,
+      "step": 3860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825372561200365e-06,
+      "loss": 0.9744,
+      "step": 3861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820778600294991e-06,
+      "loss": 0.8835,
+      "step": 3862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816184891753747e-06,
+      "loss": 0.9371,
+      "step": 3863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811591436559732e-06,
+      "loss": 0.8968,
+      "step": 3864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806998235695983e-06,
+      "loss": 0.8967,
+      "step": 3865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80240529014549e-06,
+      "loss": 0.9517,
+      "step": 3866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797812600891187e-06,
+      "loss": 0.9513,
+      "step": 3867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793220168915946e-06,
+      "loss": 0.928,
+      "step": 3868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788627995202593e-06,
+      "loss": 0.35,
+      "step": 3869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78403608073389e-06,
+      "loss": 0.8522,
+      "step": 3870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779444426492556e-06,
+      "loss": 0.886,
+      "step": 3871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774853033461243e-06,
+      "loss": 0.8874,
+      "step": 3872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770261902622552e-06,
+      "loss": 0.8726,
+      "step": 3873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765671034959027e-06,
+      "loss": 0.9717,
+      "step": 3874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761080431453154e-06,
+      "loss": 0.9164,
+      "step": 3875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.756490093087365e-06,
+      "loss": 0.8819,
+      "step": 3876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751900020844034e-06,
+      "loss": 0.9248,
+      "step": 3877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74731021570548e-06,
+      "loss": 0.9408,
+      "step": 3878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74272067865396e-06,
+      "loss": 0.9348,
+      "step": 3879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738131410671678e-06,
+      "loss": 0.9616,
+      "step": 3880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733542412740782e-06,
+      "loss": 0.9745,
+      "step": 3881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728953685843354e-06,
+      "loss": 0.9447,
+      "step": 3882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724365230961427e-06,
+      "loss": 0.362,
+      "step": 3883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719777049076972e-06,
+      "loss": 0.8828,
+      "step": 3884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715189141171904e-06,
+      "loss": 0.9748,
+      "step": 3885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.710601508228071e-06,
+      "loss": 0.9733,
+      "step": 3886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706014151227273e-06,
+      "loss": 0.9056,
+      "step": 3887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701427071151244e-06,
+      "loss": 0.9195,
+      "step": 3888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69684026898166e-06,
+      "loss": 0.9293,
+      "step": 3889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692253745700144e-06,
+      "loss": 0.87,
+      "step": 3890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687667502288249e-06,
+      "loss": 0.9554,
+      "step": 3891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683081539727476e-06,
+      "loss": 0.9349,
+      "step": 3892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67849585899926e-06,
+      "loss": 0.9852,
+      "step": 3893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673910461084976e-06,
+      "loss": 0.9356,
+      "step": 3894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669325346965947e-06,
+      "loss": 0.9653,
+      "step": 3895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664740517623424e-06,
+      "loss": 0.9316,
+      "step": 3896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660155974038608e-06,
+      "loss": 0.9567,
+      "step": 3897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.655571717192628e-06,
+      "loss": 0.921,
+      "step": 3898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650987748066561e-06,
+      "loss": 0.9658,
+      "step": 3899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.646404067641415e-06,
+      "loss": 0.8677,
+      "step": 3900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641820676898139e-06,
+      "loss": 0.3592,
+      "step": 3901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637237576817623e-06,
+      "loss": 0.9455,
+      "step": 3902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63265476838069e-06,
+      "loss": 0.9141,
+      "step": 3903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628072252568104e-06,
+      "loss": 0.9129,
+      "step": 3904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623490030360565e-06,
+      "loss": 0.932,
+      "step": 3905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618908102738712e-06,
+      "loss": 0.9128,
+      "step": 3906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614326470683116e-06,
+      "loss": 0.9027,
+      "step": 3907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60974513517429e-06,
+      "loss": 0.963,
+      "step": 3908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60516409719268e-06,
+      "loss": 0.9929,
+      "step": 3909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60058335771867e-06,
+      "loss": 0.9183,
+      "step": 3910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.596002917732583e-06,
+      "loss": 0.9274,
+      "step": 3911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591422778214672e-06,
+      "loss": 0.3559,
+      "step": 3912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58684294014513e-06,
+      "loss": 0.8995,
+      "step": 3913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582263404504085e-06,
+      "loss": 0.9123,
+      "step": 3914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577684172271596e-06,
+      "loss": 0.9806,
+      "step": 3915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573105244427664e-06,
+      "loss": 0.9041,
+      "step": 3916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56852662195222e-06,
+      "loss": 0.9362,
+      "step": 3917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563948305825127e-06,
+      "loss": 0.9666,
+      "step": 3918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559370297026195e-06,
+      "loss": 0.9653,
+      "step": 3919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554792596535152e-06,
+      "loss": 0.8721,
+      "step": 3920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550215205331673e-06,
+      "loss": 0.925,
+      "step": 3921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545638124395357e-06,
+      "loss": 0.9366,
+      "step": 3922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541061354705743e-06,
+      "loss": 1.008,
+      "step": 3923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5364848972423e-06,
+      "loss": 1.0113,
+      "step": 3924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.531908752984435e-06,
+      "loss": 0.9979,
+      "step": 3925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527332922911481e-06,
+      "loss": 0.9095,
+      "step": 3926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522757408002711e-06,
+      "loss": 0.9849,
+      "step": 3927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518182209237326e-06,
+      "loss": 0.3398,
+      "step": 3928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513607327594451e-06,
+      "loss": 0.9605,
+      "step": 3929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509032764053167e-06,
+      "loss": 0.9006,
+      "step": 3930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504458519592466e-06,
+      "loss": 1.0288,
+      "step": 3931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49988459519128e-06,
+      "loss": 0.8585,
+      "step": 3932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.495310991828463e-06,
+      "loss": 0.983,
+      "step": 3933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.490737710482816e-06,
+      "loss": 0.9708,
+      "step": 3934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48616475213306e-06,
+      "loss": 0.9405,
+      "step": 3935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48159211775785e-06,
+      "loss": 0.9822,
+      "step": 3936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.477019808335768e-06,
+      "loss": 0.9593,
+      "step": 3937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472447824845338e-06,
+      "loss": 0.9534,
+      "step": 3938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467876168265e-06,
+      "loss": 0.9679,
+      "step": 3939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463304839573132e-06,
+      "loss": 0.9098,
+      "step": 3940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458733839748038e-06,
+      "loss": 0.9335,
+      "step": 3941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454163169767956e-06,
+      "loss": 0.9687,
+      "step": 3942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44959283061105e-06,
+      "loss": 0.9592,
+      "step": 3943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445022823255414e-06,
+      "loss": 0.8819,
+      "step": 3944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440453148679073e-06,
+      "loss": 0.9542,
+      "step": 3945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43588380785998e-06,
+      "loss": 0.9792,
+      "step": 3946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431314801776009e-06,
+      "loss": 1.0142,
+      "step": 3947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426746131404974e-06,
+      "loss": 0.9849,
+      "step": 3948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422177797724614e-06,
+      "loss": 0.9592,
+      "step": 3949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41760980171259e-06,
+      "loss": 0.9707,
+      "step": 3950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413042144346495e-06,
+      "loss": 0.9471,
+      "step": 3951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408474826603852e-06,
+      "loss": 0.9737,
+      "step": 3952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.403907849462109e-06,
+      "loss": 0.9447,
+      "step": 3953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399341213898638e-06,
+      "loss": 0.9301,
+      "step": 3954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394774920890741e-06,
+      "loss": 0.9342,
+      "step": 3955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390208971415646e-06,
+      "loss": 0.938,
+      "step": 3956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38564336645051e-06,
+      "loss": 0.958,
+      "step": 3957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.381078106972413e-06,
+      "loss": 0.95,
+      "step": 3958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37651319395836e-06,
+      "loss": 0.9759,
+      "step": 3959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37194862838529e-06,
+      "loss": 0.9406,
+      "step": 3960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367384411230056e-06,
+      "loss": 0.9689,
+      "step": 3961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.362820543469445e-06,
+      "loss": 0.9409,
+      "step": 3962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358257026080161e-06,
+      "loss": 0.9601,
+      "step": 3963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353693860038847e-06,
+      "loss": 0.9463,
+      "step": 3964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349131046322054e-06,
+      "loss": 0.8961,
+      "step": 3965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344568585906271e-06,
+      "loss": 0.9592,
+      "step": 3966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340006479767906e-06,
+      "loss": 0.9385,
+      "step": 3967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33544472888329e-06,
+      "loss": 0.9806,
+      "step": 3968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330883334228676e-06,
+      "loss": 0.8743,
+      "step": 3969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32632229678025e-06,
+      "loss": 0.9491,
+      "step": 3970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32176161751411e-06,
+      "loss": 0.9587,
+      "step": 3971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317201297406288e-06,
+      "loss": 0.951,
+      "step": 3972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31264133743273e-06,
+      "loss": 0.9224,
+      "step": 3973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308081738569312e-06,
+      "loss": 0.9832,
+      "step": 3974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30352250179183e-06,
+      "loss": 0.972,
+      "step": 3975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298963628075998e-06,
+      "loss": 0.9557,
+      "step": 3976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294405118397459e-06,
+      "loss": 0.9048,
+      "step": 3977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289846973731777e-06,
+      "loss": 0.9132,
+      "step": 3978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285289195054435e-06,
+      "loss": 0.9365,
+      "step": 3979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28073178334084e-06,
+      "loss": 0.903,
+      "step": 3980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276174739566319e-06,
+      "loss": 0.9511,
+      "step": 3981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.271618064706124e-06,
+      "loss": 0.9981,
+      "step": 3982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26706175973542e-06,
+      "loss": 0.9278,
+      "step": 3983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262505825629298e-06,
+      "loss": 0.9012,
+      "step": 3984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.257950263362773e-06,
+      "loss": 0.943,
+      "step": 3985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253395073910776e-06,
+      "loss": 0.8971,
+      "step": 3986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.248840258248156e-06,
+      "loss": 1.0068,
+      "step": 3987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244285817349688e-06,
+      "loss": 0.8942,
+      "step": 3988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239731752190063e-06,
+      "loss": 0.9267,
+      "step": 3989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235178063743895e-06,
+      "loss": 0.9267,
+      "step": 3990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.230624752985711e-06,
+      "loss": 0.9415,
+      "step": 3991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226071820889963e-06,
+      "loss": 0.9273,
+      "step": 3992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221519268431024e-06,
+      "loss": 0.953,
+      "step": 3993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216967096583172e-06,
+      "loss": 1.0107,
+      "step": 3994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21241530632062e-06,
+      "loss": 0.927,
+      "step": 3995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207863898617491e-06,
+      "loss": 0.9628,
+      "step": 3996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.203312874447829e-06,
+      "loss": 0.9374,
+      "step": 3997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198762234785593e-06,
+      "loss": 0.9015,
+      "step": 3998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194211980604661e-06,
+      "loss": 0.9616,
+      "step": 3999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189662112878833e-06,
+      "loss": 0.9707,
+      "step": 4000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.185112632581817e-06,
+      "loss": 0.9109,
+      "step": 4001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180563540687246e-06,
+      "loss": 0.8843,
+      "step": 4002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176014838168664e-06,
+      "loss": 0.918,
+      "step": 4003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17146652599954e-06,
+      "loss": 0.3388,
+      "step": 4004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16691860515325e-06,
+      "loss": 0.9339,
+      "step": 4005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.162371076603092e-06,
+      "loss": 0.9577,
+      "step": 4006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157823941322281e-06,
+      "loss": 0.9459,
+      "step": 4007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153277200283945e-06,
+      "loss": 0.9387,
+      "step": 4008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148730854461122e-06,
+      "loss": 0.8959,
+      "step": 4009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144184904826776e-06,
+      "loss": 0.9476,
+      "step": 4010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139639352353782e-06,
+      "loss": 0.8957,
+      "step": 4011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135094198014929e-06,
+      "loss": 0.9624,
+      "step": 4012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130549442782921e-06,
+      "loss": 0.943,
+      "step": 4013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126005087630378e-06,
+      "loss": 0.9617,
+      "step": 4014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121461133529835e-06,
+      "loss": 0.8846,
+      "step": 4015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116917581453736e-06,
+      "loss": 0.9756,
+      "step": 4016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112374432374442e-06,
+      "loss": 0.9154,
+      "step": 4017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107831687264231e-06,
+      "loss": 0.9346,
+      "step": 4018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103289347095292e-06,
+      "loss": 0.9617,
+      "step": 4019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098747412839728e-06,
+      "loss": 0.9081,
+      "step": 4020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09420588546955e-06,
+      "loss": 0.8691,
+      "step": 4021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.089664765956693e-06,
+      "loss": 0.9567,
+      "step": 4022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085124055272992e-06,
+      "loss": 0.3836,
+      "step": 4023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080583754390201e-06,
+      "loss": 0.9344,
+      "step": 4024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07604386427999e-06,
+      "loss": 1.0047,
+      "step": 4025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071504385913931e-06,
+      "loss": 0.3585,
+      "step": 4026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06696532026352e-06,
+      "loss": 0.3891,
+      "step": 4027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062426668300154e-06,
+      "loss": 0.9826,
+      "step": 4028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057888430995151e-06,
+      "loss": 0.9467,
+      "step": 4029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053350609319728e-06,
+      "loss": 0.9599,
+      "step": 4030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048813204245023e-06,
+      "loss": 0.9828,
+      "step": 4031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044276216742083e-06,
+      "loss": 0.9152,
+      "step": 4032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039739647781867e-06,
+      "loss": 0.9673,
+      "step": 4033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035203498335238e-06,
+      "loss": 0.8719,
+      "step": 4034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030667769372974e-06,
+      "loss": 0.9646,
+      "step": 4035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026132461865767e-06,
+      "loss": 0.9996,
+      "step": 4036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021597576784207e-06,
+      "loss": 0.9135,
+      "step": 4037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017063115098805e-06,
+      "loss": 0.9425,
+      "step": 4038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012529077779975e-06,
+      "loss": 0.9942,
+      "step": 4039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007995465798046e-06,
+      "loss": 0.9215,
+      "step": 4040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.003462280123248e-06,
+      "loss": 0.9341,
+      "step": 4041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998929521725727e-06,
+      "loss": 0.9056,
+      "step": 4042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994397191575537e-06,
+      "loss": 0.972,
+      "step": 4043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98986529064263e-06,
+      "loss": 0.9348,
+      "step": 4044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98533381989688e-06,
+      "loss": 0.9512,
+      "step": 4045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.980802780308058e-06,
+      "loss": 0.9669,
+      "step": 4046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976272172845857e-06,
+      "loss": 0.904,
+      "step": 4047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.971741998479862e-06,
+      "loss": 0.9743,
+      "step": 4048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967212258179571e-06,
+      "loss": 0.9034,
+      "step": 4049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962682952914392e-06,
+      "loss": 0.9551,
+      "step": 4050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958154083653638e-06,
+      "loss": 0.3612,
+      "step": 4051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953625651366525e-06,
+      "loss": 0.3806,
+      "step": 4052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949097657022183e-06,
+      "loss": 0.9093,
+      "step": 4053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944570101589642e-06,
+      "loss": 0.8908,
+      "step": 4054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940042986037842e-06,
+      "loss": 0.9398,
+      "step": 4055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935516311335622e-06,
+      "loss": 0.9761,
+      "step": 4056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.930990078451736e-06,
+      "loss": 0.9221,
+      "step": 4057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926464288354838e-06,
+      "loss": 0.3868,
+      "step": 4058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.921938942013488e-06,
+      "loss": 0.9002,
+      "step": 4059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917414040396151e-06,
+      "loss": 0.9643,
+      "step": 4060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912889584471199e-06,
+      "loss": 0.989,
+      "step": 4061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908365575206908e-06,
+      "loss": 0.8945,
+      "step": 4062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903842013571452e-06,
+      "loss": 0.3636,
+      "step": 4063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899318900532917e-06,
+      "loss": 0.3369,
+      "step": 4064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894796237059292e-06,
+      "loss": 0.9258,
+      "step": 4065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890274024118467e-06,
+      "loss": 0.9362,
+      "step": 4066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885752262678236e-06,
+      "loss": 0.9319,
+      "step": 4067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881230953706298e-06,
+      "loss": 0.3749,
+      "step": 4068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876710098170259e-06,
+      "loss": 0.9557,
+      "step": 4069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872189697037614e-06,
+      "loss": 0.9361,
+      "step": 4070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.867669751275775e-06,
+      "loss": 0.9695,
+      "step": 4071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863150261852053e-06,
+      "loss": 0.9612,
+      "step": 4072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858631229733656e-06,
+      "loss": 0.9893,
+      "step": 4073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854112655887702e-06,
+      "loss": 0.9921,
+      "step": 4074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849594541281205e-06,
+      "loss": 0.9759,
+      "step": 4075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.845076886881085e-06,
+      "loss": 0.9953,
+      "step": 4076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840559693654158e-06,
+      "loss": 0.9399,
+      "step": 4077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836042962567146e-06,
+      "loss": 0.8616,
+      "step": 4078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831526694586669e-06,
+      "loss": 0.8965,
+      "step": 4079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.827010890679251e-06,
+      "loss": 0.9566,
+      "step": 4080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822495551811316e-06,
+      "loss": 0.8947,
+      "step": 4081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817980678949187e-06,
+      "loss": 0.9234,
+      "step": 4082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81346627305909e-06,
+      "loss": 0.9126,
+      "step": 4083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.808952335107145e-06,
+      "loss": 0.9613,
+      "step": 4084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804438866059377e-06,
+      "loss": 0.9112,
+      "step": 4085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79992586688171e-06,
+      "loss": 0.9946,
+      "step": 4086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795413338539969e-06,
+      "loss": 0.8888,
+      "step": 4087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790901281999873e-06,
+      "loss": 0.9511,
+      "step": 4088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786389698227047e-06,
+      "loss": 0.9917,
+      "step": 4089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781878588187008e-06,
+      "loss": 0.921,
+      "step": 4090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777367952845175e-06,
+      "loss": 0.9763,
+      "step": 4091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772857793166865e-06,
+      "loss": 0.9319,
+      "step": 4092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768348110117295e-06,
+      "loss": 0.8896,
+      "step": 4093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763838904661575e-06,
+      "loss": 0.9877,
+      "step": 4094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759330177764718e-06,
+      "loss": 0.9277,
+      "step": 4095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.754821930391636e-06,
+      "loss": 1.0415,
+      "step": 4096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.75031416350713e-06,
+      "loss": 0.3598,
+      "step": 4097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745806878075908e-06,
+      "loss": 0.8888,
+      "step": 4098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741300075062564e-06,
+      "loss": 0.9198,
+      "step": 4099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.7367937554316e-06,
+      "loss": 1.0032,
+      "step": 4100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.732287920147405e-06,
+      "loss": 0.8867,
+      "step": 4101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727782570174274e-06,
+      "loss": 0.934,
+      "step": 4102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.723277706476391e-06,
+      "loss": 0.9864,
+      "step": 4103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.718773330017835e-06,
+      "loss": 0.9548,
+      "step": 4104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714269441762587e-06,
+      "loss": 0.9608,
+      "step": 4105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.709766042674518e-06,
+      "loss": 0.9478,
+      "step": 4106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.7052631337174e-06,
+      "loss": 0.9448,
+      "step": 4107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700760715854894e-06,
+      "loss": 0.9996,
+      "step": 4108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.696258790050562e-06,
+      "loss": 0.9435,
+      "step": 4109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691757357267849e-06,
+      "loss": 0.9118,
+      "step": 4110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68725641847011e-06,
+      "loss": 0.969,
+      "step": 4111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.682755974620584e-06,
+      "loss": 0.9713,
+      "step": 4112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678256026682407e-06,
+      "loss": 0.9383,
+      "step": 4113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67375657561861e-06,
+      "loss": 0.9651,
+      "step": 4114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.669257622392115e-06,
+      "loss": 0.9309,
+      "step": 4115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664759167965743e-06,
+      "loss": 0.9386,
+      "step": 4116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660261213302198e-06,
+      "loss": 0.9614,
+      "step": 4117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65576375936409e-06,
+      "loss": 0.358,
+      "step": 4118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651266807113908e-06,
+      "loss": 0.9449,
+      "step": 4119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646770357514046e-06,
+      "loss": 0.9501,
+      "step": 4120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642274411526783e-06,
+      "loss": 0.956,
+      "step": 4121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637778970114295e-06,
+      "loss": 0.9611,
+      "step": 4122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633284034238645e-06,
+      "loss": 0.939,
+      "step": 4123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628789604861791e-06,
+      "loss": 0.9735,
+      "step": 4124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.624295682945582e-06,
+      "loss": 0.9101,
+      "step": 4125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619802269451756e-06,
+      "loss": 1.0099,
+      "step": 4126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615309365341948e-06,
+      "loss": 0.944,
+      "step": 4127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610816971577678e-06,
+      "loss": 0.9965,
+      "step": 4128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606325089120361e-06,
+      "loss": 0.9217,
+      "step": 4129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601833718931302e-06,
+      "loss": 0.9983,
+      "step": 4130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.59734286197169e-06,
+      "loss": 0.922,
+      "step": 4131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592852519202613e-06,
+      "loss": 0.9484,
+      "step": 4132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588362691585045e-06,
+      "loss": 0.8935,
+      "step": 4133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.583873380079849e-06,
+      "loss": 0.9299,
+      "step": 4134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5793845856477796e-06,
+      "loss": 0.9196,
+      "step": 4135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57489630924948e-06,
+      "loss": 0.9409,
+      "step": 4136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.570408551845484e-06,
+      "loss": 0.9335,
+      "step": 4137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565921314396209e-06,
+      "loss": 0.9545,
+      "step": 4138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5614345978619655e-06,
+      "loss": 0.9196,
+      "step": 4139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556948403202951e-06,
+      "loss": 0.9428,
+      "step": 4140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5524627313792545e-06,
+      "loss": 0.9421,
+      "step": 4141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54797758335085e-06,
+      "loss": 0.9851,
+      "step": 4142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543492960077599e-06,
+      "loss": 0.932,
+      "step": 4143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539008862519253e-06,
+      "loss": 0.9987,
+      "step": 4144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534525291635451e-06,
+      "loss": 0.9278,
+      "step": 4145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530042248385714e-06,
+      "loss": 1.0209,
+      "step": 4146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525559733729456e-06,
+      "loss": 0.9323,
+      "step": 4147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.521077748625978e-06,
+      "loss": 0.904,
+      "step": 4148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516596294034463e-06,
+      "loss": 0.9249,
+      "step": 4149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.512115370913983e-06,
+      "loss": 0.9761,
+      "step": 4150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507634980223499e-06,
+      "loss": 0.9695,
+      "step": 4151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5031551229218565e-06,
+      "loss": 0.9244,
+      "step": 4152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498675799967781e-06,
+      "loss": 0.8538,
+      "step": 4153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49419701231989e-06,
+      "loss": 0.9892,
+      "step": 4154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489718760936687e-06,
+      "loss": 0.9373,
+      "step": 4155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.485241046776557e-06,
+      "loss": 0.3338,
+      "step": 4156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480763870797771e-06,
+      "loss": 0.9767,
+      "step": 4157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476287233958488e-06,
+      "loss": 0.8907,
+      "step": 4158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471811137216749e-06,
+      "loss": 0.9371,
+      "step": 4159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.467335581530476e-06,
+      "loss": 0.9048,
+      "step": 4160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462860567857482e-06,
+      "loss": 0.9849,
+      "step": 4161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458386097155457e-06,
+      "loss": 1.002,
+      "step": 4162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45391217038198e-06,
+      "loss": 0.9365,
+      "step": 4163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449438788494516e-06,
+      "loss": 0.9457,
+      "step": 4164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444965952450407e-06,
+      "loss": 0.917,
+      "step": 4165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44049366320688e-06,
+      "loss": 0.8721,
+      "step": 4166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.436021921721046e-06,
+      "loss": 0.9205,
+      "step": 4167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431550728949899e-06,
+      "loss": 0.8989,
+      "step": 4168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.427080085850314e-06,
+      "loss": 0.9516,
+      "step": 4169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4226099933790515e-06,
+      "loss": 1.0425,
+      "step": 4170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4181404524927495e-06,
+      "loss": 0.961,
+      "step": 4171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4136714641479314e-06,
+      "loss": 0.9447,
+      "step": 4172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409203029301001e-06,
+      "loss": 0.9284,
+      "step": 4173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404735148908244e-06,
+      "loss": 0.9611,
+      "step": 4174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40026782392583e-06,
+      "loss": 0.9091,
+      "step": 4175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395801055309804e-06,
+      "loss": 0.925,
+      "step": 4176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3913348440161e-06,
+      "loss": 0.96,
+      "step": 4177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386869191000521e-06,
+      "loss": 0.9245,
+      "step": 4178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382404097218763e-06,
+      "loss": 0.9794,
+      "step": 4179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3779395636263925e-06,
+      "loss": 0.9585,
+      "step": 4180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3734755911788644e-06,
+      "loss": 0.3635,
+      "step": 4181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3690121808315075e-06,
+      "loss": 0.921,
+      "step": 4182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364549333539533e-06,
+      "loss": 1.0201,
+      "step": 4183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360087050258034e-06,
+      "loss": 0.8675,
+      "step": 4184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355625331941971e-06,
+      "loss": 0.9619,
+      "step": 4185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3511641795462e-06,
+      "loss": 0.9388,
+      "step": 4186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346703594025446e-06,
+      "loss": 0.9487,
+      "step": 4187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342243576334315e-06,
+      "loss": 0.8996,
+      "step": 4188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337784127427291e-06,
+      "loss": 0.91,
+      "step": 4189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333325248258738e-06,
+      "loss": 0.9322,
+      "step": 4190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328866939782895e-06,
+      "loss": 0.9817,
+      "step": 4191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3244092029538834e-06,
+      "loss": 0.9487,
+      "step": 4192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319952038725697e-06,
+      "loss": 0.93,
+      "step": 4193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.31549544805221e-06,
+      "loss": 0.8852,
+      "step": 4194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311039431887175e-06,
+      "loss": 0.9538,
+      "step": 4195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3065839911842175e-06,
+      "loss": 0.9482,
+      "step": 4196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302129126896844e-06,
+      "loss": 0.919,
+      "step": 4197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297674839978437e-06,
+      "loss": 0.9402,
+      "step": 4198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2932211313822545e-06,
+      "loss": 0.9399,
+      "step": 4199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288768002061427e-06,
+      "loss": 0.371,
+      "step": 4200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284315452968969e-06,
+      "loss": 0.8926,
+      "step": 4201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279863485057764e-06,
+      "loss": 0.9415,
+      "step": 4202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275412099280575e-06,
+      "loss": 0.934,
+      "step": 4203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27096129659004e-06,
+      "loss": 0.9039,
+      "step": 4204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266511077938669e-06,
+      "loss": 0.9592,
+      "step": 4205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.262061444278854e-06,
+      "loss": 0.932,
+      "step": 4206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.257612396562849e-06,
+      "loss": 0.914,
+      "step": 4207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2531639357428e-06,
+      "loss": 0.9245,
+      "step": 4208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.248716062770711e-06,
+      "loss": 0.9459,
+      "step": 4209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.244268778598471e-06,
+      "loss": 0.9977,
+      "step": 4210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239822084177838e-06,
+      "loss": 0.9127,
+      "step": 4211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.235375980460446e-06,
+      "loss": 0.9931,
+      "step": 4212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.230930468397804e-06,
+      "loss": 0.9423,
+      "step": 4213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226485548941286e-06,
+      "loss": 0.9501,
+      "step": 4214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222041223042149e-06,
+      "loss": 0.9334,
+      "step": 4215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2175974916515204e-06,
+      "loss": 0.8803,
+      "step": 4216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213154355720398e-06,
+      "loss": 0.9386,
+      "step": 4217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208711816199651e-06,
+      "loss": 0.9826,
+      "step": 4218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204269874040029e-06,
+      "loss": 0.3711,
+      "step": 4219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199828530192145e-06,
+      "loss": 0.9832,
+      "step": 4220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195387785606485e-06,
+      "loss": 0.8894,
+      "step": 4221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.190947641233405e-06,
+      "loss": 0.9363,
+      "step": 4222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1865080980231495e-06,
+      "loss": 0.913,
+      "step": 4223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182069156925815e-06,
+      "loss": 0.9275,
+      "step": 4224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17763081889137e-06,
+      "loss": 0.9032,
+      "step": 4225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.173193084869665e-06,
+      "loss": 0.9083,
+      "step": 4226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168755955810414e-06,
+      "loss": 0.9132,
+      "step": 4227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1643194326632034e-06,
+      "loss": 0.9361,
+      "step": 4228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.15988351637749e-06,
+      "loss": 0.9478,
+      "step": 4229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155448207902598e-06,
+      "loss": 0.9168,
+      "step": 4230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.15101350818773e-06,
+      "loss": 0.947,
+      "step": 4231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146579418181943e-06,
+      "loss": 0.9217,
+      "step": 4232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1421459388341786e-06,
+      "loss": 1.0157,
+      "step": 4233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13771307109324e-06,
+      "loss": 0.939,
+      "step": 4234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.133280815907801e-06,
+      "loss": 0.9422,
+      "step": 4235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128849174226409e-06,
+      "loss": 0.8651,
+      "step": 4236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.12441814699747e-06,
+      "loss": 0.944,
+      "step": 4237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119987735169265e-06,
+      "loss": 0.8683,
+      "step": 4238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.115557939689947e-06,
+      "loss": 0.9378,
+      "step": 4239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111128761507529e-06,
+      "loss": 0.93,
+      "step": 4240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106700201569896e-06,
+      "loss": 0.9461,
+      "step": 4241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102272260824799e-06,
+      "loss": 0.9454,
+      "step": 4242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.09784494021986e-06,
+      "loss": 0.9525,
+      "step": 4243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093418240702565e-06,
+      "loss": 0.9418,
+      "step": 4244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088992163220267e-06,
+      "loss": 0.9636,
+      "step": 4245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0845667087201904e-06,
+      "loss": 0.8722,
+      "step": 4246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08014187814942e-06,
+      "loss": 0.9281,
+      "step": 4247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07571767245491e-06,
+      "loss": 0.9779,
+      "step": 4248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071294092583478e-06,
+      "loss": 0.8948,
+      "step": 4249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0668711394818166e-06,
+      "loss": 0.945,
+      "step": 4250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.062448814096474e-06,
+      "loss": 0.928,
+      "step": 4251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058027117373869e-06,
+      "loss": 0.9771,
+      "step": 4252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0536060502602875e-06,
+      "loss": 0.9442,
+      "step": 4253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049185613701875e-06,
+      "loss": 0.9324,
+      "step": 4254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.044765808644645e-06,
+      "loss": 0.3405,
+      "step": 4255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040346636034476e-06,
+      "loss": 0.9025,
+      "step": 4256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0359280968171125e-06,
+      "loss": 0.9594,
+      "step": 4257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.031510191938163e-06,
+      "loss": 0.9132,
+      "step": 4258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027092922343099e-06,
+      "loss": 0.9348,
+      "step": 4259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022676288977257e-06,
+      "loss": 0.9339,
+      "step": 4260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0182602927858346e-06,
+      "loss": 0.3446,
+      "step": 4261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013844934713896e-06,
+      "loss": 0.8753,
+      "step": 4262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.009430215706367e-06,
+      "loss": 0.8937,
+      "step": 4263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005016136708041e-06,
+      "loss": 0.9575,
+      "step": 4264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000602698663569e-06,
+      "loss": 0.9758,
+      "step": 4265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996189902517468e-06,
+      "loss": 0.9766,
+      "step": 4266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991777749214117e-06,
+      "loss": 0.9325,
+      "step": 4267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.987366239697753e-06,
+      "loss": 0.9047,
+      "step": 4268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982955374912483e-06,
+      "loss": 0.9464,
+      "step": 4269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978545155802272e-06,
+      "loss": 0.8806,
+      "step": 4270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974135583310948e-06,
+      "loss": 0.9628,
+      "step": 4271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969726658382197e-06,
+      "loss": 0.9254,
+      "step": 4272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9653183819595694e-06,
+      "loss": 0.9605,
+      "step": 4273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.96091075498648e-06,
+      "loss": 0.9761,
+      "step": 4274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956503778406198e-06,
+      "loss": 0.9056,
+      "step": 4275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952097453161857e-06,
+      "loss": 0.9268,
+      "step": 4276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9476917801964505e-06,
+      "loss": 0.9441,
+      "step": 4277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943286760452834e-06,
+      "loss": 0.9543,
+      "step": 4278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9388823948737205e-06,
+      "loss": 0.9622,
+      "step": 4279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934478684401685e-06,
+      "loss": 0.9103,
+      "step": 4280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.930075629979162e-06,
+      "loss": 0.8965,
+      "step": 4281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9256732325484444e-06,
+      "loss": 0.3965,
+      "step": 4282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921271493051688e-06,
+      "loss": 0.8952,
+      "step": 4283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916870412430903e-06,
+      "loss": 0.9176,
+      "step": 4284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912469991627959e-06,
+      "loss": 1.0129,
+      "step": 4285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.908070231584593e-06,
+      "loss": 0.9251,
+      "step": 4286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903671133242384e-06,
+      "loss": 0.863,
+      "step": 4287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899272697542787e-06,
+      "loss": 0.8894,
+      "step": 4288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894874925427104e-06,
+      "loss": 0.9954,
+      "step": 4289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890477817836497e-06,
+      "loss": 0.9694,
+      "step": 4290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88608137571199e-06,
+      "loss": 0.9321,
+      "step": 4291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881685599994461e-06,
+      "loss": 0.9208,
+      "step": 4292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877290491624648e-06,
+      "loss": 0.3945,
+      "step": 4293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872896051543139e-06,
+      "loss": 0.9478,
+      "step": 4294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868502280690388e-06,
+      "loss": 0.9077,
+      "step": 4295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864109180006701e-06,
+      "loss": 0.924,
+      "step": 4296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8597167504322415e-06,
+      "loss": 0.9504,
+      "step": 4297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855324992907031e-06,
+      "loss": 0.9507,
+      "step": 4298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8509339083709425e-06,
+      "loss": 0.9317,
+      "step": 4299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846543497763714e-06,
+      "loss": 0.9221,
+      "step": 4300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842153762024926e-06,
+      "loss": 1.0075,
+      "step": 4301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837764702094027e-06,
+      "loss": 0.9421,
+      "step": 4302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8333763189103134e-06,
+      "loss": 0.9567,
+      "step": 4303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828988613412939e-06,
+      "loss": 0.9795,
+      "step": 4304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8246015865409155e-06,
+      "loss": 0.9024,
+      "step": 4305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8202152392331054e-06,
+      "loss": 0.3869,
+      "step": 4306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.815829572428228e-06,
+      "loss": 0.855,
+      "step": 4307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811444587064853e-06,
+      "loss": 0.9673,
+      "step": 4308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.807060284081408e-06,
+      "loss": 0.9312,
+      "step": 4309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802676664416175e-06,
+      "loss": 0.9004,
+      "step": 4310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.798293729007291e-06,
+      "loss": 0.911,
+      "step": 4311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793911478792739e-06,
+      "loss": 0.9225,
+      "step": 4312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789529914710362e-06,
+      "loss": 1.0061,
+      "step": 4313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785149037697859e-06,
+      "loss": 0.8962,
+      "step": 4314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780768848692771e-06,
+      "loss": 0.926,
+      "step": 4315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776389348632504e-06,
+      "loss": 0.9088,
+      "step": 4316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7720105384543066e-06,
+      "loss": 0.9441,
+      "step": 4317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767632419095287e-06,
+      "loss": 0.9281,
+      "step": 4318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7632549914924014e-06,
+      "loss": 0.9541,
+      "step": 4319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758878256582459e-06,
+      "loss": 0.9489,
+      "step": 4320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754502215302125e-06,
+      "loss": 0.9093,
+      "step": 4321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7501268685879065e-06,
+      "loss": 0.9135,
+      "step": 4322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7457522173761695e-06,
+      "loss": 0.9422,
+      "step": 4323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.741378262603129e-06,
+      "loss": 0.9136,
+      "step": 4324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737005005204851e-06,
+      "loss": 0.8956,
+      "step": 4325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732632446117255e-06,
+      "loss": 0.9415,
+      "step": 4326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728260586276105e-06,
+      "loss": 0.9177,
+      "step": 4327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723889426617021e-06,
+      "loss": 0.9178,
+      "step": 4328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719518968075473e-06,
+      "loss": 0.9014,
+      "step": 4329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715149211586775e-06,
+      "loss": 0.3642,
+      "step": 4330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7107801580860945e-06,
+      "loss": 0.9132,
+      "step": 4331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706411808508453e-06,
+      "loss": 0.9895,
+      "step": 4332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.702044163788713e-06,
+      "loss": 0.9591,
+      "step": 4333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6976772248615916e-06,
+      "loss": 0.9048,
+      "step": 4334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693310992661655e-06,
+      "loss": 0.9807,
+      "step": 4335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688945468123318e-06,
+      "loss": 0.8426,
+      "step": 4336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684580652180837e-06,
+      "loss": 0.9122,
+      "step": 4337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680216545768328e-06,
+      "loss": 0.9059,
+      "step": 4338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67585314981974e-06,
+      "loss": 0.9598,
+      "step": 4339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.671490465268895e-06,
+      "loss": 0.983,
+      "step": 4340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667128493049437e-06,
+      "loss": 0.9598,
+      "step": 4341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.662767234094869e-06,
+      "loss": 0.9238,
+      "step": 4342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658406689338541e-06,
+      "loss": 0.896,
+      "step": 4343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654046859713651e-06,
+      "loss": 0.8581,
+      "step": 4344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6496877461532395e-06,
+      "loss": 0.9475,
+      "step": 4345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6453293495901975e-06,
+      "loss": 0.9287,
+      "step": 4346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6409716709572645e-06,
+      "loss": 0.8856,
+      "step": 4347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63661471118702e-06,
+      "loss": 0.9239,
+      "step": 4348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632258471211892e-06,
+      "loss": 0.9149,
+      "step": 4349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.627902951964158e-06,
+      "loss": 0.9217,
+      "step": 4350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623548154375938e-06,
+      "loss": 0.9792,
+      "step": 4351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619194079379198e-06,
+      "loss": 0.9199,
+      "step": 4352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6148407279057524e-06,
+      "loss": 1.0066,
+      "step": 4353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610488100887258e-06,
+      "loss": 0.899,
+      "step": 4354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606136199255211e-06,
+      "loss": 0.9392,
+      "step": 4355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6017850239409634e-06,
+      "loss": 0.9762,
+      "step": 4356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.597434575875705e-06,
+      "loss": 0.89,
+      "step": 4357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593084855990471e-06,
+      "loss": 0.9983,
+      "step": 4358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58873586521614e-06,
+      "loss": 0.9416,
+      "step": 4359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584387604483438e-06,
+      "loss": 0.9872,
+      "step": 4360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580040074722934e-06,
+      "loss": 0.9863,
+      "step": 4361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575693276865034e-06,
+      "loss": 0.9971,
+      "step": 4362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.571347211839995e-06,
+      "loss": 0.9535,
+      "step": 4363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5670018805779145e-06,
+      "loss": 0.8602,
+      "step": 4364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562657284008732e-06,
+      "loss": 0.862,
+      "step": 4365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558313423062233e-06,
+      "loss": 0.8905,
+      "step": 4366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553970298668041e-06,
+      "loss": 0.942,
+      "step": 4367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549627911755627e-06,
+      "loss": 0.9488,
+      "step": 4368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545286263254298e-06,
+      "loss": 0.9117,
+      "step": 4369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540945354093209e-06,
+      "loss": 0.9496,
+      "step": 4370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.536605185201352e-06,
+      "loss": 0.9566,
+      "step": 4371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5322657575075645e-06,
+      "loss": 0.9728,
+      "step": 4372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527927071940523e-06,
+      "loss": 0.9374,
+      "step": 4373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523589129428745e-06,
+      "loss": 0.9702,
+      "step": 4374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519251930900591e-06,
+      "loss": 0.9038,
+      "step": 4375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514915477284263e-06,
+      "loss": 0.9783,
+      "step": 4376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510579769507795e-06,
+      "loss": 0.9857,
+      "step": 4377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506244808499074e-06,
+      "loss": 0.9682,
+      "step": 4378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501910595185819e-06,
+      "loss": 0.8512,
+      "step": 4379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4975771304955915e-06,
+      "loss": 0.8718,
+      "step": 4380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493244415355792e-06,
+      "loss": 0.9517,
+      "step": 4381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4889124506936606e-06,
+      "loss": 0.9519,
+      "step": 4382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484581237436282e-06,
+      "loss": 0.8899,
+      "step": 4383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480250776510568e-06,
+      "loss": 0.9438,
+      "step": 4384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4759210688432805e-06,
+      "loss": 0.9435,
+      "step": 4385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.471592115361016e-06,
+      "loss": 0.9105,
+      "step": 4386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467263916990212e-06,
+      "loss": 0.863,
+      "step": 4387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462936474657138e-06,
+      "loss": 0.9329,
+      "step": 4388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45860978928791e-06,
+      "loss": 0.9913,
+      "step": 4389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454283861808478e-06,
+      "loss": 0.982,
+      "step": 4390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.449958693144627e-06,
+      "loss": 0.9337,
+      "step": 4391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445634284221984e-06,
+      "loss": 0.9733,
+      "step": 4392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441310635966012e-06,
+      "loss": 0.968,
+      "step": 4393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43698774930201e-06,
+      "loss": 0.9038,
+      "step": 4394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432665625155119e-06,
+      "loss": 0.8963,
+      "step": 4395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428344264450309e-06,
+      "loss": 0.8771,
+      "step": 4396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42402366811239e-06,
+      "loss": 0.3727,
+      "step": 4397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4197038370660135e-06,
+      "loss": 0.9349,
+      "step": 4398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41538477223566e-06,
+      "loss": 0.9195,
+      "step": 4399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411066474545649e-06,
+      "loss": 0.9644,
+      "step": 4400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406748944920137e-06,
+      "loss": 0.9537,
+      "step": 4401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402432184283112e-06,
+      "loss": 0.883,
+      "step": 4402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398116193558401e-06,
+      "loss": 0.9564,
+      "step": 4403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393800973669666e-06,
+      "loss": 0.9088,
+      "step": 4404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3894865255404025e-06,
+      "loss": 0.8598,
+      "step": 4405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3851728500939434e-06,
+      "loss": 0.9608,
+      "step": 4406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380859948253453e-06,
+      "loss": 0.8535,
+      "step": 4407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3765478209419355e-06,
+      "loss": 0.8728,
+      "step": 4408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372236469082218e-06,
+      "loss": 0.9716,
+      "step": 4409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3679258935969755e-06,
+      "loss": 0.8832,
+      "step": 4410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363616095408707e-06,
+      "loss": 0.9289,
+      "step": 4411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359307075439749e-06,
+      "loss": 0.9728,
+      "step": 4412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3549988346122735e-06,
+      "loss": 0.9263,
+      "step": 4413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3506913738482814e-06,
+      "loss": 0.9151,
+      "step": 4414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3463846940696105e-06,
+      "loss": 0.858,
+      "step": 4415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342078796197926e-06,
+      "loss": 0.8909,
+      "step": 4416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337773681154731e-06,
+      "loss": 0.9744,
+      "step": 4417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333469349861361e-06,
+      "loss": 0.9125,
+      "step": 4418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329165803238979e-06,
+      "loss": 0.9242,
+      "step": 4419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324863042208586e-06,
+      "loss": 0.9436,
+      "step": 4420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.320561067691012e-06,
+      "loss": 0.8941,
+      "step": 4421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316259880606915e-06,
+      "loss": 0.9411,
+      "step": 4422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311959481876795e-06,
+      "loss": 0.905,
+      "step": 4423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30765987242097e-06,
+      "loss": 0.9383,
+      "step": 4424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303361053159599e-06,
+      "loss": 0.9946,
+      "step": 4425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299063025012667e-06,
+      "loss": 0.8832,
+      "step": 4426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294765788899992e-06,
+      "loss": 0.9756,
+      "step": 4427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2904693457412235e-06,
+      "loss": 0.3625,
+      "step": 4428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286173696455835e-06,
+      "loss": 0.9416,
+      "step": 4429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281878841963141e-06,
+      "loss": 0.963,
+      "step": 4430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277584783182273e-06,
+      "loss": 0.92,
+      "step": 4431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273291521032203e-06,
+      "loss": 0.9861,
+      "step": 4432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268999056431726e-06,
+      "loss": 0.9191,
+      "step": 4433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26470739029947e-06,
+      "loss": 0.94,
+      "step": 4434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2604165235538885e-06,
+      "loss": 0.955,
+      "step": 4435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2561264571132685e-06,
+      "loss": 0.9039,
+      "step": 4436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2518371918957246e-06,
+      "loss": 0.9385,
+      "step": 4437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247548728819195e-06,
+      "loss": 0.9747,
+      "step": 4438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2432610688014505e-06,
+      "loss": 0.8991,
+      "step": 4439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238974212760091e-06,
+      "loss": 0.9685,
+      "step": 4440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.234688161612541e-06,
+      "loss": 0.9257,
+      "step": 4441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230402916276057e-06,
+      "loss": 0.9489,
+      "step": 4442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22611847766772e-06,
+      "loss": 0.9433,
+      "step": 4443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22183484670444e-06,
+      "loss": 0.913,
+      "step": 4444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217552024302947e-06,
+      "loss": 0.8485,
+      "step": 4445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2132700113798105e-06,
+      "loss": 0.988,
+      "step": 4446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2089888088514174e-06,
+      "loss": 0.8985,
+      "step": 4447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204708417633986e-06,
+      "loss": 0.8848,
+      "step": 4448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200428838643556e-06,
+      "loss": 0.9469,
+      "step": 4449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196150072795999e-06,
+      "loss": 0.8634,
+      "step": 4450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191872121007012e-06,
+      "loss": 0.9672,
+      "step": 4451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187594984192109e-06,
+      "loss": 0.9508,
+      "step": 4452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183318663266642e-06,
+      "loss": 0.9205,
+      "step": 4453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179043159145779e-06,
+      "loss": 0.9885,
+      "step": 4454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17476847274452e-06,
+      "loss": 0.9449,
+      "step": 4455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170494604977683e-06,
+      "loss": 0.8406,
+      "step": 4456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.166221556759919e-06,
+      "loss": 0.9449,
+      "step": 4457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161949329005696e-06,
+      "loss": 0.9369,
+      "step": 4458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.157677922629314e-06,
+      "loss": 0.9231,
+      "step": 4459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153407338544888e-06,
+      "loss": 0.9084,
+      "step": 4460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149137577666365e-06,
+      "loss": 0.9124,
+      "step": 4461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144868640907513e-06,
+      "loss": 0.9477,
+      "step": 4462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1406005291819205e-06,
+      "loss": 0.9173,
+      "step": 4463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1363332434030035e-06,
+      "loss": 0.9124,
+      "step": 4464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.132066784484001e-06,
+      "loss": 0.9982,
+      "step": 4465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127801153337973e-06,
+      "loss": 0.9563,
+      "step": 4466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1235363508778055e-06,
+      "loss": 0.3524,
+      "step": 4467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1192723780162024e-06,
+      "loss": 0.9273,
+      "step": 4468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115009235665695e-06,
+      "loss": 0.9273,
+      "step": 4469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.110746924738633e-06,
+      "loss": 0.9517,
+      "step": 4470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106485446147191e-06,
+      "loss": 0.9588,
+      "step": 4471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102224800803363e-06,
+      "loss": 0.8974,
+      "step": 4472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0979649896189675e-06,
+      "loss": 0.9427,
+      "step": 4473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.093706013505642e-06,
+      "loss": 0.9573,
+      "step": 4474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089447873374845e-06,
+      "loss": 0.9313,
+      "step": 4475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08519057013786e-06,
+      "loss": 0.9723,
+      "step": 4476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080934104705789e-06,
+      "loss": 0.8783,
+      "step": 4477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.076678477989551e-06,
+      "loss": 0.9331,
+      "step": 4478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072423690899889e-06,
+      "loss": 0.9292,
+      "step": 4479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06816974434737e-06,
+      "loss": 0.9077,
+      "step": 4480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063916639242375e-06,
+      "loss": 0.9767,
+      "step": 4481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059664376495107e-06,
+      "loss": 0.8868,
+      "step": 4482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.05541295701559e-06,
+      "loss": 0.9647,
+      "step": 4483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051162381713666e-06,
+      "loss": 0.9308,
+      "step": 4484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046912651498997e-06,
+      "loss": 0.9892,
+      "step": 4485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042663767281065e-06,
+      "loss": 0.8893,
+      "step": 4486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.038415729969167e-06,
+      "loss": 0.9059,
+      "step": 4487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0341685404724246e-06,
+      "loss": 0.9251,
+      "step": 4488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029922199699774e-06,
+      "loss": 0.9958,
+      "step": 4489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025676708559971e-06,
+      "loss": 0.9739,
+      "step": 4490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0214320679615925e-06,
+      "loss": 1.0213,
+      "step": 4491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0171882788130235e-06,
+      "loss": 0.9668,
+      "step": 4492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012945342022477e-06,
+      "loss": 0.965,
+      "step": 4493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008703258497981e-06,
+      "loss": 0.9458,
+      "step": 4494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0044620291473796e-06,
+      "loss": 0.9866,
+      "step": 4495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000221654878336e-06,
+      "loss": 0.9316,
+      "step": 4496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995982136598324e-06,
+      "loss": 0.9292,
+      "step": 4497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991743475214647e-06,
+      "loss": 0.8957,
+      "step": 4498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9875056716344105e-06,
+      "loss": 0.9577,
+      "step": 4499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983268726764544e-06,
+      "loss": 0.9231,
+      "step": 4500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979032641511795e-06,
+      "loss": 0.9328,
+      "step": 4501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974797416782723e-06,
+      "loss": 0.947,
+      "step": 4502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.970563053483705e-06,
+      "loss": 0.9485,
+      "step": 4503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966329552520931e-06,
+      "loss": 0.9437,
+      "step": 4504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9620969148004145e-06,
+      "loss": 0.9448,
+      "step": 4505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957865141227972e-06,
+      "loss": 0.8999,
+      "step": 4506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.953634232709247e-06,
+      "loss": 0.9859,
+      "step": 4507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949404190149688e-06,
+      "loss": 0.957,
+      "step": 4508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945175014454565e-06,
+      "loss": 0.9585,
+      "step": 4509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94094670652896e-06,
+      "loss": 0.9074,
+      "step": 4510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93671926727777e-06,
+      "loss": 1.0101,
+      "step": 4511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932492697605705e-06,
+      "loss": 0.9639,
+      "step": 4512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9282669984172895e-06,
+      "loss": 0.9249,
+      "step": 4513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9240421706168575e-06,
+      "loss": 0.9534,
+      "step": 4514
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919818215108568e-06,
+      "loss": 0.898,
+      "step": 4515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.915595132796382e-06,
+      "loss": 0.8802,
+      "step": 4516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.911372924584081e-06,
+      "loss": 0.9278,
+      "step": 4517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907151591375251e-06,
+      "loss": 0.9403,
+      "step": 4518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9029311340733e-06,
+      "loss": 0.9359,
+      "step": 4519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89871155358144e-06,
+      "loss": 0.9586,
+      "step": 4520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894492850802704e-06,
+      "loss": 0.9376,
+      "step": 4521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.890275026639929e-06,
+      "loss": 0.9704,
+      "step": 4522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886058081995771e-06,
+      "loss": 0.9052,
+      "step": 4523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881842017772696e-06,
+      "loss": 0.8868,
+      "step": 4524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877626834872976e-06,
+      "loss": 0.8562,
+      "step": 4525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873412534198699e-06,
+      "loss": 0.9551,
+      "step": 4526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869199116651766e-06,
+      "loss": 1.0223,
+      "step": 4527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8649865831338845e-06,
+      "loss": 0.8995,
+      "step": 4528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860774934546578e-06,
+      "loss": 0.9005,
+      "step": 4529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856564171791175e-06,
+      "loss": 0.9359,
+      "step": 4530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8523542957688204e-06,
+      "loss": 0.9302,
+      "step": 4531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848145307380462e-06,
+      "loss": 0.8952,
+      "step": 4532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843937207526864e-06,
+      "loss": 0.9606,
+      "step": 4533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839729997108596e-06,
+      "loss": 0.9385,
+      "step": 4534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8355236770260425e-06,
+      "loss": 0.4046,
+      "step": 4535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831318248179396e-06,
+      "loss": 0.8852,
+      "step": 4536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827113711468648e-06,
+      "loss": 0.9537,
+      "step": 4537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8229100677936165e-06,
+      "loss": 0.9435,
+      "step": 4538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818707318053913e-06,
+      "loss": 0.9551,
+      "step": 4539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.814505463148972e-06,
+      "loss": 0.8675,
+      "step": 4540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.810304503978021e-06,
+      "loss": 0.9479,
+      "step": 4541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80610444144011e-06,
+      "loss": 0.9538,
+      "step": 4542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801905276434088e-06,
+      "loss": 0.9272,
+      "step": 4543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797707009858608e-06,
+      "loss": 0.8431,
+      "step": 4544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7935096426121496e-06,
+      "loss": 0.9556,
+      "step": 4545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789313175592976e-06,
+      "loss": 0.8796,
+      "step": 4546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785117609699181e-06,
+      "loss": 0.9791,
+      "step": 4547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780922945828644e-06,
+      "loss": 0.8284,
+      "step": 4548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.776729184879069e-06,
+      "loss": 0.9366,
+      "step": 4549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7725363277479525e-06,
+      "loss": 0.9192,
+      "step": 4550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768344375332606e-06,
+      "loss": 0.8856,
+      "step": 4551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7641533285301485e-06,
+      "loss": 0.9422,
+      "step": 4552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7599631882374985e-06,
+      "loss": 0.9518,
+      "step": 4553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755773955351388e-06,
+      "loss": 0.8943,
+      "step": 4554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751585630768346e-06,
+      "loss": 0.9573,
+      "step": 4555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747398215384718e-06,
+      "loss": 0.9347,
+      "step": 4556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.743211710096648e-06,
+      "loss": 0.8936,
+      "step": 4557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739026115800079e-06,
+      "loss": 0.9339,
+      "step": 4558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7348414333907765e-06,
+      "loss": 0.9193,
+      "step": 4559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730657663764292e-06,
+      "loss": 0.9283,
+      "step": 4560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7264748078159984e-06,
+      "loss": 0.9023,
+      "step": 4561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722292866441057e-06,
+      "loss": 0.9756,
+      "step": 4562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718111840534451e-06,
+      "loss": 0.8841,
+      "step": 4563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713931730990949e-06,
+      "loss": 0.9615,
+      "step": 4564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709752538705134e-06,
+      "loss": 0.975,
+      "step": 4565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705574264571397e-06,
+      "loss": 0.9707,
+      "step": 4566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701396909483921e-06,
+      "loss": 0.9282,
+      "step": 4567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697220474336701e-06,
+      "loss": 0.8605,
+      "step": 4568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.693044960023528e-06,
+      "loss": 0.9337,
+      "step": 4569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.688870367438009e-06,
+      "loss": 0.9592,
+      "step": 4570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684696697473537e-06,
+      "loss": 0.9219,
+      "step": 4571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680523951023316e-06,
+      "loss": 0.9756,
+      "step": 4572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.676352128980352e-06,
+      "loss": 0.9351,
+      "step": 4573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672181232237459e-06,
+      "loss": 0.98,
+      "step": 4574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668011261687238e-06,
+      "loss": 0.9674,
+      "step": 4575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663842218222111e-06,
+      "loss": 0.9399,
+      "step": 4576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659674102734281e-06,
+      "loss": 0.9179,
+      "step": 4577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6555069161157695e-06,
+      "loss": 0.8478,
+      "step": 4578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651340659258386e-06,
+      "loss": 0.9296,
+      "step": 4579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647175333053757e-06,
+      "loss": 0.9427,
+      "step": 4580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.643010938393294e-06,
+      "loss": 0.9794,
+      "step": 4581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638847476168212e-06,
+      "loss": 0.901,
+      "step": 4582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63468494726954e-06,
+      "loss": 0.9112,
+      "step": 4583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630523352588086e-06,
+      "loss": 0.9439,
+      "step": 4584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626362693014478e-06,
+      "loss": 0.8402,
+      "step": 4585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6222029694391275e-06,
+      "loss": 0.9232,
+      "step": 4586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6180441827522625e-06,
+      "loss": 0.9224,
+      "step": 4587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613886333843892e-06,
+      "loss": 0.9412,
+      "step": 4588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60972942360384e-06,
+      "loss": 0.9712,
+      "step": 4589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6055734529217235e-06,
+      "loss": 0.9414,
+      "step": 4590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601418422686949e-06,
+      "loss": 0.8785,
+      "step": 4591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597264333788743e-06,
+      "loss": 0.9497,
+      "step": 4592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593111187116108e-06,
+      "loss": 0.9452,
+      "step": 4593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588958983557864e-06,
+      "loss": 0.9041,
+      "step": 4594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584807724002613e-06,
+      "loss": 0.9045,
+      "step": 4595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58065740933877e-06,
+      "loss": 0.9517,
+      "step": 4596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576508040454536e-06,
+      "loss": 0.9138,
+      "step": 4597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572359618237908e-06,
+      "loss": 0.8788,
+      "step": 4598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.568212143576698e-06,
+      "loss": 0.9264,
+      "step": 4599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564065617358493e-06,
+      "loss": 0.9184,
+      "step": 4600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559920040470695e-06,
+      "loss": 0.9695,
+      "step": 4601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555775413800488e-06,
+      "loss": 0.881,
+      "step": 4602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551631738234869e-06,
+      "loss": 0.3762,
+      "step": 4603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547489014660616e-06,
+      "loss": 0.9456,
+      "step": 4604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5433472439643075e-06,
+      "loss": 0.9216,
+      "step": 4605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539206427032329e-06,
+      "loss": 0.9422,
+      "step": 4606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535066564750841e-06,
+      "loss": 0.9618,
+      "step": 4607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530927658005825e-06,
+      "loss": 0.3901,
+      "step": 4608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526789707683033e-06,
+      "loss": 0.9258,
+      "step": 4609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522652714668033e-06,
+      "loss": 0.9332,
+      "step": 4610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5185166798461756e-06,
+      "loss": 0.9384,
+      "step": 4611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.514381604102607e-06,
+      "loss": 0.9388,
+      "step": 4612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510247488322277e-06,
+      "loss": 0.9745,
+      "step": 4613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506114333389917e-06,
+      "loss": 0.8951,
+      "step": 4614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501982140190068e-06,
+      "loss": 0.9188,
+      "step": 4615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.49785090960705e-06,
+      "loss": 0.937,
+      "step": 4616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.49372064252499e-06,
+      "loss": 0.8663,
+      "step": 4617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4895913398278e-06,
+      "loss": 0.8745,
+      "step": 4618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.485463002399184e-06,
+      "loss": 0.9316,
+      "step": 4619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481335631122653e-06,
+      "loss": 0.9096,
+      "step": 4620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477209226881491e-06,
+      "loss": 0.9513,
+      "step": 4621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473083790558798e-06,
+      "loss": 0.9669,
+      "step": 4622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468959323037445e-06,
+      "loss": 0.9589,
+      "step": 4623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4648358252001145e-06,
+      "loss": 0.9087,
+      "step": 4624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4607132979292675e-06,
+      "loss": 0.9282,
+      "step": 4625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456591742107158e-06,
+      "loss": 0.8413,
+      "step": 4626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452471158615846e-06,
+      "loss": 0.9217,
+      "step": 4627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448351548337166e-06,
+      "loss": 0.9297,
+      "step": 4628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44423291215276e-06,
+      "loss": 0.3655,
+      "step": 4629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440115250944044e-06,
+      "loss": 0.9071,
+      "step": 4630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.435998565592246e-06,
+      "loss": 0.915,
+      "step": 4631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431882856978363e-06,
+      "loss": 0.8926,
+      "step": 4632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427768125983201e-06,
+      "loss": 0.8742,
+      "step": 4633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423654373487353e-06,
+      "loss": 0.9355,
+      "step": 4634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41954160037119e-06,
+      "loss": 0.9083,
+      "step": 4635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415429807514891e-06,
+      "loss": 0.9175,
+      "step": 4636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4113189957984155e-06,
+      "loss": 0.813,
+      "step": 4637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40720916610151e-06,
+      "loss": 0.8836,
+      "step": 4638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403100319303722e-06,
+      "loss": 0.9366,
+      "step": 4639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398992456284378e-06,
+      "loss": 0.9479,
+      "step": 4640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3948855779226e-06,
+      "loss": 0.9453,
+      "step": 4641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390779685097294e-06,
+      "loss": 0.9114,
+      "step": 4642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386674778687167e-06,
+      "loss": 0.8774,
+      "step": 4643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382570859570697e-06,
+      "loss": 0.8778,
+      "step": 4644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378467928626163e-06,
+      "loss": 0.9253,
+      "step": 4645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.374365986731633e-06,
+      "loss": 0.8492,
+      "step": 4646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370265034764953e-06,
+      "loss": 0.3166,
+      "step": 4647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366165073603773e-06,
+      "loss": 0.9158,
+      "step": 4648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362066104125514e-06,
+      "loss": 0.9081,
+      "step": 4649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3579681272074006e-06,
+      "loss": 0.9913,
+      "step": 4650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353871143726431e-06,
+      "loss": 0.365,
+      "step": 4651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349775154559397e-06,
+      "loss": 0.9032,
+      "step": 4652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345680160582883e-06,
+      "loss": 0.945,
+      "step": 4653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.341586162673248e-06,
+      "loss": 0.9959,
+      "step": 4654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337493161706653e-06,
+      "loss": 0.9543,
+      "step": 4655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333401158559028e-06,
+      "loss": 0.9069,
+      "step": 4656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329310154106109e-06,
+      "loss": 0.9012,
+      "step": 4657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325220149223402e-06,
+      "loss": 0.8838,
+      "step": 4658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.321131144786205e-06,
+      "loss": 0.9492,
+      "step": 4659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317043141669606e-06,
+      "loss": 1.0123,
+      "step": 4660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31295614074847e-06,
+      "loss": 0.9903,
+      "step": 4661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.308870142897457e-06,
+      "loss": 0.9271,
+      "step": 4662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304785148991005e-06,
+      "loss": 0.9485,
+      "step": 4663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300701159903345e-06,
+      "loss": 0.9072,
+      "step": 4664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2966181765084825e-06,
+      "loss": 0.8841,
+      "step": 4665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292536199680212e-06,
+      "loss": 0.8591,
+      "step": 4666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288455230292119e-06,
+      "loss": 1.0,
+      "step": 4667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284375269217562e-06,
+      "loss": 0.3554,
+      "step": 4668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2802963173296984e-06,
+      "loss": 0.9741,
+      "step": 4669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276218375501451e-06,
+      "loss": 0.9175,
+      "step": 4670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2721414446055454e-06,
+      "loss": 0.9651,
+      "step": 4671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268065525514478e-06,
+      "loss": 0.9258,
+      "step": 4672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263990619100531e-06,
+      "loss": 0.8991,
+      "step": 4673
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259916726235777e-06,
+      "loss": 0.9102,
+      "step": 4674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255843847792057e-06,
+      "loss": 0.9387,
+      "step": 4675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2517719846410145e-06,
+      "loss": 0.9158,
+      "step": 4676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247701137654058e-06,
+      "loss": 0.9071,
+      "step": 4677
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2436313077023905e-06,
+      "loss": 0.9236,
+      "step": 4678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2395624956569916e-06,
+      "loss": 0.9014,
+      "step": 4679
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2354947023886184e-06,
+      "loss": 0.9083,
+      "step": 4680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231427928767825e-06,
+      "loss": 0.9338,
+      "step": 4681
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227362175664929e-06,
+      "loss": 0.8655,
+      "step": 4682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22329744395005e-06,
+      "loss": 0.8993,
+      "step": 4683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219233734493064e-06,
+      "loss": 0.9262,
+      "step": 4684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215171048163654e-06,
+      "loss": 0.9448,
+      "step": 4685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2111093858312676e-06,
+      "loss": 0.3689,
+      "step": 4686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.207048748365131e-06,
+      "loss": 0.9178,
+      "step": 4687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202989136634269e-06,
+      "loss": 0.9214,
+      "step": 4688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198930551507464e-06,
+      "loss": 0.9373,
+      "step": 4689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194872993853297e-06,
+      "loss": 0.9416,
+      "step": 4690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1908164645401274e-06,
+      "loss": 0.9227,
+      "step": 4691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1867609644360774e-06,
+      "loss": 0.9218,
+      "step": 4692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182706494409071e-06,
+      "loss": 0.9032,
+      "step": 4693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178653055326794e-06,
+      "loss": 0.9212,
+      "step": 4694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.174600648056727e-06,
+      "loss": 0.9006,
+      "step": 4695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.170549273466115e-06,
+      "loss": 0.9511,
+      "step": 4696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166498932421995e-06,
+      "loss": 0.9659,
+      "step": 4697
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162449625791176e-06,
+      "loss": 0.8433,
+      "step": 4698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15840135444024e-06,
+      "loss": 0.9923,
+      "step": 4699
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154354119235563e-06,
+      "loss": 0.9417,
+      "step": 4700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1503079210432816e-06,
+      "loss": 0.9198,
+      "step": 4701
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146262760729328e-06,
+      "loss": 0.9543,
+      "step": 4702
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1422186391593955e-06,
+      "loss": 0.9709,
+      "step": 4703
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138175557198969e-06,
+      "loss": 0.9329,
+      "step": 4704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134133515713302e-06,
+      "loss": 0.9016,
+      "step": 4705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130092515567426e-06,
+      "loss": 0.9569,
+      "step": 4706
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126052557626158e-06,
+      "loss": 0.9419,
+      "step": 4707
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122013642754079e-06,
+      "loss": 0.9243,
+      "step": 4708
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11797577181556e-06,
+      "loss": 0.9589,
+      "step": 4709
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113938945674735e-06,
+      "loss": 0.9226,
+      "step": 4710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10990316519553e-06,
+      "loss": 0.9248,
+      "step": 4711
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105868431241634e-06,
+      "loss": 0.3534,
+      "step": 4712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101834744676515e-06,
+      "loss": 0.9283,
+      "step": 4713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.097802106363423e-06,
+      "loss": 0.9155,
+      "step": 4714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0937705171653755e-06,
+      "loss": 0.9444,
+      "step": 4715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089739977945173e-06,
+      "loss": 0.886,
+      "step": 4716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.085710489565383e-06,
+      "loss": 0.8952,
+      "step": 4717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081682052888359e-06,
+      "loss": 0.8803,
+      "step": 4718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077654668776222e-06,
+      "loss": 0.9538,
+      "step": 4719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0736283380908605e-06,
+      "loss": 0.9219,
+      "step": 4720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069603061693958e-06,
+      "loss": 0.9501,
+      "step": 4721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0655788404469485e-06,
+      "loss": 0.9205,
+      "step": 4722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061555675211063e-06,
+      "loss": 0.9008,
+      "step": 4723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0575335668472865e-06,
+      "loss": 0.9313,
+      "step": 4724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.053512516216396e-06,
+      "loss": 0.9422,
+      "step": 4725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049492524178927e-06,
+      "loss": 0.8846,
+      "step": 4726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045473591595191e-06,
+      "loss": 0.912,
+      "step": 4727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041455719325287e-06,
+      "loss": 0.8475,
+      "step": 4728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037438908229064e-06,
+      "loss": 0.9274,
+      "step": 4729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0334231591661665e-06,
+      "loss": 0.8964,
+      "step": 4730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029408472995994e-06,
+      "loss": 0.8913,
+      "step": 4731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025394850577733e-06,
+      "loss": 0.9417,
+      "step": 4732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02138229277033e-06,
+      "loss": 0.9731,
+      "step": 4733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.017370800432508e-06,
+      "loss": 0.9147,
+      "step": 4734
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0133603744227675e-06,
+      "loss": 0.9272,
+      "step": 4735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00935101559937e-06,
+      "loss": 0.8771,
+      "step": 4736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0053427248203655e-06,
+      "loss": 0.9318,
+      "step": 4737
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001335502943552e-06,
+      "loss": 0.9361,
+      "step": 4738
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9973293508265216e-06,
+      "loss": 0.9595,
+      "step": 4739
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9933242693266235e-06,
+      "loss": 0.9534,
+      "step": 4740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9893202593009785e-06,
+      "loss": 0.9328,
+      "step": 4741
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9853173216064874e-06,
+      "loss": 0.935,
+      "step": 4742
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981315457099808e-06,
+      "loss": 0.9385,
+      "step": 4743
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977314666637385e-06,
+      "loss": 0.9128,
+      "step": 4744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973314951075414e-06,
+      "loss": 0.9409,
+      "step": 4745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96931631126988e-06,
+      "loss": 0.9929,
+      "step": 4746
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9653187480765256e-06,
+      "loss": 0.9299,
+      "step": 4747
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961322262350859e-06,
+      "loss": 0.97,
+      "step": 4748
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.95732685494817e-06,
+      "loss": 0.9195,
+      "step": 4749
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.953332526723512e-06,
+      "loss": 0.9478,
+      "step": 4750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949339278531712e-06,
+      "loss": 0.9528,
+      "step": 4751
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9453471112273566e-06,
+      "loss": 0.8914,
+      "step": 4752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.941356025664803e-06,
+      "loss": 0.8929,
+      "step": 4753
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9373660226981865e-06,
+      "loss": 0.8952,
+      "step": 4754
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933377103181399e-06,
+      "loss": 0.9478,
+      "step": 4755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92938926796811e-06,
+      "loss": 0.943,
+      "step": 4756
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925402517911748e-06,
+      "loss": 1.0008,
+      "step": 4757
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9214168538655195e-06,
+      "loss": 0.8846,
+      "step": 4758
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917432276682389e-06,
+      "loss": 0.9615,
+      "step": 4759
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913448787215091e-06,
+      "loss": 0.9414,
+      "step": 4760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909466386316133e-06,
+      "loss": 0.9657,
+      "step": 4761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905485074837779e-06,
+      "loss": 0.8784,
+      "step": 4762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901504853632073e-06,
+      "loss": 0.8978,
+      "step": 4763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897525723550812e-06,
+      "loss": 0.9139,
+      "step": 4764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893547685445574e-06,
+      "loss": 0.9134,
+      "step": 4765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88957074016769e-06,
+      "loss": 0.9609,
+      "step": 4766
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88559488856826e-06,
+      "loss": 0.9297,
+      "step": 4767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881620131498159e-06,
+      "loss": 0.8795,
+      "step": 4768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877646469808015e-06,
+      "loss": 0.9169,
+      "step": 4769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.873673904348235e-06,
+      "loss": 0.9592,
+      "step": 4770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869702435968976e-06,
+      "loss": 0.8797,
+      "step": 4771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865732065520177e-06,
+      "loss": 0.9833,
+      "step": 4772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861762793851529e-06,
+      "loss": 0.921,
+      "step": 4773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857794621812487e-06,
+      "loss": 0.9479,
+      "step": 4774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853827550252287e-06,
+      "loss": 0.8879,
+      "step": 4775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8498615800199086e-06,
+      "loss": 0.9173,
+      "step": 4776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845896711964114e-06,
+      "loss": 0.3791,
+      "step": 4777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841932946933413e-06,
+      "loss": 0.9569,
+      "step": 4778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837970285776095e-06,
+      "loss": 0.9652,
+      "step": 4779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834008729340203e-06,
+      "loss": 0.9012,
+      "step": 4780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830048278473542e-06,
+      "loss": 0.9376,
+      "step": 4781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826088934023689e-06,
+      "loss": 0.9611,
+      "step": 4782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822130696837978e-06,
+      "loss": 0.8746,
+      "step": 4783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81817356776351e-06,
+      "loss": 0.9797,
+      "step": 4784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814217547647143e-06,
+      "loss": 0.9093,
+      "step": 4785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810262637335507e-06,
+      "loss": 0.8953,
+      "step": 4786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806308837674986e-06,
+      "loss": 0.9289,
+      "step": 4787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.802356149511723e-06,
+      "loss": 1.0047,
+      "step": 4788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7984045736916406e-06,
+      "loss": 0.951,
+      "step": 4789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7944541110604024e-06,
+      "loss": 0.9619,
+      "step": 4790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79050476246345e-06,
+      "loss": 0.9018,
+      "step": 4791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.786556528745976e-06,
+      "loss": 0.9268,
+      "step": 4792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782609410752942e-06,
+      "loss": 0.9905,
+      "step": 4793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778663409329065e-06,
+      "loss": 0.9703,
+      "step": 4794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774718525318824e-06,
+      "loss": 0.931,
+      "step": 4795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770774759566462e-06,
+      "loss": 0.9003,
+      "step": 4796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.766832112915979e-06,
+      "loss": 0.9388,
+      "step": 4797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762890586211142e-06,
+      "loss": 0.9738,
+      "step": 4798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758950180295465e-06,
+      "loss": 0.8711,
+      "step": 4799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.755010896012242e-06,
+      "loss": 0.932,
+      "step": 4800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751072734204509e-06,
+      "loss": 0.9181,
+      "step": 4801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747135695715066e-06,
+      "loss": 0.9396,
+      "step": 4802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743199781386483e-06,
+      "loss": 0.9746,
+      "step": 4803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7392649920610725e-06,
+      "loss": 0.9598,
+      "step": 4804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7353313285809235e-06,
+      "loss": 0.9138,
+      "step": 4805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731398791787869e-06,
+      "loss": 0.9158,
+      "step": 4806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727467382523511e-06,
+      "loss": 0.9593,
+      "step": 4807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723537101629211e-06,
+      "loss": 0.8937,
+      "step": 4808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719607949946077e-06,
+      "loss": 0.9536,
+      "step": 4809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71567992831499e-06,
+      "loss": 0.8695,
+      "step": 4810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711753037576576e-06,
+      "loss": 0.9358,
+      "step": 4811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707827278571234e-06,
+      "loss": 0.9143,
+      "step": 4812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.703902652139107e-06,
+      "loss": 0.9183,
+      "step": 4813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.699979159120097e-06,
+      "loss": 0.9105,
+      "step": 4814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696056800353875e-06,
+      "loss": 0.8835,
+      "step": 4815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692135576679855e-06,
+      "loss": 0.8799,
+      "step": 4816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68821548893722e-06,
+      "loss": 0.8608,
+      "step": 4817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6842965379649e-06,
+      "loss": 0.8849,
+      "step": 4818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680378724601591e-06,
+      "loss": 0.8935,
+      "step": 4819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6764620496857374e-06,
+      "loss": 0.978,
+      "step": 4820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672546514055541e-06,
+      "loss": 0.9064,
+      "step": 4821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668632118548969e-06,
+      "loss": 0.9294,
+      "step": 4822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66471886400373e-06,
+      "loss": 0.9311,
+      "step": 4823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660806751257304e-06,
+      "loss": 0.9049,
+      "step": 4824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656895781146912e-06,
+      "loss": 0.9369,
+      "step": 4825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652985954509543e-06,
+      "loss": 0.9034,
+      "step": 4826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649077272181932e-06,
+      "loss": 0.9668,
+      "step": 4827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64516973500057e-06,
+      "loss": 0.9423,
+      "step": 4828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.641263343801713e-06,
+      "loss": 0.9012,
+      "step": 4829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637358099421356e-06,
+      "loss": 0.927,
+      "step": 4830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6334540026952645e-06,
+      "loss": 0.8515,
+      "step": 4831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629551054458945e-06,
+      "loss": 0.8752,
+      "step": 4832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625649255547669e-06,
+      "loss": 0.9116,
+      "step": 4833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621748606796455e-06,
+      "loss": 0.9267,
+      "step": 4834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617849109040073e-06,
+      "loss": 0.9638,
+      "step": 4835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613950763113059e-06,
+      "loss": 0.9087,
+      "step": 4836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610053569849687e-06,
+      "loss": 0.9205,
+      "step": 4837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6061575300839985e-06,
+      "loss": 0.9457,
+      "step": 4838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6022626446497755e-06,
+      "loss": 0.8737,
+      "step": 4839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598368914380565e-06,
+      "loss": 0.9631,
+      "step": 4840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5944763401096585e-06,
+      "loss": 0.9441,
+      "step": 4841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590584922670096e-06,
+      "loss": 0.8653,
+      "step": 4842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.586694662894687e-06,
+      "loss": 0.9594,
+      "step": 4843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582805561615974e-06,
+      "loss": 0.9088,
+      "step": 4844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.578917619666268e-06,
+      "loss": 0.3267,
+      "step": 4845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575030837877614e-06,
+      "loss": 0.9398,
+      "step": 4846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5711452170818295e-06,
+      "loss": 0.3571,
+      "step": 4847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567260758110466e-06,
+      "loss": 0.9296,
+      "step": 4848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563377461794833e-06,
+      "loss": 0.9167,
+      "step": 4849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559495328965997e-06,
+      "loss": 0.9295,
+      "step": 4850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555614360454761e-06,
+      "loss": 0.9506,
+      "step": 4851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551734557091697e-06,
+      "loss": 0.9824,
+      "step": 4852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547855919707109e-06,
+      "loss": 0.9007,
+      "step": 4853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543978449131071e-06,
+      "loss": 0.9461,
+      "step": 4854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540102146193391e-06,
+      "loss": 0.8508,
+      "step": 4855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536227011723631e-06,
+      "loss": 0.9092,
+      "step": 4856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.53235304655111e-06,
+      "loss": 0.9579,
+      "step": 4857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528480251504888e-06,
+      "loss": 0.8962,
+      "step": 4858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524608627413782e-06,
+      "loss": 0.9544,
+      "step": 4859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520738175106351e-06,
+      "loss": 0.9392,
+      "step": 4860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516868895410912e-06,
+      "loss": 0.9224,
+      "step": 4861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513000789155524e-06,
+      "loss": 0.8999,
+      "step": 4862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509133857167993e-06,
+      "loss": 0.9033,
+      "step": 4863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505268100275884e-06,
+      "loss": 0.9515,
+      "step": 4864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501403519306498e-06,
+      "loss": 0.9571,
+      "step": 4865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497540115086895e-06,
+      "loss": 0.8723,
+      "step": 4866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493677888443879e-06,
+      "loss": 0.9184,
+      "step": 4867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4898168402039984e-06,
+      "loss": 0.8797,
+      "step": 4868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485956971193558e-06,
+      "loss": 0.9119,
+      "step": 4869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4820982822385975e-06,
+      "loss": 0.9721,
+      "step": 4870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4782407741649195e-06,
+      "loss": 0.9164,
+      "step": 4871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474384447798058e-06,
+      "loss": 0.9816,
+      "step": 4872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47052930396331e-06,
+      "loss": 0.9062,
+      "step": 4873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466675343485708e-06,
+      "loss": 0.9336,
+      "step": 4874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46282256719003e-06,
+      "loss": 0.9351,
+      "step": 4875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458970975900814e-06,
+      "loss": 0.8704,
+      "step": 4876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455120570442326e-06,
+      "loss": 0.9299,
+      "step": 4877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.451271351638598e-06,
+      "loss": 0.9879,
+      "step": 4878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.447423320313388e-06,
+      "loss": 0.9082,
+      "step": 4879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443576477290218e-06,
+      "loss": 0.8433,
+      "step": 4880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439730823392345e-06,
+      "loss": 0.9412,
+      "step": 4881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435886359442769e-06,
+      "loss": 0.9214,
+      "step": 4882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432043086264248e-06,
+      "loss": 0.8926,
+      "step": 4883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428201004679271e-06,
+      "loss": 0.8716,
+      "step": 4884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4243601155100856e-06,
+      "loss": 0.9652,
+      "step": 4885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420520419578667e-06,
+      "loss": 0.9363,
+      "step": 4886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4166819177067575e-06,
+      "loss": 0.9134,
+      "step": 4887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412844610715824e-06,
+      "loss": 0.9122,
+      "step": 4888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.409008499427082e-06,
+      "loss": 0.8957,
+      "step": 4889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405173584661503e-06,
+      "loss": 0.9403,
+      "step": 4890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401339867239785e-06,
+      "loss": 0.8945,
+      "step": 4891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3975073479823874e-06,
+      "loss": 0.3482,
+      "step": 4892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3936760277094955e-06,
+      "loss": 0.9166,
+      "step": 4893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389845907241055e-06,
+      "loss": 0.3625,
+      "step": 4894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386016987396742e-06,
+      "loss": 0.9392,
+      "step": 4895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382189268995977e-06,
+      "loss": 0.8605,
+      "step": 4896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378362752857934e-06,
+      "loss": 0.9838,
+      "step": 4897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374537439801515e-06,
+      "loss": 0.9535,
+      "step": 4898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.37071333064538e-06,
+      "loss": 0.9535,
+      "step": 4899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366890426207915e-06,
+      "loss": 0.8781,
+      "step": 4900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363068727307263e-06,
+      "loss": 0.9104,
+      "step": 4901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359248234761302e-06,
+      "loss": 0.9165,
+      "step": 4902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355428949387644e-06,
+      "loss": 0.9572,
+      "step": 4903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351610872003661e-06,
+      "loss": 0.9174,
+      "step": 4904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3477940034264485e-06,
+      "loss": 0.964,
+      "step": 4905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343978344472859e-06,
+      "loss": 0.967,
+      "step": 4906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.34016389595947e-06,
+      "loss": 0.9355,
+      "step": 4907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336350658702617e-06,
+      "loss": 0.9443,
+      "step": 4908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332538633518358e-06,
+      "loss": 0.9658,
+      "step": 4909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328727821222511e-06,
+      "loss": 0.9216,
+      "step": 4910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3249182226306195e-06,
+      "loss": 0.9185,
+      "step": 4911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321109838557969e-06,
+      "loss": 0.8863,
+      "step": 4912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317302669819594e-06,
+      "loss": 0.933,
+      "step": 4913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313496717230259e-06,
+      "loss": 0.905,
+      "step": 4914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309691981604478e-06,
+      "loss": 0.8702,
+      "step": 4915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3058884637564935e-06,
+      "loss": 0.9512,
+      "step": 4916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3020861645002975e-06,
+      "loss": 0.916,
+      "step": 4917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298285084649615e-06,
+      "loss": 0.8798,
+      "step": 4918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.294485225017908e-06,
+      "loss": 0.9822,
+      "step": 4919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290686586418389e-06,
+      "loss": 0.898,
+      "step": 4920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286889169663991e-06,
+      "loss": 0.9041,
+      "step": 4921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283092975567408e-06,
+      "loss": 0.9437,
+      "step": 4922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279298004941048e-06,
+      "loss": 0.8918,
+      "step": 4923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275504258597075e-06,
+      "loss": 0.9732,
+      "step": 4924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.27171173734739e-06,
+      "loss": 0.876,
+      "step": 4925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267920442003619e-06,
+      "loss": 0.929,
+      "step": 4926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26413037337714e-06,
+      "loss": 0.9136,
+      "step": 4927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.260341532279061e-06,
+      "loss": 0.9108,
+      "step": 4928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2565539195202225e-06,
+      "loss": 0.9037,
+      "step": 4929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252767535911215e-06,
+      "loss": 0.9013,
+      "step": 4930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248982382262356e-06,
+      "loss": 0.9453,
+      "step": 4931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245198459383705e-06,
+      "loss": 1.0015,
+      "step": 4932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2414157680850525e-06,
+      "loss": 0.9567,
+      "step": 4933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237634309175934e-06,
+      "loss": 0.9779,
+      "step": 4934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233854083465614e-06,
+      "loss": 0.8828,
+      "step": 4935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23007509176309e-06,
+      "loss": 0.346,
+      "step": 4936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22629733487711e-06,
+      "loss": 0.8984,
+      "step": 4937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222520813616139e-06,
+      "loss": 0.9545,
+      "step": 4938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218745528788396e-06,
+      "loss": 0.9257,
+      "step": 4939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214971481201818e-06,
+      "loss": 0.3745,
+      "step": 4940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211198671664094e-06,
+      "loss": 0.9186,
+      "step": 4941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.207427100982635e-06,
+      "loss": 0.9413,
+      "step": 4942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203656769964588e-06,
+      "loss": 0.3788,
+      "step": 4943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199887679416846e-06,
+      "loss": 0.9142,
+      "step": 4944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196119830146021e-06,
+      "loss": 0.9135,
+      "step": 4945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192353222958475e-06,
+      "loss": 0.8572,
+      "step": 4946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188587858660288e-06,
+      "loss": 0.9345,
+      "step": 4947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1848237380572896e-06,
+      "loss": 0.9324,
+      "step": 4948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181060861955034e-06,
+      "loss": 0.9033,
+      "step": 4949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177299231158806e-06,
+      "loss": 0.9471,
+      "step": 4950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.173538846473635e-06,
+      "loss": 0.908,
+      "step": 4951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169779708704272e-06,
+      "loss": 0.9279,
+      "step": 4952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1660218186552145e-06,
+      "loss": 0.923,
+      "step": 4953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162265177130676e-06,
+      "loss": 0.9108,
+      "step": 4954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158509784934621e-06,
+      "loss": 0.9657,
+      "step": 4955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154755642870731e-06,
+      "loss": 0.8704,
+      "step": 4956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151002751742431e-06,
+      "loss": 0.9789,
+      "step": 4957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147251112352873e-06,
+      "loss": 0.9597,
+      "step": 4958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143500725504936e-06,
+      "loss": 0.946,
+      "step": 4959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139751592001248e-06,
+      "loss": 0.8867,
+      "step": 4960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.136003712644146e-06,
+      "loss": 0.8609,
+      "step": 4961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132257088235721e-06,
+      "loss": 0.953,
+      "step": 4962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128511719577777e-06,
+      "loss": 0.9065,
+      "step": 4963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124767607471862e-06,
+      "loss": 0.9162,
+      "step": 4964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1210247527192496e-06,
+      "loss": 0.8692,
+      "step": 4965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117283156120938e-06,
+      "loss": 0.9296,
+      "step": 4966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113542818477673e-06,
+      "loss": 0.3332,
+      "step": 4967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109803740589912e-06,
+      "loss": 0.885,
+      "step": 4968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10606592325786e-06,
+      "loss": 0.8686,
+      "step": 4969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102329367281437e-06,
+      "loss": 0.9366,
+      "step": 4970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0985940734603065e-06,
+      "loss": 0.9507,
+      "step": 4971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094860042593852e-06,
+      "loss": 0.9106,
+      "step": 4972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091127275481186e-06,
+      "loss": 0.8998,
+      "step": 4973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087395772921163e-06,
+      "loss": 0.9747,
+      "step": 4974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083665535712351e-06,
+      "loss": 0.9509,
+      "step": 4975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0799365646530605e-06,
+      "loss": 0.9431,
+      "step": 4976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0762088605413205e-06,
+      "loss": 0.8861,
+      "step": 4977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072482424174901e-06,
+      "loss": 0.9396,
+      "step": 4978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0687572563512864e-06,
+      "loss": 0.8883,
+      "step": 4979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065033357867696e-06,
+      "loss": 0.907,
+      "step": 4980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061310729521085e-06,
+      "loss": 0.9491,
+      "step": 4981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057589372108122e-06,
+      "loss": 1.053,
+      "step": 4982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053869286425214e-06,
+      "loss": 0.9148,
+      "step": 4983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0501504732684984e-06,
+      "loss": 0.891,
+      "step": 4984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046432933433827e-06,
+      "loss": 0.8619,
+      "step": 4985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042716667716795e-06,
+      "loss": 0.899,
+      "step": 4986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03900167691271e-06,
+      "loss": 0.911,
+      "step": 4987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035287961816619e-06,
+      "loss": 0.9647,
+      "step": 4988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031575523223291e-06,
+      "loss": 0.9638,
+      "step": 4989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027864361927216e-06,
+      "loss": 0.864,
+      "step": 4990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024154478722624e-06,
+      "loss": 0.9008,
+      "step": 4991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.020445874403456e-06,
+      "loss": 0.9758,
+      "step": 4992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016738549763395e-06,
+      "loss": 0.9643,
+      "step": 4993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013032505595835e-06,
+      "loss": 0.8903,
+      "step": 4994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0093277426939105e-06,
+      "loss": 0.9234,
+      "step": 4995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005624261850469e-06,
+      "loss": 0.9289,
+      "step": 4996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001922063858089e-06,
+      "loss": 0.8496,
+      "step": 4997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998221149509079e-06,
+      "loss": 0.9212,
+      "step": 4998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994521519595464e-06,
+      "loss": 0.9283,
+      "step": 4999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990823174909003e-06,
+      "loss": 0.9288,
+      "step": 5000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.987126116241169e-06,
+      "loss": 0.8858,
+      "step": 5001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983430344383175e-06,
+      "loss": 0.8358,
+      "step": 5002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97973586012594e-06,
+      "loss": 0.8977,
+      "step": 5003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976042664260127e-06,
+      "loss": 0.9381,
+      "step": 5004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972350757576108e-06,
+      "loss": 0.8884,
+      "step": 5005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968660140863983e-06,
+      "loss": 0.9653,
+      "step": 5006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964970814913584e-06,
+      "loss": 0.9656,
+      "step": 5007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9612827805144525e-06,
+      "loss": 0.878,
+      "step": 5008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95759603845587e-06,
+      "loss": 0.8554,
+      "step": 5009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953910589526824e-06,
+      "loss": 0.9601,
+      "step": 5010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950226434516044e-06,
+      "loss": 0.889,
+      "step": 5011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946543574211965e-06,
+      "loss": 0.9372,
+      "step": 5012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942862009402754e-06,
+      "loss": 0.909,
+      "step": 5013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939181740876304e-06,
+      "loss": 0.887,
+      "step": 5014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935502769420217e-06,
+      "loss": 0.9269,
+      "step": 5015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931825095821837e-06,
+      "loss": 0.9261,
+      "step": 5016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928148720868212e-06,
+      "loss": 0.9034,
+      "step": 5017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924473645346125e-06,
+      "loss": 0.9506,
+      "step": 5018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920799870042075e-06,
+      "loss": 0.9364,
+      "step": 5019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917127395742276e-06,
+      "loss": 0.9251,
+      "step": 5020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913456223232681e-06,
+      "loss": 0.9227,
+      "step": 5021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909786353298947e-06,
+      "loss": 0.9056,
+      "step": 5022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.906117786726465e-06,
+      "loss": 0.9412,
+      "step": 5023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9024505243003376e-06,
+      "loss": 0.9087,
+      "step": 5024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898784566805396e-06,
+      "loss": 0.9435,
+      "step": 5025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895119915026187e-06,
+      "loss": 0.912,
+      "step": 5026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891456569746977e-06,
+      "loss": 0.9029,
+      "step": 5027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8877945317517596e-06,
+      "loss": 0.951,
+      "step": 5028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884133801824239e-06,
+      "loss": 0.9501,
+      "step": 5029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880474380747851e-06,
+      "loss": 0.9171,
+      "step": 5030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876816269305739e-06,
+      "loss": 0.9172,
+      "step": 5031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873159468280779e-06,
+      "loss": 0.9633,
+      "step": 5032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869503978455555e-06,
+      "loss": 0.9695,
+      "step": 5033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865849800612373e-06,
+      "loss": 0.9646,
+      "step": 5034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862196935533268e-06,
+      "loss": 0.9902,
+      "step": 5035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858545383999979e-06,
+      "loss": 0.9067,
+      "step": 5036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854895146793976e-06,
+      "loss": 0.9101,
+      "step": 5037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851246224696438e-06,
+      "loss": 0.9424,
+      "step": 5038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847598618488276e-06,
+      "loss": 0.8948,
+      "step": 5039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843952328950103e-06,
+      "loss": 0.9548,
+      "step": 5040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840307356862257e-06,
+      "loss": 0.9669,
+      "step": 5041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8366637030048046e-06,
+      "loss": 0.9256,
+      "step": 5042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.833021368157517e-06,
+      "loss": 0.9249,
+      "step": 5043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829380353099882e-06,
+      "loss": 0.8689,
+      "step": 5044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825740658611117e-06,
+      "loss": 0.9555,
+      "step": 5045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822102285470143e-06,
+      "loss": 0.9381,
+      "step": 5046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818465234455612e-06,
+      "loss": 0.9205,
+      "step": 5047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8148295063458795e-06,
+      "loss": 0.923,
+      "step": 5048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811195101919032e-06,
+      "loss": 0.8982,
+      "step": 5049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8075620219528565e-06,
+      "loss": 0.8993,
+      "step": 5050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8039302672248723e-06,
+      "loss": 1.0013,
+      "step": 5051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800299838512307e-06,
+      "loss": 0.9218,
+      "step": 5052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7966707365920995e-06,
+      "loss": 0.9569,
+      "step": 5053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7930429622409183e-06,
+      "loss": 0.9003,
+      "step": 5054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7894165162351337e-06,
+      "loss": 0.3772,
+      "step": 5055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7857913993508445e-06,
+      "loss": 0.9296,
+      "step": 5056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782167612363853e-06,
+      "loss": 0.9722,
+      "step": 5057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7785451560496876e-06,
+      "loss": 0.9252,
+      "step": 5058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7749240311835856e-06,
+      "loss": 0.8541,
+      "step": 5059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7713042385404965e-06,
+      "loss": 0.9398,
+      "step": 5060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767685778895095e-06,
+      "loss": 0.8974,
+      "step": 5061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7640686530217584e-06,
+      "loss": 0.9403,
+      "step": 5062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7604528616945914e-06,
+      "loss": 0.9269,
+      "step": 5063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7568384056873995e-06,
+      "loss": 0.8906,
+      "step": 5064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.753225285773716e-06,
+      "loss": 0.9004,
+      "step": 5065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749613502726779e-06,
+      "loss": 0.9589,
+      "step": 5066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.746003057319537e-06,
+      "loss": 0.9339,
+      "step": 5067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742393950324669e-06,
+      "loss": 0.9807,
+      "step": 5068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7387861825145456e-06,
+      "loss": 0.904,
+      "step": 5069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7351797546612713e-06,
+      "loss": 0.9623,
+      "step": 5070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7315746675366484e-06,
+      "loss": 0.9003,
+      "step": 5071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.727970921912203e-06,
+      "loss": 0.9035,
+      "step": 5072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724368518559168e-06,
+      "loss": 0.8988,
+      "step": 5073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7207674582484864e-06,
+      "loss": 0.8458,
+      "step": 5074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7171677417508243e-06,
+      "loss": 0.9412,
+      "step": 5075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7135693698365484e-06,
+      "loss": 0.9384,
+      "step": 5076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7099723432757484e-06,
+      "loss": 0.8981,
+      "step": 5077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7063766628382145e-06,
+      "loss": 0.8931,
+      "step": 5078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7027823292934627e-06,
+      "loss": 0.9312,
+      "step": 5079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6991893434107094e-06,
+      "loss": 0.8859,
+      "step": 5080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6955977059588822e-06,
+      "loss": 0.9261,
+      "step": 5081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692007417706631e-06,
+      "loss": 0.9177,
+      "step": 5082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6884184794223056e-06,
+      "loss": 0.9627,
+      "step": 5083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.684830891873975e-06,
+      "loss": 0.8892,
+      "step": 5084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.681244655829411e-06,
+      "loss": 0.9484,
+      "step": 5085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6776597720561068e-06,
+      "loss": 0.8699,
+      "step": 5086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6740762413212574e-06,
+      "loss": 0.8962,
+      "step": 5087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6704940643917675e-06,
+      "loss": 0.8847,
+      "step": 5088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6669132420342624e-06,
+      "loss": 0.9257,
+      "step": 5089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.663333775015063e-06,
+      "loss": 0.938,
+      "step": 5090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6597556641002165e-06,
+      "loss": 0.9242,
+      "step": 5091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6561789100554622e-06,
+      "loss": 0.9844,
+      "step": 5092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.652603513646268e-06,
+      "loss": 0.8985,
+      "step": 5093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6490294756377942e-06,
+      "loss": 0.9505,
+      "step": 5094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645456796794917e-06,
+      "loss": 0.9505,
+      "step": 5095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6418854778822275e-06,
+      "loss": 0.9219,
+      "step": 5096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.638315519664015e-06,
+      "loss": 0.3773,
+      "step": 5097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.634746922904289e-06,
+      "loss": 0.8703,
+      "step": 5098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6311796883667548e-06,
+      "loss": 0.9086,
+      "step": 5099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6276138168148367e-06,
+      "loss": 0.9152,
+      "step": 5100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.624049309011667e-06,
+      "loss": 0.9138,
+      "step": 5101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6204861657200775e-06,
+      "loss": 0.9009,
+      "step": 5102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.616924387702618e-06,
+      "loss": 0.9274,
+      "step": 5103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6133639757215356e-06,
+      "loss": 0.9259,
+      "step": 5104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609804930538798e-06,
+      "loss": 0.8883,
+      "step": 5105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6062472529160697e-06,
+      "loss": 0.9503,
+      "step": 5106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.602690943614723e-06,
+      "loss": 0.9381,
+      "step": 5107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5991360033958467e-06,
+      "loss": 0.995,
+      "step": 5108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595582433020224e-06,
+      "loss": 0.3337,
+      "step": 5109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.592030233248358e-06,
+      "loss": 0.9163,
+      "step": 5110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588479404840446e-06,
+      "loss": 0.8691,
+      "step": 5111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.584929948556403e-06,
+      "loss": 0.8785,
+      "step": 5112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5813818651558417e-06,
+      "loss": 0.9157,
+      "step": 5113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5778351553980818e-06,
+      "loss": 0.9303,
+      "step": 5114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5742898200421583e-06,
+      "loss": 0.8575,
+      "step": 5115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5707458598467983e-06,
+      "loss": 0.8959,
+      "step": 5116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.567203275570448e-06,
+      "loss": 1.0107,
+      "step": 5117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5636620679712453e-06,
+      "loss": 0.3537,
+      "step": 5118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5601222378070497e-06,
+      "loss": 0.9625,
+      "step": 5119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5565837858354124e-06,
+      "loss": 0.9379,
+      "step": 5120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5530467128135914e-06,
+      "loss": 0.8566,
+      "step": 5121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5495110194985584e-06,
+      "loss": 0.9439,
+      "step": 5122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5459767066469786e-06,
+      "loss": 0.9637,
+      "step": 5123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542443775015234e-06,
+      "loss": 0.934,
+      "step": 5124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5389122253593955e-06,
+      "loss": 0.898,
+      "step": 5125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.535382058435256e-06,
+      "loss": 0.908,
+      "step": 5126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531853274998298e-06,
+      "loss": 0.9343,
+      "step": 5127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.528325875803712e-06,
+      "loss": 0.8832,
+      "step": 5128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5247998616063984e-06,
+      "loss": 0.9037,
+      "step": 5129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5212752331609514e-06,
+      "loss": 0.8733,
+      "step": 5130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5177519912216783e-06,
+      "loss": 0.8702,
+      "step": 5131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5142301365425803e-06,
+      "loss": 0.9385,
+      "step": 5132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5107096698773723e-06,
+      "loss": 0.9088,
+      "step": 5133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5071905919794622e-06,
+      "loss": 0.9452,
+      "step": 5134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5036729036019623e-06,
+      "loss": 0.9133,
+      "step": 5135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5001566054976966e-06,
+      "loss": 0.9324,
+      "step": 5136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4966416984191788e-06,
+      "loss": 0.9709,
+      "step": 5137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493128183118637e-06,
+      "loss": 0.9691,
+      "step": 5138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4896160603479876e-06,
+      "loss": 0.8732,
+      "step": 5139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4861053308588664e-06,
+      "loss": 0.3629,
+      "step": 5140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4825959954025955e-06,
+      "loss": 0.9211,
+      "step": 5141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479088054730204e-06,
+      "loss": 0.8803,
+      "step": 5142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.475581509592427e-06,
+      "loss": 0.9437,
+      "step": 5143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4720763607396923e-06,
+      "loss": 0.9997,
+      "step": 5144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4685726089221405e-06,
+      "loss": 0.9277,
+      "step": 5145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465070254889599e-06,
+      "loss": 0.9769,
+      "step": 5146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4615692993916093e-06,
+      "loss": 0.9238,
+      "step": 5147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458069743177407e-06,
+      "loss": 0.9067,
+      "step": 5148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4545715869959227e-06,
+      "loss": 0.8845,
+      "step": 5149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.451074831595803e-06,
+      "loss": 0.3666,
+      "step": 5150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.447579477725378e-06,
+      "loss": 1.0245,
+      "step": 5151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444085526132691e-06,
+      "loss": 0.8681,
+      "step": 5152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4405929775654732e-06,
+      "loss": 0.912,
+      "step": 5153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.43710183277117e-06,
+      "loss": 0.9536,
+      "step": 5154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433612092496913e-06,
+      "loss": 0.8908,
+      "step": 5155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.430123757489535e-06,
+      "loss": 0.9527,
+      "step": 5156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4266368284955796e-06,
+      "loss": 0.8714,
+      "step": 5157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.423151306261271e-06,
+      "loss": 0.9139,
+      "step": 5158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4196671915325565e-06,
+      "loss": 0.9502,
+      "step": 5159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4161844850550596e-06,
+      "loss": 0.9608,
+      "step": 5160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4127031875741102e-06,
+      "loss": 0.9208,
+      "step": 5161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409223299834743e-06,
+      "loss": 0.91,
+      "step": 5162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4057448225816804e-06,
+      "loss": 0.9004,
+      "step": 5163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.402267756559353e-06,
+      "loss": 0.9773,
+      "step": 5164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3987921025118797e-06,
+      "loss": 0.8943,
+      "step": 5165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3953178611830885e-06,
+      "loss": 0.9669,
+      "step": 5166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3918450333164943e-06,
+      "loss": 0.9244,
+      "step": 5167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.388373619655312e-06,
+      "loss": 0.8937,
+      "step": 5168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3849036209424614e-06,
+      "loss": 0.9207,
+      "step": 5169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3814350379205472e-06,
+      "loss": 0.885,
+      "step": 5170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3779678713318863e-06,
+      "loss": 0.8407,
+      "step": 5171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3745021219184747e-06,
+      "loss": 0.918,
+      "step": 5172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.371037790422023e-06,
+      "loss": 0.8919,
+      "step": 5173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3675748775839246e-06,
+      "loss": 0.883,
+      "step": 5174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.364113384145272e-06,
+      "loss": 0.9095,
+      "step": 5175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.360653310846864e-06,
+      "loss": 0.9403,
+      "step": 5176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3571946584291803e-06,
+      "loss": 0.8995,
+      "step": 5177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.353737427632411e-06,
+      "loss": 0.9384,
+      "step": 5178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.350281619196427e-06,
+      "loss": 0.9536,
+      "step": 5179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.346827233860811e-06,
+      "loss": 0.9223,
+      "step": 5180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.343374272364829e-06,
+      "loss": 0.9346,
+      "step": 5181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.339922735447443e-06,
+      "loss": 0.8981,
+      "step": 5182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33647262384732e-06,
+      "loss": 0.3442,
+      "step": 5183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3330239383028095e-06,
+      "loss": 0.9539,
+      "step": 5184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.329576679551967e-06,
+      "loss": 0.8934,
+      "step": 5185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3261308483325315e-06,
+      "loss": 0.8805,
+      "step": 5186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3226864453819486e-06,
+      "loss": 0.8679,
+      "step": 5187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.319243471437348e-06,
+      "loss": 0.8868,
+      "step": 5188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3158019272355556e-06,
+      "loss": 0.929,
+      "step": 5189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3123618135130996e-06,
+      "loss": 0.9719,
+      "step": 5190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3089231310061875e-06,
+      "loss": 0.9417,
+      "step": 5191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.305485880450737e-06,
+      "loss": 0.9399,
+      "step": 5192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.302050062582345e-06,
+      "loss": 0.8895,
+      "step": 5193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.298615678136312e-06,
+      "loss": 0.9497,
+      "step": 5194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295182727847626e-06,
+      "loss": 0.9069,
+      "step": 5195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2917512124509655e-06,
+      "loss": 0.8914,
+      "step": 5196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288321132680714e-06,
+      "loss": 0.9028,
+      "step": 5197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2848924892709322e-06,
+      "loss": 0.958,
+      "step": 5198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2814652829553883e-06,
+      "loss": 0.9373,
+      "step": 5199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278039514467528e-06,
+      "loss": 0.8908,
+      "step": 5200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.274615184540505e-06,
+      "loss": 0.9691,
+      "step": 5201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2711922939071515e-06,
+      "loss": 0.9129,
+      "step": 5202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2677708432999976e-06,
+      "loss": 0.9501,
+      "step": 5203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2643508334512685e-06,
+      "loss": 0.3419,
+      "step": 5204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.260932265092872e-06,
+      "loss": 0.886,
+      "step": 5205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.25751513895642e-06,
+      "loss": 0.9308,
+      "step": 5206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254099455773201e-06,
+      "loss": 0.914,
+      "step": 5207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2506852162742098e-06,
+      "loss": 0.9207,
+      "step": 5208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247272421190121e-06,
+      "loss": 0.9346,
+      "step": 5209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2438610712513006e-06,
+      "loss": 0.8707,
+      "step": 5210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.240451167187815e-06,
+      "loss": 0.925,
+      "step": 5211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2370427097294088e-06,
+      "loss": 0.3298,
+      "step": 5212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2336356996055285e-06,
+      "loss": 0.8685,
+      "step": 5213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2302301375453003e-06,
+      "loss": 0.9346,
+      "step": 5214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.226826024277551e-06,
+      "loss": 0.8993,
+      "step": 5215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2234233605307885e-06,
+      "loss": 0.8637,
+      "step": 5216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2200221470332084e-06,
+      "loss": 0.8885,
+      "step": 5217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.216622384512713e-06,
+      "loss": 0.9661,
+      "step": 5218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2132240736968743e-06,
+      "loss": 0.8751,
+      "step": 5219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2098272153129663e-06,
+      "loss": 0.9173,
+      "step": 5220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2064318100879465e-06,
+      "loss": 0.8966,
+      "step": 5221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2030378587484578e-06,
+      "loss": 0.9176,
+      "step": 5222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1996453620208445e-06,
+      "loss": 0.8874,
+      "step": 5223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1962543206311235e-06,
+      "loss": 0.8977,
+      "step": 5224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.192864735305017e-06,
+      "loss": 0.9216,
+      "step": 5225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1894766067679207e-06,
+      "loss": 0.9811,
+      "step": 5226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1860899357449294e-06,
+      "loss": 0.9286,
+      "step": 5227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.18270472296082e-06,
+      "loss": 0.9011,
+      "step": 5228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.179320969140055e-06,
+      "loss": 0.8948,
+      "step": 5229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1759386750067955e-06,
+      "loss": 0.8982,
+      "step": 5230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.172557841284877e-06,
+      "loss": 0.9679,
+      "step": 5231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1691784686978343e-06,
+      "loss": 0.8974,
+      "step": 5232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1658005579688787e-06,
+      "loss": 0.9128,
+      "step": 5233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1624241098209186e-06,
+      "loss": 0.9478,
+      "step": 5234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159049124976543e-06,
+      "loss": 0.904,
+      "step": 5235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.155675604158026e-06,
+      "loss": 0.9465,
+      "step": 5236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152303548087339e-06,
+      "loss": 0.8735,
+      "step": 5237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1489329574861237e-06,
+      "loss": 0.9416,
+      "step": 5238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1455638330757256e-06,
+      "loss": 0.9016,
+      "step": 5239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1421961755771623e-06,
+      "loss": 0.9321,
+      "step": 5240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.138829985711147e-06,
+      "loss": 0.3611,
+      "step": 5241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135465264198074e-06,
+      "loss": 0.9329,
+      "step": 5242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1321020117580215e-06,
+      "loss": 0.919,
+      "step": 5243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1287402291107618e-06,
+      "loss": 0.9364,
+      "step": 5244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1253799169757394e-06,
+      "loss": 0.8976,
+      "step": 5245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.122021076072099e-06,
+      "loss": 0.8265,
+      "step": 5246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1186637071186575e-06,
+      "loss": 0.8802,
+      "step": 5247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1153078108339296e-06,
+      "loss": 0.9286,
+      "step": 5248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.111953387936102e-06,
+      "loss": 0.8733,
+      "step": 5249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10860043914305e-06,
+      "loss": 0.935,
+      "step": 5250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1052489651723416e-06,
+      "loss": 0.9213,
+      "step": 5251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.101898966741217e-06,
+      "loss": 0.9245,
+      "step": 5252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0985504445666113e-06,
+      "loss": 1.007,
+      "step": 5253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.095203399365134e-06,
+      "loss": 0.8905,
+      "step": 5254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091857831853088e-06,
+      "loss": 0.8496,
+      "step": 5255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.088513742746454e-06,
+      "loss": 0.931,
+      "step": 5256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0851711327608923e-06,
+      "loss": 0.9119,
+      "step": 5257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.08183000261176e-06,
+      "loss": 0.9227,
+      "step": 5258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.078490353014081e-06,
+      "loss": 0.8946,
+      "step": 5259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0751521846825794e-06,
+      "loss": 0.888,
+      "step": 5260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0718154983316473e-06,
+      "loss": 0.9457,
+      "step": 5261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.06848029467537e-06,
+      "loss": 0.9622,
+      "step": 5262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.06514657442751e-06,
+      "loss": 0.9278,
+      "step": 5263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.061814338301511e-06,
+      "loss": 0.9201,
+      "step": 5264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058483587010507e-06,
+      "loss": 0.9634,
+      "step": 5265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055154321267304e-06,
+      "loss": 0.88,
+      "step": 5266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0518265417844006e-06,
+      "loss": 0.9305,
+      "step": 5267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.048500249273967e-06,
+      "loss": 0.937,
+      "step": 5268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0451754444478644e-06,
+      "loss": 0.9625,
+      "step": 5269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0418521280176293e-06,
+      "loss": 0.8699,
+      "step": 5270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.038530300694479e-06,
+      "loss": 0.8984,
+      "step": 5271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0352099631893208e-06,
+      "loss": 0.8802,
+      "step": 5272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0318911162127295e-06,
+      "loss": 0.9546,
+      "step": 5273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0285737604749777e-06,
+      "loss": 0.8633,
+      "step": 5274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.025257896686001e-06,
+      "loss": 0.8681,
+      "step": 5275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0219435255554296e-06,
+      "loss": 0.9226,
+      "step": 5276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.018630647792571e-06,
+      "loss": 0.8566,
+      "step": 5277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0153192641064044e-06,
+      "loss": 0.9941,
+      "step": 5278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0120093752056045e-06,
+      "loss": 0.9189,
+      "step": 5279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.008700981798509e-06,
+      "loss": 0.9295,
+      "step": 5280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005394084593153e-06,
+      "loss": 0.9747,
+      "step": 5281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0020886842972375e-06,
+      "loss": 0.9198,
+      "step": 5282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.998784781618145e-06,
+      "loss": 0.9555,
+      "step": 5283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9954823772629494e-06,
+      "loss": 0.926,
+      "step": 5284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.992181471938387e-06,
+      "loss": 0.924,
+      "step": 5285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.988882066350888e-06,
+      "loss": 0.9414,
+      "step": 5286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9855841612065493e-06,
+      "loss": 0.9225,
+      "step": 5287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.982287757211161e-06,
+      "loss": 0.9054,
+      "step": 5288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9789928550701773e-06,
+      "loss": 0.8967,
+      "step": 5289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9756994554887354e-06,
+      "loss": 0.2845,
+      "step": 5290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9724075591716594e-06,
+      "loss": 0.3457,
+      "step": 5291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9691171668234387e-06,
+      "loss": 0.9153,
+      "step": 5292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965828279148254e-06,
+      "loss": 0.8704,
+      "step": 5293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9625408968499503e-06,
+      "loss": 0.912,
+      "step": 5294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9592550206320637e-06,
+      "loss": 0.3394,
+      "step": 5295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.955970651197798e-06,
+      "loss": 0.9384,
+      "step": 5296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.952687789250037e-06,
+      "loss": 0.8707,
+      "step": 5297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9494064354913475e-06,
+      "loss": 0.9133,
+      "step": 5298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.946126590623963e-06,
+      "loss": 0.3404,
+      "step": 5299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942848255349807e-06,
+      "loss": 0.9623,
+      "step": 5300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939571430370468e-06,
+      "loss": 0.9352,
+      "step": 5301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9362961163872207e-06,
+      "loss": 0.9103,
+      "step": 5302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9330223141010095e-06,
+      "loss": 0.8379,
+      "step": 5303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9297500242124556e-06,
+      "loss": 0.9264,
+      "step": 5304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926479247421864e-06,
+      "loss": 0.9248,
+      "step": 5305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.923209984429206e-06,
+      "loss": 0.3483,
+      "step": 5306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.919942235934139e-06,
+      "loss": 0.9223,
+      "step": 5307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.916676002635984e-06,
+      "loss": 0.8867,
+      "step": 5308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.913411285233753e-06,
+      "loss": 0.9409,
+      "step": 5309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9101480844261197e-06,
+      "loss": 0.8893,
+      "step": 5310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9068864009114382e-06,
+      "loss": 0.902,
+      "step": 5311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9036262353877433e-06,
+      "loss": 0.9601,
+      "step": 5312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9003675885527326e-06,
+      "loss": 0.9585,
+      "step": 5313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8971104611037957e-06,
+      "loss": 0.8749,
+      "step": 5314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8938548537379783e-06,
+      "loss": 0.8798,
+      "step": 5315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8906007671520185e-06,
+      "loss": 0.8667,
+      "step": 5316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.887348202042317e-06,
+      "loss": 0.9325,
+      "step": 5317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8840971591049482e-06,
+      "loss": 0.8699,
+      "step": 5318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.880847639035671e-06,
+      "loss": 0.9221,
+      "step": 5319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.877599642529908e-06,
+      "loss": 0.9527,
+      "step": 5320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8743531702827643e-06,
+      "loss": 0.9041,
+      "step": 5321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8711082229890095e-06,
+      "loss": 0.9273,
+      "step": 5322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8678648013430977e-06,
+      "loss": 0.898,
+      "step": 5323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8646229060391473e-06,
+      "loss": 0.951,
+      "step": 5324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8613825377709515e-06,
+      "loss": 0.9303,
+      "step": 5325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8581436972319843e-06,
+      "loss": 0.852,
+      "step": 5326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85490638511538e-06,
+      "loss": 0.9238,
+      "step": 5327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8516706021139594e-06,
+      "loss": 0.9045,
+      "step": 5328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848436348920205e-06,
+      "loss": 0.9154,
+      "step": 5329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8452036262262807e-06,
+      "loss": 0.9352,
+      "step": 5330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841972434724013e-06,
+      "loss": 0.9541,
+      "step": 5331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8387427751049134e-06,
+      "loss": 0.9028,
+      "step": 5332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8355146480601537e-06,
+      "loss": 0.9068,
+      "step": 5333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8322880542805784e-06,
+      "loss": 0.8987,
+      "step": 5334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8290629944567184e-06,
+      "loss": 0.8638,
+      "step": 5335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8258394692787615e-06,
+      "loss": 0.9545,
+      "step": 5336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822617479436567e-06,
+      "loss": 0.3611,
+      "step": 5337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8193970256196768e-06,
+      "loss": 0.9188,
+      "step": 5338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816178108517291e-06,
+      "loss": 0.8328,
+      "step": 5339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8129607288182947e-06,
+      "loss": 0.9388,
+      "step": 5340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8097448872112287e-06,
+      "loss": 0.9195,
+      "step": 5341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806530584384318e-06,
+      "loss": 0.8689,
+      "step": 5342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8033178210254517e-06,
+      "loss": 0.9331,
+      "step": 5343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8001065978221876e-06,
+      "loss": 0.9437,
+      "step": 5344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.79689691546176e-06,
+      "loss": 0.9478,
+      "step": 5345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.793688774631066e-06,
+      "loss": 0.9346,
+      "step": 5346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7904821760166835e-06,
+      "loss": 0.9633,
+      "step": 5347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7872771203048467e-06,
+      "loss": 0.9195,
+      "step": 5348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7840736081814736e-06,
+      "loss": 0.991,
+      "step": 5349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.780871640332141e-06,
+      "loss": 0.9575,
+      "step": 5350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777671217442097e-06,
+      "loss": 0.9516,
+      "step": 5351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774472340196267e-06,
+      "loss": 0.9245,
+      "step": 5352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.771275009279235e-06,
+      "loss": 0.905,
+      "step": 5353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7680792253752633e-06,
+      "loss": 0.9022,
+      "step": 5354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764884989168273e-06,
+      "loss": 0.9387,
+      "step": 5355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761692301341866e-06,
+      "loss": 0.8872,
+      "step": 5356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7585011625793046e-06,
+      "loss": 0.915,
+      "step": 5357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7553115735635184e-06,
+      "loss": 0.9116,
+      "step": 5358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7521235349771125e-06,
+      "loss": 0.9152,
+      "step": 5359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7489370475023527e-06,
+      "loss": 0.9097,
+      "step": 5360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7457521118211816e-06,
+      "loss": 0.9177,
+      "step": 5361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.742568728615197e-06,
+      "loss": 0.8914,
+      "step": 5362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73938689856568e-06,
+      "loss": 0.9044,
+      "step": 5363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.736206622353568e-06,
+      "loss": 0.8552,
+      "step": 5364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.733027900659465e-06,
+      "loss": 0.9288,
+      "step": 5365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729850734163653e-06,
+      "loss": 0.881,
+      "step": 5366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7266751235460675e-06,
+      "loss": 0.939,
+      "step": 5367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7235010694863263e-06,
+      "loss": 0.9725,
+      "step": 5368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7203285726636985e-06,
+      "loss": 1.0138,
+      "step": 5369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.717157633757134e-06,
+      "loss": 0.9797,
+      "step": 5370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.713988253445239e-06,
+      "loss": 0.9382,
+      "step": 5371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.710820432406286e-06,
+      "loss": 0.9241,
+      "step": 5372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.707654171318226e-06,
+      "loss": 0.9577,
+      "step": 5373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.70448947085866e-06,
+      "loss": 0.9005,
+      "step": 5374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701326331704869e-06,
+      "loss": 0.9109,
+      "step": 5375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6981647545337876e-06,
+      "loss": 0.8945,
+      "step": 5376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.695004740022029e-06,
+      "loss": 0.9336,
+      "step": 5377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6918462888458573e-06,
+      "loss": 0.8675,
+      "step": 5378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6886894016812184e-06,
+      "loss": 0.9083,
+      "step": 5379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.685534079203709e-06,
+      "loss": 0.8833,
+      "step": 5380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6823803220885957e-06,
+      "loss": 0.9136,
+      "step": 5381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.679228131010818e-06,
+      "loss": 0.9051,
+      "step": 5382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6760775066449652e-06,
+      "loss": 0.9127,
+      "step": 5383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6729284496653076e-06,
+      "loss": 0.9775,
+      "step": 5384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6697809607457657e-06,
+      "loss": 0.9588,
+      "step": 5385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6666350405599363e-06,
+      "loss": 0.9417,
+      "step": 5386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663490689781072e-06,
+      "loss": 0.8942,
+      "step": 5387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6603479090820896e-06,
+      "loss": 0.9873,
+      "step": 5388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6572066991355795e-06,
+      "loss": 0.898,
+      "step": 5389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654067060613783e-06,
+      "loss": 0.9499,
+      "step": 5390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6509289941886163e-06,
+      "loss": 0.9237,
+      "step": 5391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6477925005316497e-06,
+      "loss": 0.3471,
+      "step": 5392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.644657580314124e-06,
+      "loss": 0.9424,
+      "step": 5393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6415242342069425e-06,
+      "loss": 0.8492,
+      "step": 5394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6383924628806656e-06,
+      "loss": 0.8404,
+      "step": 5395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6352622670055273e-06,
+      "loss": 0.8386,
+      "step": 5396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6321336472514124e-06,
+      "loss": 0.9578,
+      "step": 5397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6290066042878726e-06,
+      "loss": 0.9115,
+      "step": 5398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625881138784131e-06,
+      "loss": 0.9516,
+      "step": 5399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622757251409057e-06,
+      "loss": 0.9297,
+      "step": 5400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6196349428311983e-06,
+      "loss": 0.9265,
+      "step": 5401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.616514213718753e-06,
+      "loss": 0.842,
+      "step": 5402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613395064739588e-06,
+      "loss": 0.9421,
+      "step": 5403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61027749656123e-06,
+      "loss": 0.9313,
+      "step": 5404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6071615098508618e-06,
+      "loss": 0.9261,
+      "step": 5405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6040471052753404e-06,
+      "loss": 0.9197,
+      "step": 5406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60093428350117e-06,
+      "loss": 0.9107,
+      "step": 5407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5978230451945287e-06,
+      "loss": 0.9139,
+      "step": 5408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594713391021244e-06,
+      "loss": 0.9358,
+      "step": 5409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5916053216468173e-06,
+      "loss": 0.9093,
+      "step": 5410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5884988377363996e-06,
+      "loss": 0.913,
+      "step": 5411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5853939399548035e-06,
+      "loss": 0.9619,
+      "step": 5412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5822906289665117e-06,
+      "loss": 0.8808,
+      "step": 5413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5791889054356567e-06,
+      "loss": 0.8751,
+      "step": 5414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5760887700260394e-06,
+      "loss": 0.8766,
+      "step": 5415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572990223401113e-06,
+      "loss": 0.922,
+      "step": 5416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5698932662239984e-06,
+      "loss": 0.8594,
+      "step": 5417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5667978991574717e-06,
+      "loss": 0.9633,
+      "step": 5418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563704122863966e-06,
+      "loss": 0.9464,
+      "step": 5419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5606119380055828e-06,
+      "loss": 0.864,
+      "step": 5420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.557521345244073e-06,
+      "loss": 0.9221,
+      "step": 5421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.554432345240857e-06,
+      "loss": 0.9092,
+      "step": 5422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5513449386570034e-06,
+      "loss": 0.9247,
+      "step": 5423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.548259126153252e-06,
+      "loss": 0.8944,
+      "step": 5424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545174908389987e-06,
+      "loss": 0.9686,
+      "step": 5425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5420922860272655e-06,
+      "loss": 0.9142,
+      "step": 5426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.539011259724795e-06,
+      "loss": 0.9246,
+      "step": 5427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.53593183014194e-06,
+      "loss": 0.8437,
+      "step": 5428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532853997937732e-06,
+      "loss": 0.9642,
+      "step": 5429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5297777637708498e-06,
+      "loss": 0.9071,
+      "step": 5430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5267031282996402e-06,
+      "loss": 0.9378,
+      "step": 5431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.523630092182098e-06,
+      "loss": 0.8503,
+      "step": 5432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.520558656075888e-06,
+      "loss": 0.9486,
+      "step": 5433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5174888206383195e-06,
+      "loss": 0.908,
+      "step": 5434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514420586526366e-06,
+      "loss": 0.8875,
+      "step": 5435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5113539543966616e-06,
+      "loss": 0.855,
+      "step": 5436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5082889249054878e-06,
+      "loss": 0.8623,
+      "step": 5437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5052254987087943e-06,
+      "loss": 0.9237,
+      "step": 5438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502163676462177e-06,
+      "loss": 0.8842,
+      "step": 5439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4991034588209007e-06,
+      "loss": 0.8719,
+      "step": 5440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4960448464398745e-06,
+      "loss": 0.9468,
+      "step": 5441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4929878399736683e-06,
+      "loss": 0.9128,
+      "step": 5442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4899324400765146e-06,
+      "loss": 0.9295,
+      "step": 5443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.48687864740229e-06,
+      "loss": 0.947,
+      "step": 5444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4838264626045417e-06,
+      "loss": 0.9543,
+      "step": 5445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4807758863364584e-06,
+      "loss": 0.8619,
+      "step": 5446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.477726919250897e-06,
+      "loss": 0.8603,
+      "step": 5447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4746795620003604e-06,
+      "loss": 0.9498,
+      "step": 5448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47163381523701e-06,
+      "loss": 0.9168,
+      "step": 5449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4685896796126673e-06,
+      "loss": 0.8768,
+      "step": 5450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4655471557788e-06,
+      "loss": 0.3618,
+      "step": 5451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.46250624438654e-06,
+      "loss": 0.3373,
+      "step": 5452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4594669460866715e-06,
+      "loss": 0.8748,
+      "step": 5453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.456429261529626e-06,
+      "loss": 0.9232,
+      "step": 5454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.453393191365502e-06,
+      "loss": 0.9463,
+      "step": 5455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4503587362440405e-06,
+      "loss": 0.9088,
+      "step": 5456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44732589681465e-06,
+      "loss": 0.891,
+      "step": 5457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4442946737263783e-06,
+      "loss": 0.8992,
+      "step": 5458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441265067627937e-06,
+      "loss": 0.9494,
+      "step": 5459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4382370791676923e-06,
+      "loss": 0.8953,
+      "step": 5460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.435210708993656e-06,
+      "loss": 0.9327,
+      "step": 5461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4321859577535057e-06,
+      "loss": 0.9462,
+      "step": 5462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429162826094559e-06,
+      "loss": 1.0017,
+      "step": 5463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4261413146638003e-06,
+      "loss": 0.8728,
+      "step": 5464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.423121424107856e-06,
+      "loss": 0.9217,
+      "step": 5465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420103155073009e-06,
+      "loss": 0.8833,
+      "step": 5466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.417086508205201e-06,
+      "loss": 0.8918,
+      "step": 5467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.414071484150017e-06,
+      "loss": 0.9531,
+      "step": 5468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4110580835527053e-06,
+      "loss": 0.9027,
+      "step": 5469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.408046307058155e-06,
+      "loss": 0.9269,
+      "step": 5470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.40503615531092e-06,
+      "loss": 0.8613,
+      "step": 5471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4020276289551934e-06,
+      "loss": 0.8679,
+      "step": 5472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3990207286348343e-06,
+      "loss": 0.9641,
+      "step": 5473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3960154549933434e-06,
+      "loss": 0.9148,
+      "step": 5474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393011808673874e-06,
+      "loss": 0.9334,
+      "step": 5475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390009790319239e-06,
+      "loss": 0.9211,
+      "step": 5476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3870094005718925e-06,
+      "loss": 0.944,
+      "step": 5477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3840106400739494e-06,
+      "loss": 0.9671,
+      "step": 5478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3810135094671695e-06,
+      "loss": 0.8915,
+      "step": 5479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378018009392968e-06,
+      "loss": 0.9075,
+      "step": 5480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375024140492409e-06,
+      "loss": 0.8997,
+      "step": 5481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3720319034062032e-06,
+      "loss": 0.8869,
+      "step": 5482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3690412987747236e-06,
+      "loss": 0.9562,
+      "step": 5483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3660523272379807e-06,
+      "loss": 0.8711,
+      "step": 5484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.363064989435647e-06,
+      "loss": 0.94,
+      "step": 5485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3600792860070356e-06,
+      "loss": 0.9163,
+      "step": 5486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357095217591118e-06,
+      "loss": 0.9645,
+      "step": 5487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.354112784826511e-06,
+      "loss": 0.9015,
+      "step": 5488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3511319883514782e-06,
+      "loss": 0.9245,
+      "step": 5489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3481528288039433e-06,
+      "loss": 0.8986,
+      "step": 5490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.345175306821468e-06,
+      "loss": 0.3453,
+      "step": 5491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.342199423041276e-06,
+      "loss": 0.9353,
+      "step": 5492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3392251781002252e-06,
+      "loss": 0.9476,
+      "step": 5493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3362525726348396e-06,
+      "loss": 0.9184,
+      "step": 5494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3332816072812794e-06,
+      "loss": 0.9279,
+      "step": 5495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3303122826753553e-06,
+      "loss": 0.9161,
+      "step": 5496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3273445994525378e-06,
+      "loss": 0.8816,
+      "step": 5497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3243785582479296e-06,
+      "loss": 0.9183,
+      "step": 5498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321414159696297e-06,
+      "loss": 0.9382,
+      "step": 5499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.318451404432044e-06,
+      "loss": 0.9435,
+      "step": 5500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315490293089232e-06,
+      "loss": 0.8848,
+      "step": 5501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312530826301562e-06,
+      "loss": 0.8632,
+      "step": 5502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3095730047023858e-06,
+      "loss": 0.9287,
+      "step": 5503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3066168289247094e-06,
+      "loss": 0.9382,
+      "step": 5504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3036622996011746e-06,
+      "loss": 0.91,
+      "step": 5505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3007094173640844e-06,
+      "loss": 0.9211,
+      "step": 5506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2977581828453765e-06,
+      "loss": 0.8665,
+      "step": 5507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2948085966766488e-06,
+      "loss": 0.8999,
+      "step": 5508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2918606594891333e-06,
+      "loss": 0.9237,
+      "step": 5509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2889143719137143e-06,
+      "loss": 0.3434,
+      "step": 5510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285969734580933e-06,
+      "loss": 0.9083,
+      "step": 5511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2830267481209623e-06,
+      "loss": 0.9715,
+      "step": 5512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2800854131636276e-06,
+      "loss": 0.8866,
+      "step": 5513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2771457303384048e-06,
+      "loss": 0.8522,
+      "step": 5514
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2742077002744088e-06,
+      "loss": 0.8707,
+      "step": 5515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.271271323600409e-06,
+      "loss": 0.8959,
+      "step": 5516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2683366009448125e-06,
+      "loss": 0.8964,
+      "step": 5517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2654035329356826e-06,
+      "loss": 0.875,
+      "step": 5518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262472120200716e-06,
+      "loss": 0.8872,
+      "step": 5519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2595423633672673e-06,
+      "loss": 0.9659,
+      "step": 5520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256614263062329e-06,
+      "loss": 0.9551,
+      "step": 5521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2536878199125366e-06,
+      "loss": 0.9597,
+      "step": 5522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2507630345441844e-06,
+      "loss": 0.9267,
+      "step": 5523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2478399075831957e-06,
+      "loss": 0.8502,
+      "step": 5524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2449184396551516e-06,
+      "loss": 0.8835,
+      "step": 5525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2419986313852684e-06,
+      "loss": 0.8733,
+      "step": 5526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239080483398416e-06,
+      "loss": 0.9548,
+      "step": 5527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.236163996319103e-06,
+      "loss": 0.9622,
+      "step": 5528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233249170771481e-06,
+      "loss": 0.9337,
+      "step": 5529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2303360073793546e-06,
+      "loss": 0.8844,
+      "step": 5530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2274245067661614e-06,
+      "loss": 0.345,
+      "step": 5531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2245146695549957e-06,
+      "loss": 0.9144,
+      "step": 5532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2216064963685833e-06,
+      "loss": 0.8497,
+      "step": 5533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2186999878293046e-06,
+      "loss": 0.8686,
+      "step": 5534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2157951445591775e-06,
+      "loss": 0.9134,
+      "step": 5535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212891967179861e-06,
+      "loss": 0.8937,
+      "step": 5536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209990456312667e-06,
+      "loss": 0.8855,
+      "step": 5537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207090612578542e-06,
+      "loss": 0.9164,
+      "step": 5538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2041924365980803e-06,
+      "loss": 0.9588,
+      "step": 5539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.201295928991517e-06,
+      "loss": 0.9097,
+      "step": 5540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1984010903787333e-06,
+      "loss": 0.9741,
+      "step": 5541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1955079213792495e-06,
+      "loss": 0.876,
+      "step": 5542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1926164226122283e-06,
+      "loss": 0.9005,
+      "step": 5543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1897265946964806e-06,
+      "loss": 0.9029,
+      "step": 5544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1868384382504514e-06,
+      "loss": 0.8855,
+      "step": 5545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183951953892238e-06,
+      "loss": 0.3742,
+      "step": 5546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181067142239569e-06,
+      "loss": 0.951,
+      "step": 5547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1781840039098255e-06,
+      "loss": 0.9437,
+      "step": 5548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1753025395200233e-06,
+      "loss": 0.3488,
+      "step": 5549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1724227496868188e-06,
+      "loss": 0.9289,
+      "step": 5550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169544635026518e-06,
+      "loss": 0.956,
+      "step": 5551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1666681961550593e-06,
+      "loss": 0.8514,
+      "step": 5552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163793433688032e-06,
+      "loss": 0.9635,
+      "step": 5553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.160920348240656e-06,
+      "loss": 0.8957,
+      "step": 5554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1580489404278025e-06,
+      "loss": 0.3524,
+      "step": 5555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.155179210863978e-06,
+      "loss": 0.9074,
+      "step": 5556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1523111601633274e-06,
+      "loss": 0.8871,
+      "step": 5557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.149444788939644e-06,
+      "loss": 0.3432,
+      "step": 5558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.146580097806352e-06,
+      "loss": 0.927,
+      "step": 5559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1437170873765288e-06,
+      "loss": 0.8953,
+      "step": 5560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1408557582628775e-06,
+      "loss": 0.8913,
+      "step": 5561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1379961110777547e-06,
+      "loss": 0.9272,
+      "step": 5562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1351381464331487e-06,
+      "loss": 0.876,
+      "step": 5563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132281864940686e-06,
+      "loss": 0.9181,
+      "step": 5564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1294272672116433e-06,
+      "loss": 0.8683,
+      "step": 5565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.126574353856926e-06,
+      "loss": 0.9576,
+      "step": 5566
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.123723125487086e-06,
+      "loss": 0.8945,
+      "step": 5567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1208735827123106e-06,
+      "loss": 0.8331,
+      "step": 5568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1180257261424276e-06,
+      "loss": 0.9219,
+      "step": 5569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.115179556386907e-06,
+      "loss": 0.9594,
+      "step": 5570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112335074054852e-06,
+      "loss": 0.9,
+      "step": 5571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109492279755012e-06,
+      "loss": 0.9297,
+      "step": 5572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1066511740957653e-06,
+      "loss": 0.9543,
+      "step": 5573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103811757685139e-06,
+      "loss": 0.9265,
+      "step": 5574
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1009740311307935e-06,
+      "loss": 0.8464,
+      "step": 5575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0981379950400237e-06,
+      "loss": 0.9277,
+      "step": 5576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.095303650019773e-06,
+      "loss": 0.9378,
+      "step": 5577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092470996676612e-06,
+      "loss": 0.9057,
+      "step": 5578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08964003561676e-06,
+      "loss": 0.9322,
+      "step": 5579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0868107674460614e-06,
+      "loss": 0.8617,
+      "step": 5580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0839831927700125e-06,
+      "loss": 0.9161,
+      "step": 5581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.081157312193738e-06,
+      "loss": 0.9398,
+      "step": 5582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0783331263219976e-06,
+      "loss": 0.9637,
+      "step": 5583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.075510635759198e-06,
+      "loss": 0.9138,
+      "step": 5584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0726898411093744e-06,
+      "loss": 0.9537,
+      "step": 5585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0698707429762066e-06,
+      "loss": 0.9182,
+      "step": 5586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0670533419630025e-06,
+      "loss": 0.9476,
+      "step": 5587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064237638672717e-06,
+      "loss": 0.9064,
+      "step": 5588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.061423633707933e-06,
+      "loss": 0.8732,
+      "step": 5589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058611327670872e-06,
+      "loss": 0.3605,
+      "step": 5590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055800721163396e-06,
+      "loss": 0.9113,
+      "step": 5591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0529918147869976e-06,
+      "loss": 0.939,
+      "step": 5592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0501846091428113e-06,
+      "loss": 0.9069,
+      "step": 5593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047379104831602e-06,
+      "loss": 0.9406,
+      "step": 5594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044575302453776e-06,
+      "loss": 0.9311,
+      "step": 5595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041773202609373e-06,
+      "loss": 0.9111,
+      "step": 5596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.038972805898062e-06,
+      "loss": 0.9585,
+      "step": 5597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0361741129191605e-06,
+      "loss": 0.9033,
+      "step": 5598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0333771242716082e-06,
+      "loss": 0.8943,
+      "step": 5599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030581840553991e-06,
+      "loss": 0.9148,
+      "step": 5600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027788262364523e-06,
+      "loss": 0.912,
+      "step": 5601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.024996390301057e-06,
+      "loss": 0.9078,
+      "step": 5602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0222062249610775e-06,
+      "loss": 0.9405,
+      "step": 5603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0194177669417037e-06,
+      "loss": 0.8728,
+      "step": 5604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016631016839696e-06,
+      "loss": 0.8498,
+      "step": 5605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0138459752514385e-06,
+      "loss": 0.9,
+      "step": 5606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0110626427729607e-06,
+      "loss": 0.9352,
+      "step": 5607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008281019999916e-06,
+      "loss": 0.8877,
+      "step": 5608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.005501107527602e-06,
+      "loss": 0.9121,
+      "step": 5609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0027229059509447e-06,
+      "loss": 0.9162,
+      "step": 5610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999946415864499e-06,
+      "loss": 0.8737,
+      "step": 5611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9971716378624664e-06,
+      "loss": 0.9012,
+      "step": 5612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994398572538667e-06,
+      "loss": 0.9241,
+      "step": 5613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9916272204865706e-06,
+      "loss": 0.9075,
+      "step": 5614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988857582299265e-06,
+      "loss": 0.8814,
+      "step": 5615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9860896585694834e-06,
+      "loss": 0.9084,
+      "step": 5616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9833234498895836e-06,
+      "loss": 0.8909,
+      "step": 5617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9805589568515573e-06,
+      "loss": 0.87,
+      "step": 5618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977796180047037e-06,
+      "loss": 0.3513,
+      "step": 5619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.975035120067275e-06,
+      "loss": 0.896,
+      "step": 5620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972275777503171e-06,
+      "loss": 0.8449,
+      "step": 5621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9695181529452423e-06,
+      "loss": 0.9226,
+      "step": 5622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9667622469836513e-06,
+      "loss": 0.9625,
+      "step": 5623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9640080602081856e-06,
+      "loss": 0.9923,
+      "step": 5624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961255593208262e-06,
+      "loss": 0.9413,
+      "step": 5625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9585048465729407e-06,
+      "loss": 0.8684,
+      "step": 5626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9557558208908976e-06,
+      "loss": 0.9331,
+      "step": 5627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9530085167504587e-06,
+      "loss": 0.8992,
+      "step": 5628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9502629347395684e-06,
+      "loss": 0.8789,
+      "step": 5629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.947519075445804e-06,
+      "loss": 0.8895,
+      "step": 5630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94477693945638e-06,
+      "loss": 0.9683,
+      "step": 5631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9420365273581343e-06,
+      "loss": 0.9367,
+      "step": 5632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.939297839737545e-06,
+      "loss": 0.9013,
+      "step": 5633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936560877180712e-06,
+      "loss": 0.9165,
+      "step": 5634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9338256402733734e-06,
+      "loss": 0.3708,
+      "step": 5635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9310921296008942e-06,
+      "loss": 0.8951,
+      "step": 5636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.928360345748267e-06,
+      "loss": 0.9785,
+      "step": 5637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9256302893001244e-06,
+      "loss": 0.9025,
+      "step": 5638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9229019608407164e-06,
+      "loss": 0.9365,
+      "step": 5639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9201753609539376e-06,
+      "loss": 0.9189,
+      "step": 5640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9174504902232983e-06,
+      "loss": 0.8708,
+      "step": 5641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914727349231953e-06,
+      "loss": 0.8553,
+      "step": 5642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912005938562673e-06,
+      "loss": 0.8897,
+      "step": 5643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9092862587978656e-06,
+      "loss": 0.9196,
+      "step": 5644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9065683105195697e-06,
+      "loss": 0.9061,
+      "step": 5645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9038520943094474e-06,
+      "loss": 0.9018,
+      "step": 5646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9011376107487989e-06,
+      "loss": 0.8471,
+      "step": 5647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8984248604185417e-06,
+      "loss": 0.9328,
+      "step": 5648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8957138438992363e-06,
+      "loss": 0.3558,
+      "step": 5649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8930045617710613e-06,
+      "loss": 0.9215,
+      "step": 5650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8902970146138256e-06,
+      "loss": 0.9864,
+      "step": 5651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.887591203006973e-06,
+      "loss": 0.9013,
+      "step": 5652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8848871275295677e-06,
+      "loss": 0.9316,
+      "step": 5653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8821847887603116e-06,
+      "loss": 0.928,
+      "step": 5654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.879484187277525e-06,
+      "loss": 0.9107,
+      "step": 5655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.876785323659166e-06,
+      "loss": 0.8912,
+      "step": 5656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874088198482814e-06,
+      "loss": 0.9197,
+      "step": 5657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8713928123256742e-06,
+      "loss": 0.8722,
+      "step": 5658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8686991657645904e-06,
+      "loss": 0.9501,
+      "step": 5659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8660072593760225e-06,
+      "loss": 0.9685,
+      "step": 5660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8633170937360666e-06,
+      "loss": 0.8734,
+      "step": 5661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.860628669420439e-06,
+      "loss": 0.8357,
+      "step": 5662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8579419870044924e-06,
+      "loss": 0.8452,
+      "step": 5663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8552570470631982e-06,
+      "loss": 0.868,
+      "step": 5664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8525738501711543e-06,
+      "loss": 0.8825,
+      "step": 5665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8498923969025962e-06,
+      "loss": 0.9838,
+      "step": 5666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8472126878313735e-06,
+      "loss": 0.3632,
+      "step": 5667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.844534723530973e-06,
+      "loss": 0.8564,
+      "step": 5668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8418585045744996e-06,
+      "loss": 0.9156,
+      "step": 5669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.839184031534692e-06,
+      "loss": 0.9241,
+      "step": 5670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8365113049839112e-06,
+      "loss": 0.916,
+      "step": 5671
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8338403254941406e-06,
+      "loss": 0.9195,
+      "step": 5672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8311710936370008e-06,
+      "loss": 0.8421,
+      "step": 5673
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.828503609983726e-06,
+      "loss": 0.9753,
+      "step": 5674
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8258378751051875e-06,
+      "loss": 0.9133,
+      "step": 5675
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.823173889571871e-06,
+      "loss": 0.9576,
+      "step": 5676
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.820511653953899e-06,
+      "loss": 0.9215,
+      "step": 5677
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8178511688210122e-06,
+      "loss": 0.8787,
+      "step": 5678
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8151924347425752e-06,
+      "loss": 0.9421,
+      "step": 5679
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8125354522875883e-06,
+      "loss": 0.3271,
+      "step": 5680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8098802220246625e-06,
+      "loss": 0.8174,
+      "step": 5681
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.807226744522048e-06,
+      "loss": 0.8415,
+      "step": 5682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8045750203476076e-06,
+      "loss": 0.9124,
+      "step": 5683
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801925050068839e-06,
+      "loss": 0.8651,
+      "step": 5684
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7992768342528555e-06,
+      "loss": 0.949,
+      "step": 5685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7966303734664026e-06,
+      "loss": 0.9685,
+      "step": 5686
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7939856682758483e-06,
+      "loss": 0.9246,
+      "step": 5687
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7913427192471788e-06,
+      "loss": 0.9059,
+      "step": 5688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.788701526946015e-06,
+      "loss": 0.9047,
+      "step": 5689
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7860620919375936e-06,
+      "loss": 0.912,
+      "step": 5690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7834244147867742e-06,
+      "loss": 0.9001,
+      "step": 5691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.78078849605805e-06,
+      "loss": 0.9566,
+      "step": 5692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7781543363155263e-06,
+      "loss": 0.8575,
+      "step": 5693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7755219361229425e-06,
+      "loss": 0.3876,
+      "step": 5694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7728912960436495e-06,
+      "loss": 0.9339,
+      "step": 5695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7702624166406356e-06,
+      "loss": 0.8743,
+      "step": 5696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7676352984765022e-06,
+      "loss": 0.9465,
+      "step": 5697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7650099421134725e-06,
+      "loss": 1.0004,
+      "step": 5698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762386348113402e-06,
+      "loss": 0.9489,
+      "step": 5699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7597645170377597e-06,
+      "loss": 0.9245,
+      "step": 5700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7571444494476464e-06,
+      "loss": 0.8687,
+      "step": 5701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7545261459037754e-06,
+      "loss": 0.8468,
+      "step": 5702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7519096069664909e-06,
+      "loss": 0.8416,
+      "step": 5703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749294833195756e-06,
+      "loss": 0.9731,
+      "step": 5704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7466818251511509e-06,
+      "loss": 0.8953,
+      "step": 5705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7440705833918913e-06,
+      "loss": 0.9045,
+      "step": 5706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7414611084767996e-06,
+      "loss": 0.847,
+      "step": 5707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7388534009643321e-06,
+      "loss": 0.9817,
+      "step": 5708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7362474614125579e-06,
+      "loss": 0.9559,
+      "step": 5709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7336432903791778e-06,
+      "loss": 0.3557,
+      "step": 5710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7310408884215036e-06,
+      "loss": 0.8884,
+      "step": 5711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7284402560964709e-06,
+      "loss": 0.9251,
+      "step": 5712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7258413939606444e-06,
+      "loss": 0.9061,
+      "step": 5713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7232443025701996e-06,
+      "loss": 0.8942,
+      "step": 5714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7206489824809414e-06,
+      "loss": 0.9477,
+      "step": 5715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.718055434248288e-06,
+      "loss": 0.9333,
+      "step": 5716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7154636584272877e-06,
+      "loss": 0.8842,
+      "step": 5717
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.712873655572601e-06,
+      "loss": 0.8441,
+      "step": 5718
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7102854262385093e-06,
+      "loss": 0.9227,
+      "step": 5719
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7076989709789226e-06,
+      "loss": 0.9012,
+      "step": 5720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051142903473616e-06,
+      "loss": 0.383,
+      "step": 5721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702531384896976e-06,
+      "loss": 0.944,
+      "step": 5722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699950255180526e-06,
+      "loss": 0.8702,
+      "step": 5723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6973709017504014e-06,
+      "loss": 0.9491,
+      "step": 5724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6947933251586057e-06,
+      "loss": 0.9654,
+      "step": 5725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6922175259567608e-06,
+      "loss": 0.849,
+      "step": 5726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6896435046961168e-06,
+      "loss": 0.928,
+      "step": 5727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.687071261927532e-06,
+      "loss": 0.8813,
+      "step": 5728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684500798201496e-06,
+      "loss": 0.942,
+      "step": 5729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6819321140681044e-06,
+      "loss": 0.9175,
+      "step": 5730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6793652100770863e-06,
+      "loss": 0.9039,
+      "step": 5731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.676800086777778e-06,
+      "loss": 0.9456,
+      "step": 5732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6742367447191387e-06,
+      "loss": 0.9477,
+      "step": 5733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6716751844497504e-06,
+      "loss": 0.9003,
+      "step": 5734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.669115406517806e-06,
+      "loss": 0.9469,
+      "step": 5735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.666557411471127e-06,
+      "loss": 0.9191,
+      "step": 5736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6640011998571415e-06,
+      "loss": 0.9273,
+      "step": 5737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6614467722229077e-06,
+      "loss": 0.9246,
+      "step": 5738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6588941291150939e-06,
+      "loss": 0.916,
+      "step": 5739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6563432710799854e-06,
+      "loss": 0.9218,
+      "step": 5740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6537941986634964e-06,
+      "loss": 0.9809,
+      "step": 5741
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6512469124111442e-06,
+      "loss": 0.9179,
+      "step": 5742
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.648701412868078e-06,
+      "loss": 1.0087,
+      "step": 5743
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.646157700579052e-06,
+      "loss": 0.9177,
+      "step": 5744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6436157760884464e-06,
+      "loss": 0.9,
+      "step": 5745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6410756399402594e-06,
+      "loss": 0.9264,
+      "step": 5746
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6385372926780984e-06,
+      "loss": 0.9136,
+      "step": 5747
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6360007348451968e-06,
+      "loss": 0.895,
+      "step": 5748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6334659669843955e-06,
+      "loss": 0.9013,
+      "step": 5749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6309329896381655e-06,
+      "loss": 0.9233,
+      "step": 5750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6284018033485816e-06,
+      "loss": 0.8828,
+      "step": 5751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6258724086573408e-06,
+      "loss": 0.9395,
+      "step": 5752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.623344806105761e-06,
+      "loss": 0.9156,
+      "step": 5753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6208189962347654e-06,
+      "loss": 0.8616,
+      "step": 5754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.618294979584908e-06,
+      "loss": 0.9024,
+      "step": 5755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6157727566963444e-06,
+      "loss": 0.892,
+      "step": 5756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6132523281088585e-06,
+      "loss": 0.9868,
+      "step": 5757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6107336943618435e-06,
+      "loss": 0.8625,
+      "step": 5758
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6082168559943067e-06,
+      "loss": 0.3238,
+      "step": 5759
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6057018135448798e-06,
+      "loss": 0.8856,
+      "step": 5760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6031885675517989e-06,
+      "loss": 0.9077,
+      "step": 5761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.600677118552928e-06,
+      "loss": 0.9968,
+      "step": 5762
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.598167467085735e-06,
+      "loss": 0.9155,
+      "step": 5763
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5956596136873126e-06,
+      "loss": 0.8977,
+      "step": 5764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5931535588943615e-06,
+      "loss": 0.9538,
+      "step": 5765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.590649303243198e-06,
+      "loss": 0.9038,
+      "step": 5766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5881468472697604e-06,
+      "loss": 0.9043,
+      "step": 5767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5856461915095933e-06,
+      "loss": 0.9185,
+      "step": 5768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5831473364978634e-06,
+      "loss": 0.8734,
+      "step": 5769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5806502827693437e-06,
+      "loss": 0.8875,
+      "step": 5770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5781550308584326e-06,
+      "loss": 0.9752,
+      "step": 5771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5756615812991327e-06,
+      "loss": 0.8296,
+      "step": 5772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.573169934625064e-06,
+      "loss": 0.9621,
+      "step": 5773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5706800913694653e-06,
+      "loss": 0.3577,
+      "step": 5774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5681920520651816e-06,
+      "loss": 0.9057,
+      "step": 5775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5657058172446804e-06,
+      "loss": 0.8693,
+      "step": 5776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5632213874400337e-06,
+      "loss": 0.8971,
+      "step": 5777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5607387631829374e-06,
+      "loss": 0.9414,
+      "step": 5778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.558257945004693e-06,
+      "loss": 0.8703,
+      "step": 5779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5557789334362172e-06,
+      "loss": 0.8967,
+      "step": 5780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5533017290080433e-06,
+      "loss": 0.9255,
+      "step": 5781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5508263322503125e-06,
+      "loss": 0.9174,
+      "step": 5782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.548352743692787e-06,
+      "loss": 0.8755,
+      "step": 5783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.545880963864832e-06,
+      "loss": 0.9121,
+      "step": 5784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5434109932954354e-06,
+      "loss": 0.8942,
+      "step": 5785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5409428325131926e-06,
+      "loss": 0.9307,
+      "step": 5786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5384764820463083e-06,
+      "loss": 0.9028,
+      "step": 5787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5360119424226105e-06,
+      "loss": 0.9617,
+      "step": 5788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5335492141695263e-06,
+      "loss": 0.893,
+      "step": 5789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5310882978141084e-06,
+      "loss": 0.8626,
+      "step": 5790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528629193883011e-06,
+      "loss": 0.9319,
+      "step": 5791
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526171902902508e-06,
+      "loss": 0.9373,
+      "step": 5792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.523716425398477e-06,
+      "loss": 0.9137,
+      "step": 5793
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5212627618964193e-06,
+      "loss": 0.905,
+      "step": 5794
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.518810912921438e-06,
+      "loss": 0.8672,
+      "step": 5795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5163608789982476e-06,
+      "loss": 0.9284,
+      "step": 5796
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5139126606511845e-06,
+      "loss": 0.8619,
+      "step": 5797
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5114662584041838e-06,
+      "loss": 0.913,
+      "step": 5798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5090216727808027e-06,
+      "loss": 0.8958,
+      "step": 5799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5065789043042e-06,
+      "loss": 0.8974,
+      "step": 5800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5041379534971546e-06,
+      "loss": 0.8943,
+      "step": 5801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5016988208820494e-06,
+      "loss": 0.9483,
+      "step": 5802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4992615069808814e-06,
+      "loss": 0.9103,
+      "step": 5803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4968260123152622e-06,
+      "loss": 0.9048,
+      "step": 5804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.494392337406405e-06,
+      "loss": 0.9007,
+      "step": 5805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4919604827751388e-06,
+      "loss": 0.8357,
+      "step": 5806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.489530448941906e-06,
+      "loss": 0.9126,
+      "step": 5807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4871022364267506e-06,
+      "loss": 0.8686,
+      "step": 5808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4846758457493382e-06,
+      "loss": 0.9043,
+      "step": 5809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4822512774289332e-06,
+      "loss": 0.9138,
+      "step": 5810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4798285319844207e-06,
+      "loss": 0.3469,
+      "step": 5811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.477407609934287e-06,
+      "loss": 0.8836,
+      "step": 5812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.47498851179663e-06,
+      "loss": 0.893,
+      "step": 5813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4725712380891622e-06,
+      "loss": 0.9441,
+      "step": 5814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4701557893291985e-06,
+      "loss": 0.8713,
+      "step": 5815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4677421660336721e-06,
+      "loss": 0.9057,
+      "step": 5816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4653303687191145e-06,
+      "loss": 0.8944,
+      "step": 5817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4629203979016782e-06,
+      "loss": 0.9173,
+      "step": 5818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4605122540971162e-06,
+      "loss": 0.9156,
+      "step": 5819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4581059378207907e-06,
+      "loss": 0.8702,
+      "step": 5820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4557014495876786e-06,
+      "loss": 0.8945,
+      "step": 5821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4532987899123607e-06,
+      "loss": 0.8896,
+      "step": 5822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.45089795930903e-06,
+      "loss": 0.3413,
+      "step": 5823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4484989582914833e-06,
+      "loss": 0.8336,
+      "step": 5824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.446101787373132e-06,
+      "loss": 0.9375,
+      "step": 5825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4437064470669904e-06,
+      "loss": 0.8895,
+      "step": 5826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4413129378856827e-06,
+      "loss": 0.8939,
+      "step": 5827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4389212603414438e-06,
+      "loss": 0.9032,
+      "step": 5828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4365314149461108e-06,
+      "loss": 0.9171,
+      "step": 5829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4341434022111378e-06,
+      "loss": 0.8793,
+      "step": 5830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4317572226475762e-06,
+      "loss": 0.9098,
+      "step": 5831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4293728767660942e-06,
+      "loss": 0.8822,
+      "step": 5832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4269903650769611e-06,
+      "loss": 0.9037,
+      "step": 5833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4246096880900551e-06,
+      "loss": 0.9564,
+      "step": 5834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4222308463148671e-06,
+      "loss": 0.929,
+      "step": 5835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4198538402604855e-06,
+      "loss": 0.8788,
+      "step": 5836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4174786704356158e-06,
+      "loss": 0.9326,
+      "step": 5837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4151053373485623e-06,
+      "loss": 0.9506,
+      "step": 5838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4127338415072445e-06,
+      "loss": 0.8565,
+      "step": 5839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4103641834191795e-06,
+      "loss": 0.894,
+      "step": 5840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4079963635914996e-06,
+      "loss": 0.8392,
+      "step": 5841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.405630382530937e-06,
+      "loss": 0.9132,
+      "step": 5842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4032662407438325e-06,
+      "loss": 0.9533,
+      "step": 5843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4009039387361378e-06,
+      "loss": 0.9266,
+      "step": 5844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3985434770134031e-06,
+      "loss": 0.9101,
+      "step": 5845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3961848560807934e-06,
+      "loss": 0.8863,
+      "step": 5846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3938280764430688e-06,
+      "loss": 0.872,
+      "step": 5847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3914731386046076e-06,
+      "loss": 0.9158,
+      "step": 5848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3891200430693853e-06,
+      "loss": 0.8647,
+      "step": 5849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3867687903409844e-06,
+      "loss": 0.9528,
+      "step": 5850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.384419380922597e-06,
+      "loss": 0.9411,
+      "step": 5851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3820718153170154e-06,
+      "loss": 0.8892,
+      "step": 5852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3797260940266444e-06,
+      "loss": 0.8993,
+      "step": 5853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3773822175534845e-06,
+      "loss": 0.9239,
+      "step": 5854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3750401863991514e-06,
+      "loss": 0.9152,
+      "step": 5855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3727000010648595e-06,
+      "loss": 0.9395,
+      "step": 5856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.370361662051426e-06,
+      "loss": 0.9343,
+      "step": 5857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3680251698592827e-06,
+      "loss": 0.8337,
+      "step": 5858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3656905249884555e-06,
+      "loss": 0.932,
+      "step": 5859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3633577279385845e-06,
+      "loss": 0.961,
+      "step": 5860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3610267792089038e-06,
+      "loss": 0.9267,
+      "step": 5861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3586976792982597e-06,
+      "loss": 0.8828,
+      "step": 5862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3563704287051062e-06,
+      "loss": 0.904,
+      "step": 5863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3540450279274876e-06,
+      "loss": 0.8623,
+      "step": 5864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3517214774630682e-06,
+      "loss": 0.9334,
+      "step": 5865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.349399777809105e-06,
+      "loss": 0.3678,
+      "step": 5866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3470799294624615e-06,
+      "loss": 0.9425,
+      "step": 5867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3447619329196116e-06,
+      "loss": 0.8937,
+      "step": 5868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3424457886766217e-06,
+      "loss": 0.9321,
+      "step": 5869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3401314972291736e-06,
+      "loss": 0.8493,
+      "step": 5870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3378190590725415e-06,
+      "loss": 0.9285,
+      "step": 5871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3355084747016122e-06,
+      "loss": 0.9094,
+      "step": 5872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3331997446108714e-06,
+      "loss": 0.8621,
+      "step": 5873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3308928692944046e-06,
+      "loss": 0.8804,
+      "step": 5874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3285878492459082e-06,
+      "loss": 0.9401,
+      "step": 5875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3262846849586753e-06,
+      "loss": 0.8983,
+      "step": 5876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.323983376925606e-06,
+      "loss": 0.9198,
+      "step": 5877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3216839256391978e-06,
+      "loss": 0.9356,
+      "step": 5878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.319386331591559e-06,
+      "loss": 0.8838,
+      "step": 5879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.317090595274394e-06,
+      "loss": 0.8915,
+      "step": 5880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3147967171790077e-06,
+      "loss": 0.8765,
+      "step": 5881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3125046977963163e-06,
+      "loss": 0.8985,
+      "step": 5882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3102145376168275e-06,
+      "loss": 0.9471,
+      "step": 5883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3079262371306612e-06,
+      "loss": 0.8755,
+      "step": 5884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3056397968275315e-06,
+      "loss": 0.9501,
+      "step": 5885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30335521719676e-06,
+      "loss": 0.9364,
+      "step": 5886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3010724987272639e-06,
+      "loss": 0.9137,
+      "step": 5887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2987916419075708e-06,
+      "loss": 0.8843,
+      "step": 5888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2965126472258016e-06,
+      "loss": 0.939,
+      "step": 5889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2942355151696796e-06,
+      "loss": 0.8854,
+      "step": 5890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.291960246226538e-06,
+      "loss": 0.9067,
+      "step": 5891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2896868408832985e-06,
+      "loss": 0.9029,
+      "step": 5892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2874152996264966e-06,
+      "loss": 0.9149,
+      "step": 5893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2851456229422588e-06,
+      "loss": 0.9294,
+      "step": 5894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2828778113163186e-06,
+      "loss": 0.882,
+      "step": 5895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2806118652340083e-06,
+      "loss": 0.8831,
+      "step": 5896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2783477851802572e-06,
+      "loss": 0.91,
+      "step": 5897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2760855716396058e-06,
+      "loss": 0.9255,
+      "step": 5898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2738252250961813e-06,
+      "loss": 0.9727,
+      "step": 5899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2715667460337256e-06,
+      "loss": 0.951,
+      "step": 5900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2693101349355674e-06,
+      "loss": 0.9617,
+      "step": 5901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2670553922846461e-06,
+      "loss": 0.8847,
+      "step": 5902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2648025185634971e-06,
+      "loss": 0.8561,
+      "step": 5903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2625515142542533e-06,
+      "loss": 0.9382,
+      "step": 5904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2603023798386527e-06,
+      "loss": 0.9026,
+      "step": 5905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.258055115798027e-06,
+      "loss": 0.9161,
+      "step": 5906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2558097226133158e-06,
+      "loss": 0.3355,
+      "step": 5907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2535662007650507e-06,
+      "loss": 0.9003,
+      "step": 5908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2513245507333682e-06,
+      "loss": 0.8874,
+      "step": 5909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2490847729980005e-06,
+      "loss": 0.9091,
+      "step": 5910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2468468680382773e-06,
+      "loss": 0.8959,
+      "step": 5911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2446108363331366e-06,
+      "loss": 0.8648,
+      "step": 5912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2423766783611046e-06,
+      "loss": 0.9559,
+      "step": 5913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2401443946003156e-06,
+      "loss": 0.9235,
+      "step": 5914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.237913985528495e-06,
+      "loss": 0.3321,
+      "step": 5915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2356854516229732e-06,
+      "loss": 0.9252,
+      "step": 5916
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2334587933606768e-06,
+      "loss": 0.9424,
+      "step": 5917
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2312340112181288e-06,
+      "loss": 0.3262,
+      "step": 5918
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2290111056714527e-06,
+      "loss": 0.8823,
+      "step": 5919
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2267900771963737e-06,
+      "loss": 0.8713,
+      "step": 5920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2245709262682093e-06,
+      "loss": 0.9117,
+      "step": 5921
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2223536533618807e-06,
+      "loss": 0.921,
+      "step": 5922
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2201382589519018e-06,
+      "loss": 0.9213,
+      "step": 5923
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2179247435123898e-06,
+      "loss": 0.9464,
+      "step": 5924
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2157131075170535e-06,
+      "loss": 0.9214,
+      "step": 5925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213503351439208e-06,
+      "loss": 0.9147,
+      "step": 5926
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2112954757517582e-06,
+      "loss": 0.8838,
+      "step": 5927
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.209089480927208e-06,
+      "loss": 0.9149,
+      "step": 5928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2068853674376657e-06,
+      "loss": 0.9543,
+      "step": 5929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.204683135754825e-06,
+      "loss": 0.9041,
+      "step": 5930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.20248278634999e-06,
+      "loss": 0.8843,
+      "step": 5931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2002843196940494e-06,
+      "loss": 0.3458,
+      "step": 5932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1980877362575016e-06,
+      "loss": 0.8826,
+      "step": 5933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1958930365104293e-06,
+      "loss": 0.9744,
+      "step": 5934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1937002209225235e-06,
+      "loss": 0.8468,
+      "step": 5935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1915092899630653e-06,
+      "loss": 0.8928,
+      "step": 5936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1893202441009298e-06,
+      "loss": 0.9084,
+      "step": 5937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.187133083804598e-06,
+      "loss": 0.9117,
+      "step": 5938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1849478095421385e-06,
+      "loss": 0.885,
+      "step": 5939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1827644217812229e-06,
+      "loss": 0.9402,
+      "step": 5940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1805829209891128e-06,
+      "loss": 0.9431,
+      "step": 5941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.178403307632674e-06,
+      "loss": 0.9784,
+      "step": 5942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1762255821783596e-06,
+      "loss": 0.9009,
+      "step": 5943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1740497450922229e-06,
+      "loss": 0.8643,
+      "step": 5944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1718757968399154e-06,
+      "loss": 0.9427,
+      "step": 5945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1697037378866782e-06,
+      "loss": 0.8831,
+      "step": 5946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1675335686973566e-06,
+      "loss": 0.9088,
+      "step": 5947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1653652897363809e-06,
+      "loss": 0.9357,
+      "step": 5948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.163198901467788e-06,
+      "loss": 0.8306,
+      "step": 5949
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1610344043552025e-06,
+      "loss": 0.8814,
+      "step": 5950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1588717988618435e-06,
+      "loss": 0.866,
+      "step": 5951
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1567110854505325e-06,
+      "loss": 0.8443,
+      "step": 5952
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1545522645836794e-06,
+      "loss": 0.8967,
+      "step": 5953
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1523953367232954e-06,
+      "loss": 0.9285,
+      "step": 5954
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.150240302330976e-06,
+      "loss": 0.9113,
+      "step": 5955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1480871618679256e-06,
+      "loss": 0.9124,
+      "step": 5956
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1459359157949312e-06,
+      "loss": 0.8982,
+      "step": 5957
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1437865645723788e-06,
+      "loss": 0.9546,
+      "step": 5958
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1416391086602529e-06,
+      "loss": 0.8984,
+      "step": 5959
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1394935485181247e-06,
+      "loss": 0.9097,
+      "step": 5960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1373498846051667e-06,
+      "loss": 0.9256,
+      "step": 5961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1352081173801387e-06,
+      "loss": 0.9149,
+      "step": 5962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1330682473014043e-06,
+      "loss": 0.9076,
+      "step": 5963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1309302748269102e-06,
+      "loss": 0.9005,
+      "step": 5964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1287942004142027e-06,
+      "loss": 0.9643,
+      "step": 5965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.126660024520423e-06,
+      "loss": 0.9337,
+      "step": 5966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.124527747602302e-06,
+      "loss": 0.8557,
+      "step": 5967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1223973701161695e-06,
+      "loss": 0.8539,
+      "step": 5968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1202688925179417e-06,
+      "loss": 0.8608,
+      "step": 5969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1181423152631365e-06,
+      "loss": 0.9487,
+      "step": 5970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1160176388068588e-06,
+      "loss": 0.8628,
+      "step": 5971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.113894863603806e-06,
+      "loss": 0.912,
+      "step": 5972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1117739901082768e-06,
+      "loss": 0.947,
+      "step": 5973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1096550187741517e-06,
+      "loss": 0.9096,
+      "step": 5974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1075379500549155e-06,
+      "loss": 0.936,
+      "step": 5975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1054227844036359e-06,
+      "loss": 0.8973,
+      "step": 5976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1033095222729806e-06,
+      "loss": 0.9024,
+      "step": 5977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1011981641152037e-06,
+      "loss": 0.8922,
+      "step": 5978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0990887103821579e-06,
+      "loss": 0.9092,
+      "step": 5979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0969811615252856e-06,
+      "loss": 0.9326,
+      "step": 5980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0948755179956194e-06,
+      "loss": 0.8789,
+      "step": 5981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0927717802437886e-06,
+      "loss": 0.9403,
+      "step": 5982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0906699487200113e-06,
+      "loss": 0.9494,
+      "step": 5983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.088570023874096e-06,
+      "loss": 0.9126,
+      "step": 5984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0864720061554513e-06,
+      "loss": 0.9587,
+      "step": 5985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0843758960130657e-06,
+      "loss": 0.8807,
+      "step": 5986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0822816938955317e-06,
+      "loss": 0.9078,
+      "step": 5987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0801894002510228e-06,
+      "loss": 0.9261,
+      "step": 5988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.078099015527313e-06,
+      "loss": 0.8433,
+      "step": 5989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0760105401717614e-06,
+      "loss": 0.9054,
+      "step": 5990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.07392397463132e-06,
+      "loss": 0.9657,
+      "step": 5991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0718393193525356e-06,
+      "loss": 0.8952,
+      "step": 5992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.069756574781541e-06,
+      "loss": 0.3903,
+      "step": 5993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0676757413640638e-06,
+      "loss": 0.9425,
+      "step": 5994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0655968195454202e-06,
+      "loss": 0.9475,
+      "step": 5995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06351980977052e-06,
+      "loss": 0.8651,
+      "step": 5996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0614447124838622e-06,
+      "loss": 0.8879,
+      "step": 5997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0593715281295347e-06,
+      "loss": 0.9192,
+      "step": 5998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.057300257151219e-06,
+      "loss": 0.3354,
+      "step": 5999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0552308999921856e-06,
+      "loss": 0.9212,
+      "step": 6000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0531634570952965e-06,
+      "loss": 0.9208,
+      "step": 6001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051097928903002e-06,
+      "loss": 0.8688,
+      "step": 6002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0490343158573467e-06,
+      "loss": 0.8922,
+      "step": 6003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0469726183999606e-06,
+      "loss": 0.8682,
+      "step": 6004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.044912836972064e-06,
+      "loss": 0.8767,
+      "step": 6005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0428549720144721e-06,
+      "loss": 0.8883,
+      "step": 6006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0407990239675848e-06,
+      "loss": 0.9279,
+      "step": 6007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0387449932713954e-06,
+      "loss": 0.9585,
+      "step": 6008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.036692880365483e-06,
+      "loss": 0.953,
+      "step": 6009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0346426856890212e-06,
+      "loss": 0.9164,
+      "step": 6010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0325944096807695e-06,
+      "loss": 0.8677,
+      "step": 6011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0305480527790746e-06,
+      "loss": 0.8948,
+      "step": 6012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0285036154218797e-06,
+      "loss": 0.9533,
+      "step": 6013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0264610980467094e-06,
+      "loss": 0.9093,
+      "step": 6014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0244205010906849e-06,
+      "loss": 0.8717,
+      "step": 6015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0223818249905081e-06,
+      "loss": 0.928,
+      "step": 6016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.020345070182479e-06,
+      "loss": 0.8653,
+      "step": 6017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.01831023710248e-06,
+      "loss": 0.9132,
+      "step": 6018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0162773261859816e-06,
+      "loss": 0.9339,
+      "step": 6019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0142463378680479e-06,
+      "loss": 0.8718,
+      "step": 6020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0122172725833268e-06,
+      "loss": 0.9021,
+      "step": 6021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.01019013076606e-06,
+      "loss": 0.8732,
+      "step": 6022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0081649128500703e-06,
+      "loss": 0.923,
+      "step": 6023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0061416192687768e-06,
+      "loss": 0.89,
+      "step": 6024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0041202504551805e-06,
+      "loss": 0.8893,
+      "step": 6025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0021008068418714e-06,
+      "loss": 0.906,
+      "step": 6026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0000832888610324e-06,
+      "loss": 0.8797,
+      "step": 6027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.980676969444269e-07,
+      "loss": 0.9457,
+      "step": 6028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.960540315234146e-07,
+      "loss": 0.8033,
+      "step": 6029
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.940422930289329e-07,
+      "loss": 0.8791,
+      "step": 6030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.920324818915172e-07,
+      "loss": 0.8824,
+      "step": 6031
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.900245985412826e-07,
+      "loss": 0.9127,
+      "step": 6032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.880186434079341e-07,
+      "loss": 0.9358,
+      "step": 6033
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.860146169207664e-07,
+      "loss": 0.9086,
+      "step": 6034
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.840125195086558e-07,
+      "loss": 0.9103,
+      "step": 6035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.820123516000747e-07,
+      "loss": 0.8631,
+      "step": 6036
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.800141136230723e-07,
+      "loss": 0.8686,
+      "step": 6037
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.780178060052914e-07,
+      "loss": 0.9239,
+      "step": 6038
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.760234291739624e-07,
+      "loss": 0.8663,
+      "step": 6039
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.740309835558969e-07,
+      "loss": 0.839,
+      "step": 6040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.720404695774987e-07,
+      "loss": 0.3842,
+      "step": 6041
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.700518876647535e-07,
+      "loss": 0.8756,
+      "step": 6042
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.6806523824324e-07,
+      "loss": 0.3421,
+      "step": 6043
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.660805217381165e-07,
+      "loss": 0.8758,
+      "step": 6044
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.64097738574129e-07,
+      "loss": 0.9797,
+      "step": 6045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.621168891756139e-07,
+      "loss": 0.9299,
+      "step": 6046
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.601379739664884e-07,
+      "loss": 0.8828,
+      "step": 6047
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58160993370263e-07,
+      "loss": 0.8889,
+      "step": 6048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.561859478100233e-07,
+      "loss": 0.3325,
+      "step": 6049
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.542128377084526e-07,
+      "loss": 0.9057,
+      "step": 6050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.522416634878118e-07,
+      "loss": 0.9365,
+      "step": 6051
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.502724255699492e-07,
+      "loss": 0.9355,
+      "step": 6052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.483051243763031e-07,
+      "loss": 0.9095,
+      "step": 6053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.46339760327889e-07,
+      "loss": 0.8929,
+      "step": 6054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.443763338453183e-07,
+      "loss": 0.3589,
+      "step": 6055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.42414845348777e-07,
+      "loss": 0.2992,
+      "step": 6056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.404552952580459e-07,
+      "loss": 0.9253,
+      "step": 6057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.384976839924842e-07,
+      "loss": 0.8627,
+      "step": 6058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.365420119710378e-07,
+      "loss": 0.9662,
+      "step": 6059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.345882796122408e-07,
+      "loss": 0.9058,
+      "step": 6060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.326364873342065e-07,
+      "loss": 0.9395,
+      "step": 6061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.306866355546407e-07,
+      "loss": 0.8593,
+      "step": 6062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.287387246908252e-07,
+      "loss": 0.868,
+      "step": 6063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.267927551596334e-07,
+      "loss": 0.8695,
+      "step": 6064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.248487273775197e-07,
+      "loss": 0.9671,
+      "step": 6065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.229066417605225e-07,
+      "loss": 0.9016,
+      "step": 6066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.209664987242683e-07,
+      "loss": 0.856,
+      "step": 6067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.190282986839616e-07,
+      "loss": 0.873,
+      "step": 6068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.170920420544005e-07,
+      "loss": 0.9735,
+      "step": 6069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.151577292499547e-07,
+      "loss": 0.9045,
+      "step": 6070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.13225360684592e-07,
+      "loss": 0.9475,
+      "step": 6071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.112949367718526e-07,
+      "loss": 0.9588,
+      "step": 6072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.09366457924864e-07,
+      "loss": 0.892,
+      "step": 6073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.074399245563415e-07,
+      "loss": 0.9356,
+      "step": 6074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.055153370785774e-07,
+      "loss": 0.8726,
+      "step": 6075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.035926959034546e-07,
+      "loss": 0.9353,
+      "step": 6076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.016720014424318e-07,
+      "loss": 0.8856,
+      "step": 6077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.997532541065601e-07,
+      "loss": 0.9656,
+      "step": 6078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.978364543064655e-07,
+      "loss": 0.8898,
+      "step": 6079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.959216024523587e-07,
+      "loss": 0.8809,
+      "step": 6080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.940086989540409e-07,
+      "loss": 0.8773,
+      "step": 6081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.920977442208856e-07,
+      "loss": 0.8982,
+      "step": 6082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.90188738661859e-07,
+      "loss": 0.8648,
+      "step": 6083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.882816826855012e-07,
+      "loss": 0.9019,
+      "step": 6084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.863765766999443e-07,
+      "loss": 0.9415,
+      "step": 6085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.844734211128958e-07,
+      "loss": 0.9732,
+      "step": 6086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.825722163316474e-07,
+      "loss": 0.9102,
+      "step": 6087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.806729627630772e-07,
+      "loss": 0.9155,
+      "step": 6088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.787756608136388e-07,
+      "loss": 0.9304,
+      "step": 6089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.768803108893764e-07,
+      "loss": 0.9796,
+      "step": 6090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.749869133959077e-07,
+      "loss": 0.876,
+      "step": 6091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.730954687384419e-07,
+      "loss": 0.9208,
+      "step": 6092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.712059773217629e-07,
+      "loss": 0.9823,
+      "step": 6093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.693184395502363e-07,
+      "loss": 0.8915,
+      "step": 6094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.674328558278166e-07,
+      "loss": 0.971,
+      "step": 6095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.655492265580334e-07,
+      "loss": 0.3636,
+      "step": 6096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.636675521440041e-07,
+      "loss": 0.9105,
+      "step": 6097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.617878329884222e-07,
+      "loss": 0.9029,
+      "step": 6098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.599100694935625e-07,
+      "loss": 0.9083,
+      "step": 6099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.580342620612859e-07,
+      "loss": 0.9068,
+      "step": 6100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.56160411093031e-07,
+      "loss": 0.8816,
+      "step": 6101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.542885169898208e-07,
+      "loss": 0.972,
+      "step": 6102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.524185801522555e-07,
+      "loss": 0.9389,
+      "step": 6103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.505506009805209e-07,
+      "loss": 0.8942,
+      "step": 6104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.486845798743803e-07,
+      "loss": 0.9538,
+      "step": 6105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.468205172331778e-07,
+      "loss": 0.8913,
+      "step": 6106
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.44958413455843e-07,
+      "loss": 0.7867,
+      "step": 6107
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.430982689408795e-07,
+      "loss": 0.9163,
+      "step": 6108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.412400840863799e-07,
+      "loss": 0.8526,
+      "step": 6109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.393838592900072e-07,
+      "loss": 0.9048,
+      "step": 6110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.375295949490148e-07,
+      "loss": 0.9279,
+      "step": 6111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.356772914602307e-07,
+      "loss": 0.8478,
+      "step": 6112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.338269492200635e-07,
+      "loss": 0.8907,
+      "step": 6113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.319785686245052e-07,
+      "loss": 0.8783,
+      "step": 6114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.30132150069124e-07,
+      "loss": 0.8982,
+      "step": 6115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.282876939490725e-07,
+      "loss": 0.9078,
+      "step": 6116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.264452006590795e-07,
+      "loss": 0.9173,
+      "step": 6117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.246046705934573e-07,
+      "loss": 0.9057,
+      "step": 6118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.227661041460944e-07,
+      "loss": 0.92,
+      "step": 6119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.209295017104601e-07,
+      "loss": 0.8923,
+      "step": 6120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.19094863679607e-07,
+      "loss": 0.9173,
+      "step": 6121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.172621904461608e-07,
+      "loss": 0.8821,
+      "step": 6122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.154314824023335e-07,
+      "loss": 0.8926,
+      "step": 6123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.136027399399104e-07,
+      "loss": 0.8607,
+      "step": 6124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.117759634502631e-07,
+      "loss": 0.8879,
+      "step": 6125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.099511533243342e-07,
+      "loss": 0.9516,
+      "step": 6126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.081283099526505e-07,
+      "loss": 0.8386,
+      "step": 6127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.063074337253196e-07,
+      "loss": 0.905,
+      "step": 6128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.044885250320223e-07,
+      "loss": 0.9118,
+      "step": 6129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.026715842620258e-07,
+      "loss": 0.9379,
+      "step": 6130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.008566118041672e-07,
+      "loss": 0.923,
+      "step": 6131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.990436080468711e-07,
+      "loss": 0.8517,
+      "step": 6132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.97232573378135e-07,
+      "loss": 0.8991,
+      "step": 6133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.954235081855355e-07,
+      "loss": 0.333,
+      "step": 6134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.936164128562318e-07,
+      "loss": 0.8708,
+      "step": 6135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.918112877769568e-07,
+      "loss": 0.9175,
+      "step": 6136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.900081333340259e-07,
+      "loss": 0.908,
+      "step": 6137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.882069499133272e-07,
+      "loss": 0.8929,
+      "step": 6138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.864077379003343e-07,
+      "loss": 0.921,
+      "step": 6139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.846104976800927e-07,
+      "loss": 0.9297,
+      "step": 6140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.828152296372271e-07,
+      "loss": 0.8981,
+      "step": 6141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.810219341559444e-07,
+      "loss": 0.8968,
+      "step": 6142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.792306116200232e-07,
+      "loss": 0.8758,
+      "step": 6143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.774412624128258e-07,
+      "loss": 0.9197,
+      "step": 6144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.756538869172847e-07,
+      "loss": 0.8543,
+      "step": 6145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.738684855159196e-07,
+      "loss": 0.9993,
+      "step": 6146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.720850585908213e-07,
+      "loss": 0.9195,
+      "step": 6147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.703036065236558e-07,
+      "loss": 0.3411,
+      "step": 6148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.685241296956758e-07,
+      "loss": 0.8819,
+      "step": 6149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.667466284877001e-07,
+      "loss": 0.9468,
+      "step": 6150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.649711032801343e-07,
+      "loss": 0.9029,
+      "step": 6151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.631975544529535e-07,
+      "loss": 0.8962,
+      "step": 6152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.614259823857173e-07,
+      "loss": 0.9184,
+      "step": 6153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.596563874575547e-07,
+      "loss": 0.9641,
+      "step": 6154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.578887700471771e-07,
+      "loss": 0.9071,
+      "step": 6155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.561231305328709e-07,
+      "loss": 0.9158,
+      "step": 6156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.543594692924972e-07,
+      "loss": 0.9216,
+      "step": 6157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.525977867034984e-07,
+      "loss": 0.9152,
+      "step": 6158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.508380831428896e-07,
+      "loss": 0.8858,
+      "step": 6159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.490803589872608e-07,
+      "loss": 0.8686,
+      "step": 6160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.473246146127844e-07,
+      "loss": 0.8316,
+      "step": 6161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.455708503952031e-07,
+      "loss": 0.8943,
+      "step": 6162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.43819066709841e-07,
+      "loss": 0.917,
+      "step": 6163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.420692639315929e-07,
+      "loss": 0.9004,
+      "step": 6164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.40321442434937e-07,
+      "loss": 0.8956,
+      "step": 6165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.385756025939183e-07,
+      "loss": 0.8275,
+      "step": 6166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.368317447821638e-07,
+      "loss": 0.9125,
+      "step": 6167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.35089869372877e-07,
+      "loss": 0.9321,
+      "step": 6168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.333499767388319e-07,
+      "loss": 0.9232,
+      "step": 6169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.316120672523852e-07,
+      "loss": 0.928,
+      "step": 6170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.298761412854604e-07,
+      "loss": 0.9048,
+      "step": 6171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281421992095672e-07,
+      "loss": 0.8966,
+      "step": 6172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.264102413957819e-07,
+      "loss": 0.9703,
+      "step": 6173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.246802682147569e-07,
+      "loss": 0.9195,
+      "step": 6174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.229522800367273e-07,
+      "loss": 0.8899,
+      "step": 6175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.212262772314948e-07,
+      "loss": 0.9099,
+      "step": 6176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19502260168442e-07,
+      "loss": 0.8743,
+      "step": 6177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.177802292165214e-07,
+      "loss": 0.9593,
+      "step": 6178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.16060184744266e-07,
+      "loss": 0.9345,
+      "step": 6179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.143421271197815e-07,
+      "loss": 0.9045,
+      "step": 6180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.126260567107435e-07,
+      "loss": 0.9194,
+      "step": 6181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.109119738844117e-07,
+      "loss": 0.9188,
+      "step": 6182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.091998790076116e-07,
+      "loss": 0.854,
+      "step": 6183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.074897724467511e-07,
+      "loss": 0.9548,
+      "step": 6184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.057816545678042e-07,
+      "loss": 0.8731,
+      "step": 6185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.040755257363274e-07,
+      "loss": 0.8826,
+      "step": 6186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.023713863174475e-07,
+      "loss": 0.9314,
+      "step": 6187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.006692366758616e-07,
+      "loss": 0.9238,
+      "step": 6188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.989690771758506e-07,
+      "loss": 0.8698,
+      "step": 6189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.972709081812601e-07,
+      "loss": 0.9669,
+      "step": 6190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.955747300555182e-07,
+      "loss": 0.9258,
+      "step": 6191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.938805431616169e-07,
+      "loss": 0.8465,
+      "step": 6192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.921883478621328e-07,
+      "loss": 0.8433,
+      "step": 6193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.904981445192083e-07,
+      "loss": 0.9074,
+      "step": 6194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.888099334945619e-07,
+      "loss": 0.9452,
+      "step": 6195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.871237151494881e-07,
+      "loss": 0.9396,
+      "step": 6196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.854394898448513e-07,
+      "loss": 0.9367,
+      "step": 6197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.837572579410922e-07,
+      "loss": 0.8797,
+      "step": 6198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.82077019798223e-07,
+      "loss": 0.9123,
+      "step": 6199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.803987757758312e-07,
+      "loss": 0.9839,
+      "step": 6200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.787225262330765e-07,
+      "loss": 0.8823,
+      "step": 6201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.770482715286886e-07,
+      "loss": 0.891,
+      "step": 6202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.753760120209763e-07,
+      "loss": 0.9114,
+      "step": 6203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.737057480678156e-07,
+      "loss": 0.8982,
+      "step": 6204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.720374800266627e-07,
+      "loss": 0.93,
+      "step": 6205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.703712082545366e-07,
+      "loss": 0.9279,
+      "step": 6206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.687069331080398e-07,
+      "loss": 0.912,
+      "step": 6207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.670446549433396e-07,
+      "loss": 0.9255,
+      "step": 6208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.653843741161781e-07,
+      "loss": 0.885,
+      "step": 6209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.637260909818721e-07,
+      "loss": 0.3157,
+      "step": 6210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.620698058953079e-07,
+      "loss": 0.9518,
+      "step": 6211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.604155192109473e-07,
+      "loss": 0.9688,
+      "step": 6212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.587632312828218e-07,
+      "loss": 0.9264,
+      "step": 6213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.571129424645361e-07,
+      "loss": 0.9144,
+      "step": 6214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.554646531092679e-07,
+      "loss": 0.9452,
+      "step": 6215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.538183635697637e-07,
+      "loss": 0.935,
+      "step": 6216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.521740741983485e-07,
+      "loss": 0.9484,
+      "step": 6217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.505317853469107e-07,
+      "loss": 0.9446,
+      "step": 6218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.488914973669191e-07,
+      "loss": 0.8748,
+      "step": 6219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.472532106094076e-07,
+      "loss": 0.8823,
+      "step": 6220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.456169254249855e-07,
+      "loss": 0.8863,
+      "step": 6221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.439826421638329e-07,
+      "loss": 0.8973,
+      "step": 6222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.423503611756998e-07,
+      "loss": 0.8751,
+      "step": 6223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.407200828099114e-07,
+      "loss": 0.8698,
+      "step": 6224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.390918074153607e-07,
+      "loss": 0.9148,
+      "step": 6225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.374655353405146e-07,
+      "loss": 0.9277,
+      "step": 6226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.358412669334102e-07,
+      "loss": 0.9217,
+      "step": 6227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.342190025416528e-07,
+      "loss": 0.8683,
+      "step": 6228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.325987425124258e-07,
+      "loss": 0.8715,
+      "step": 6229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.309804871924774e-07,
+      "loss": 0.9357,
+      "step": 6230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.293642369281294e-07,
+      "loss": 0.8597,
+      "step": 6231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.27749992065273e-07,
+      "loss": 0.9108,
+      "step": 6232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.26137752949375e-07,
+      "loss": 0.9094,
+      "step": 6233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.245275199254652e-07,
+      "loss": 0.9138,
+      "step": 6234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.229192933381489e-07,
+      "loss": 1.0144,
+      "step": 6235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.213130735316042e-07,
+      "loss": 0.9161,
+      "step": 6236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.197088608495727e-07,
+      "loss": 0.9268,
+      "step": 6237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.181066556353732e-07,
+      "loss": 0.8736,
+      "step": 6238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.165064582318903e-07,
+      "loss": 0.8784,
+      "step": 6239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.149082689815844e-07,
+      "loss": 0.8954,
+      "step": 6240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.133120882264786e-07,
+      "loss": 0.9289,
+      "step": 6241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.117179163081722e-07,
+      "loss": 0.9357,
+      "step": 6242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.101257535678317e-07,
+      "loss": 0.8936,
+      "step": 6243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08535600346194e-07,
+      "loss": 0.9025,
+      "step": 6244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.069474569835687e-07,
+      "loss": 0.8995,
+      "step": 6245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.053613238198297e-07,
+      "loss": 0.9306,
+      "step": 6246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.037772011944277e-07,
+      "loss": 0.9472,
+      "step": 6247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.021950894463768e-07,
+      "loss": 0.9607,
+      "step": 6248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.006149889142632e-07,
+      "loss": 0.9283,
+      "step": 6249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.99036899936245e-07,
+      "loss": 0.8741,
+      "step": 6250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.974608228500434e-07,
+      "loss": 0.9083,
+      "step": 6251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.958867579929583e-07,
+      "loss": 0.3304,
+      "step": 6252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.943147057018506e-07,
+      "loss": 0.945,
+      "step": 6253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.927446663131564e-07,
+      "loss": 0.9107,
+      "step": 6254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.91176640162876e-07,
+      "loss": 0.8623,
+      "step": 6255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.89610627586582e-07,
+      "loss": 0.9336,
+      "step": 6256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.880466289194164e-07,
+      "loss": 0.9296,
+      "step": 6257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.864846444960859e-07,
+      "loss": 0.9377,
+      "step": 6258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.849246746508752e-07,
+      "loss": 0.8776,
+      "step": 6259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.833667197176263e-07,
+      "loss": 0.9035,
+      "step": 6260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.81810780029759e-07,
+      "loss": 0.8987,
+      "step": 6261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.802568559202582e-07,
+      "loss": 0.3511,
+      "step": 6262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.787049477216777e-07,
+      "loss": 0.9127,
+      "step": 6263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.771550557661398e-07,
+      "loss": 0.9404,
+      "step": 6264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.756071803853335e-07,
+      "loss": 0.9076,
+      "step": 6265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.740613219105218e-07,
+      "loss": 0.8896,
+      "step": 6266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.725174806725298e-07,
+      "loss": 0.8494,
+      "step": 6267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.709756570017567e-07,
+      "loss": 0.9016,
+      "step": 6268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.694358512281628e-07,
+      "loss": 0.8571,
+      "step": 6269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.678980636812836e-07,
+      "loss": 0.8923,
+      "step": 6270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.66362294690218e-07,
+      "loss": 0.8577,
+      "step": 6271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648285445836344e-07,
+      "loss": 0.8606,
+      "step": 6272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.632968136897721e-07,
+      "loss": 0.8445,
+      "step": 6273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.617671023364335e-07,
+      "loss": 0.9177,
+      "step": 6274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.60239410850989e-07,
+      "loss": 0.9637,
+      "step": 6275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.587137395603825e-07,
+      "loss": 0.9516,
+      "step": 6276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.571900887911174e-07,
+      "loss": 0.9332,
+      "step": 6277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.556684588692718e-07,
+      "loss": 0.9084,
+      "step": 6278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.541488501204862e-07,
+      "loss": 0.8762,
+      "step": 6279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.526312628699737e-07,
+      "loss": 0.8546,
+      "step": 6280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.511156974425092e-07,
+      "loss": 0.8991,
+      "step": 6281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.496021541624375e-07,
+      "loss": 0.9298,
+      "step": 6282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.480906333536729e-07,
+      "loss": 0.8887,
+      "step": 6283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.465811353396921e-07,
+      "loss": 0.9582,
+      "step": 6284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.450736604435435e-07,
+      "loss": 0.8755,
+      "step": 6285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.435682089878391e-07,
+      "loss": 0.9339,
+      "step": 6286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4206478129476e-07,
+      "loss": 0.9134,
+      "step": 6287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.40563377686053e-07,
+      "loss": 0.8502,
+      "step": 6288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.390639984830315e-07,
+      "loss": 0.9141,
+      "step": 6289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.375666440065786e-07,
+      "loss": 0.9898,
+      "step": 6290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.36071314577139e-07,
+      "loss": 0.9218,
+      "step": 6291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.345780105147291e-07,
+      "loss": 0.9443,
+      "step": 6292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.330867321389266e-07,
+      "loss": 0.3193,
+      "step": 6293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.315974797688839e-07,
+      "loss": 0.9102,
+      "step": 6294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.301102537233093e-07,
+      "loss": 0.8915,
+      "step": 6295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.286250543204851e-07,
+      "loss": 0.3536,
+      "step": 6296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.271418818782581e-07,
+      "loss": 0.8983,
+      "step": 6297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.256607367140376e-07,
+      "loss": 0.8666,
+      "step": 6298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.241816191448068e-07,
+      "loss": 0.8542,
+      "step": 6299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.227045294871058e-07,
+      "loss": 0.9566,
+      "step": 6300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.212294680570495e-07,
+      "loss": 0.92,
+      "step": 6301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.197564351703122e-07,
+      "loss": 0.9541,
+      "step": 6302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.182854311421359e-07,
+      "loss": 0.9202,
+      "step": 6303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.168164562873301e-07,
+      "loss": 0.922,
+      "step": 6304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.153495109202667e-07,
+      "loss": 0.942,
+      "step": 6305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.138845953548888e-07,
+      "loss": 0.3012,
+      "step": 6306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.124217099046991e-07,
+      "loss": 0.8812,
+      "step": 6307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.109608548827693e-07,
+      "loss": 0.9348,
+      "step": 6308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.095020306017351e-07,
+      "loss": 0.9204,
+      "step": 6309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.080452373737998e-07,
+      "loss": 0.9175,
+      "step": 6310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.065904755107287e-07,
+      "loss": 0.9258,
+      "step": 6311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.051377453238537e-07,
+      "loss": 0.9649,
+      "step": 6312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.036870471240751e-07,
+      "loss": 0.888,
+      "step": 6313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.022383812218524e-07,
+      "loss": 0.873,
+      "step": 6314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.007917479272151e-07,
+      "loss": 0.9476,
+      "step": 6315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.993471475497546e-07,
+      "loss": 0.9355,
+      "step": 6316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.979045803986315e-07,
+      "loss": 0.8799,
+      "step": 6317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.964640467825654e-07,
+      "loss": 0.882,
+      "step": 6318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.950255470098431e-07,
+      "loss": 0.9272,
+      "step": 6319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.935890813883204e-07,
+      "loss": 0.9018,
+      "step": 6320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.921546502254093e-07,
+      "loss": 0.8935,
+      "step": 6321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.907222538280954e-07,
+      "loss": 0.8399,
+      "step": 6322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.892918925029211e-07,
+      "loss": 0.9286,
+      "step": 6323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.878635665559994e-07,
+      "loss": 0.9395,
+      "step": 6324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.864372762930047e-07,
+      "loss": 0.3163,
+      "step": 6325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.85013022019174e-07,
+      "loss": 0.9429,
+      "step": 6326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.835908040393133e-07,
+      "loss": 0.9385,
+      "step": 6327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.821706226577883e-07,
+      "loss": 0.9501,
+      "step": 6328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.807524781785311e-07,
+      "loss": 0.9037,
+      "step": 6329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.793363709050392e-07,
+      "loss": 0.8613,
+      "step": 6330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7792230114037e-07,
+      "loss": 0.85,
+      "step": 6331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.765102691871504e-07,
+      "loss": 0.8862,
+      "step": 6332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.751002753475653e-07,
+      "loss": 0.8434,
+      "step": 6333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.736923199233678e-07,
+      "loss": 0.3385,
+      "step": 6334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7228640321587247e-07,
+      "loss": 0.8781,
+      "step": 6335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708825255259575e-07,
+      "loss": 0.8687,
+      "step": 6336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6948068715406804e-07,
+      "loss": 0.8978,
+      "step": 6337
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.680808884002075e-07,
+      "loss": 0.896,
+      "step": 6338
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6668312956394846e-07,
+      "loss": 0.9261,
+      "step": 6339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6528741094442054e-07,
+      "loss": 0.9066,
+      "step": 6340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.638937328403226e-07,
+      "loss": 0.8948,
+      "step": 6341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6250209554991485e-07,
+      "loss": 0.8742,
+      "step": 6342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.611124993710181e-07,
+      "loss": 0.9508,
+      "step": 6343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.597249446010221e-07,
+      "loss": 0.9222,
+      "step": 6344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.583394315368717e-07,
+      "loss": 0.896,
+      "step": 6345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5695596047508394e-07,
+      "loss": 0.9016,
+      "step": 6346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5557453171172994e-07,
+      "loss": 0.873,
+      "step": 6347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.54195145542452e-07,
+      "loss": 0.9436,
+      "step": 6348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5281780226244963e-07,
+      "loss": 0.9858,
+      "step": 6349
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.514425021664848e-07,
+      "loss": 0.9018,
+      "step": 6350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5006924554888865e-07,
+      "loss": 0.8876,
+      "step": 6351
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.486980327035451e-07,
+      "loss": 0.9016,
+      "step": 6352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.473288639239115e-07,
+      "loss": 0.8687,
+      "step": 6353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4596173950299915e-07,
+      "loss": 0.8849,
+      "step": 6354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.445966597333873e-07,
+      "loss": 0.9765,
+      "step": 6355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.432336249072133e-07,
+      "loss": 0.8863,
+      "step": 6356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4187263531618173e-07,
+      "loss": 0.8998,
+      "step": 6357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.40513691251554e-07,
+      "loss": 0.9241,
+      "step": 6358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.391567930041574e-07,
+      "loss": 0.9627,
+      "step": 6359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.378019408643819e-07,
+      "loss": 0.9026,
+      "step": 6360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3644913512217556e-07,
+      "loss": 0.884,
+      "step": 6361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.350983760670535e-07,
+      "loss": 0.953,
+      "step": 6362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3374966398808894e-07,
+      "loss": 0.9265,
+      "step": 6363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.324029991739209e-07,
+      "loss": 0.3349,
+      "step": 6364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.310583819127456e-07,
+      "loss": 0.9174,
+      "step": 6365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.297158124923229e-07,
+      "loss": 0.8837,
+      "step": 6366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2837529119997833e-07,
+      "loss": 0.8954,
+      "step": 6367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.270368183225915e-07,
+      "loss": 0.9364,
+      "step": 6368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2570039414661093e-07,
+      "loss": 0.8383,
+      "step": 6369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.243660189580412e-07,
+      "loss": 0.9518,
+      "step": 6370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2303369304245387e-07,
+      "loss": 0.852,
+      "step": 6371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.217034166849765e-07,
+      "loss": 0.9233,
+      "step": 6372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.20375190170299e-07,
+      "loss": 0.9195,
+      "step": 6373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.190490137826775e-07,
+      "loss": 0.9235,
+      "step": 6374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.177248878059226e-07,
+      "loss": 0.7999,
+      "step": 6375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.164028125234121e-07,
+      "loss": 0.3263,
+      "step": 6376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.150827882180797e-07,
+      "loss": 0.918,
+      "step": 6377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1376481517242607e-07,
+      "loss": 0.8763,
+      "step": 6378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.124488936685056e-07,
+      "loss": 0.3704,
+      "step": 6379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.111350239879386e-07,
+      "loss": 0.9464,
+      "step": 6380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0982320641190787e-07,
+      "loss": 0.8904,
+      "step": 6381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0851344122115e-07,
+      "loss": 0.895,
+      "step": 6382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.072057286959696e-07,
+      "loss": 0.9193,
+      "step": 6383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0590006911622736e-07,
+      "loss": 0.8899,
+      "step": 6384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0459646276134856e-07,
+      "loss": 0.8611,
+      "step": 6385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.032949099103156e-07,
+      "loss": 0.3392,
+      "step": 6386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.019954108416713e-07,
+      "loss": 0.9178,
+      "step": 6387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0069796583352194e-07,
+      "loss": 0.9186,
+      "step": 6388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9940257516353333e-07,
+      "loss": 0.93,
+      "step": 6389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9810923910892807e-07,
+      "loss": 0.8496,
+      "step": 6390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.968179579464959e-07,
+      "loss": 0.85,
+      "step": 6391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.95528731952578e-07,
+      "loss": 0.8015,
+      "step": 6392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.942415614030848e-07,
+      "loss": 0.8792,
+      "step": 6393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.929564465734803e-07,
+      "loss": 0.9332,
+      "step": 6394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9167338773879126e-07,
+      "loss": 0.8021,
+      "step": 6395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9039238517360466e-07,
+      "loss": 0.3418,
+      "step": 6396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8911343915206343e-07,
+      "loss": 0.8346,
+      "step": 6397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8783654994787866e-07,
+      "loss": 0.924,
+      "step": 6398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8656171783431175e-07,
+      "loss": 0.8769,
+      "step": 6399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.852889430841933e-07,
+      "loss": 0.9535,
+      "step": 6400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.840182259699032e-07,
+      "loss": 0.9153,
+      "step": 6401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.827495667633907e-07,
+      "loss": 0.8872,
+      "step": 6402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.814829657361574e-07,
+      "loss": 0.8865,
+      "step": 6403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8021842315927094e-07,
+      "loss": 0.9176,
+      "step": 6404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.789559393033526e-07,
+      "loss": 0.9398,
+      "step": 6405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.776955144385852e-07,
+      "loss": 0.8703,
+      "step": 6406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.764371488347118e-07,
+      "loss": 0.3314,
+      "step": 6407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7518084276103373e-07,
+      "loss": 0.8529,
+      "step": 6408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.739265964864136e-07,
+      "loss": 0.9315,
+      "step": 6409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7267441027926896e-07,
+      "loss": 0.8615,
+      "step": 6410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7142428440758215e-07,
+      "loss": 0.9682,
+      "step": 6411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7017621913888915e-07,
+      "loss": 0.9475,
+      "step": 6412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6893021474028625e-07,
+      "loss": 0.8843,
+      "step": 6413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.676862714784335e-07,
+      "loss": 0.9091,
+      "step": 6414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6644438961954243e-07,
+      "loss": 0.863,
+      "step": 6415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.652045694293904e-07,
+      "loss": 0.8963,
+      "step": 6416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6396681117330635e-07,
+      "loss": 0.8332,
+      "step": 6417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6273111511618607e-07,
+      "loss": 0.3764,
+      "step": 6418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.61497481522477e-07,
+      "loss": 0.8934,
+      "step": 6419
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6026591065618787e-07,
+      "loss": 0.9394,
+      "step": 6420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.590364027808879e-07,
+      "loss": 0.8915,
+      "step": 6421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5780895815969994e-07,
+      "loss": 0.9041,
+      "step": 6422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5658357705531277e-07,
+      "loss": 0.9453,
+      "step": 6423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5536025972996544e-07,
+      "loss": 0.915,
+      "step": 6424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5413900644546083e-07,
+      "loss": 0.8985,
+      "step": 6425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.529198174631576e-07,
+      "loss": 0.8575,
+      "step": 6426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5170269304397375e-07,
+      "loss": 0.9217,
+      "step": 6427
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.504876334483842e-07,
+      "loss": 0.9835,
+      "step": 6428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4927463893642324e-07,
+      "loss": 0.8963,
+      "step": 6429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4806370976768313e-07,
+      "loss": 0.8899,
+      "step": 6430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4685484620131217e-07,
+      "loss": 0.9175,
+      "step": 6431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4564804849602116e-07,
+      "loss": 0.9102,
+      "step": 6432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.444433169100736e-07,
+      "loss": 0.8957,
+      "step": 6433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.432406517012932e-07,
+      "loss": 0.8782,
+      "step": 6434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4204005312706176e-07,
+      "loss": 0.9048,
+      "step": 6435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4084152144431726e-07,
+      "loss": 0.8859,
+      "step": 6436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3964505690955887e-07,
+      "loss": 0.3276,
+      "step": 6437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.384506597788373e-07,
+      "loss": 0.9275,
+      "step": 6438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.372583303077681e-07,
+      "loss": 0.8599,
+      "step": 6439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3606806875151944e-07,
+      "loss": 0.8739,
+      "step": 6440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3487987536481636e-07,
+      "loss": 0.8712,
+      "step": 6441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.336937504019455e-07,
+      "loss": 0.9265,
+      "step": 6442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3250969411674603e-07,
+      "loss": 0.9075,
+      "step": 6443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3132770676261974e-07,
+      "loss": 0.9205,
+      "step": 6444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.301477885925197e-07,
+      "loss": 0.8587,
+      "step": 6445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2896993985895965e-07,
+      "loss": 0.9523,
+      "step": 6446
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.277941608140134e-07,
+      "loss": 0.9072,
+      "step": 6447
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.266204517093041e-07,
+      "loss": 0.8768,
+      "step": 6448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.254488127960187e-07,
+      "loss": 0.8872,
+      "step": 6449
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2427924432489653e-07,
+      "loss": 0.9764,
+      "step": 6450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2311174654623854e-07,
+      "loss": 0.9336,
+      "step": 6451
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.219463197098993e-07,
+      "loss": 0.9262,
+      "step": 6452
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.207829640652882e-07,
+      "loss": 0.8638,
+      "step": 6453
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1962167986137713e-07,
+      "loss": 0.8892,
+      "step": 6454
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1846246734668964e-07,
+      "loss": 0.9147,
+      "step": 6455
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1730532676930826e-07,
+      "loss": 0.9432,
+      "step": 6456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1615025837687166e-07,
+      "loss": 0.9025,
+      "step": 6457
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.149972624165754e-07,
+      "loss": 0.9191,
+      "step": 6458
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.138463391351709e-07,
+      "loss": 0.8971,
+      "step": 6459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1269748877896467e-07,
+      "loss": 0.8942,
+      "step": 6460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1155071159382435e-07,
+      "loss": 0.9374,
+      "step": 6461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1040600782516807e-07,
+      "loss": 0.8532,
+      "step": 6462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.092633777179743e-07,
+      "loss": 0.9436,
+      "step": 6463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.081228215167742e-07,
+      "loss": 0.8957,
+      "step": 6464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.069843394656602e-07,
+      "loss": 0.8871,
+      "step": 6465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0584793180827745e-07,
+      "loss": 0.3417,
+      "step": 6466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.047135987878236e-07,
+      "loss": 0.8926,
+      "step": 6467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0358134064706113e-07,
+      "loss": 0.9147,
+      "step": 6468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.024511576283007e-07,
+      "loss": 0.8738,
+      "step": 6469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.013230499734132e-07,
+      "loss": 0.9339,
+      "step": 6470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.001970179238234e-07,
+      "loss": 0.9272,
+      "step": 6471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.990730617205129e-07,
+      "loss": 0.9567,
+      "step": 6472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.979511816040181e-07,
+      "loss": 0.9059,
+      "step": 6473
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.968313778144305e-07,
+      "loss": 0.9089,
+      "step": 6474
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9571365059140047e-07,
+      "loss": 0.9212,
+      "step": 6475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9459800017412996e-07,
+      "loss": 0.9047,
+      "step": 6476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.934844268013803e-07,
+      "loss": 0.8934,
+      "step": 6477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.923729307114642e-07,
+      "loss": 0.9419,
+      "step": 6478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.912635121422536e-07,
+      "loss": 0.9418,
+      "step": 6479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9015617133117403e-07,
+      "loss": 0.9255,
+      "step": 6480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.890509085152049e-07,
+      "loss": 0.9449,
+      "step": 6481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.879477239308848e-07,
+      "loss": 0.9435,
+      "step": 6482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8684661781430255e-07,
+      "loss": 0.9669,
+      "step": 6483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8574759040110755e-07,
+      "loss": 0.8614,
+      "step": 6484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.846506419265005e-07,
+      "loss": 0.8736,
+      "step": 6485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8355577262523916e-07,
+      "loss": 0.8429,
+      "step": 6486
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8246298273163495e-07,
+      "loss": 0.8311,
+      "step": 6487
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.813722724795531e-07,
+      "loss": 0.8522,
+      "step": 6488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.802836421024191e-07,
+      "loss": 0.933,
+      "step": 6489
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7919709183320767e-07,
+      "loss": 0.3374,
+      "step": 6490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7811262190445053e-07,
+      "loss": 0.8381,
+      "step": 6491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7703023254823327e-07,
+      "loss": 0.9005,
+      "step": 6492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7594992399620044e-07,
+      "loss": 0.8892,
+      "step": 6493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7487169647954504e-07,
+      "loss": 0.8806,
+      "step": 6494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7379555022901683e-07,
+      "loss": 0.9267,
+      "step": 6495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.727214854749238e-07,
+      "loss": 0.3534,
+      "step": 6496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.716495024471222e-07,
+      "loss": 0.9129,
+      "step": 6497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.705796013750306e-07,
+      "loss": 0.9564,
+      "step": 6498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6951178248761256e-07,
+      "loss": 0.952,
+      "step": 6499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.684460460133953e-07,
+      "loss": 0.9108,
+      "step": 6500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.673823921804541e-07,
+      "loss": 0.9135,
+      "step": 6501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6632082121642034e-07,
+      "loss": 0.8498,
+      "step": 6502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6526133334848105e-07,
+      "loss": 0.8899,
+      "step": 6503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6420392880337486e-07,
+      "loss": 0.933,
+      "step": 6504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6314860780739747e-07,
+      "loss": 0.9661,
+      "step": 6505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.62095370586396e-07,
+      "loss": 0.8833,
+      "step": 6506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6104421736577345e-07,
+      "loss": 0.8743,
+      "step": 6507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.599951483704866e-07,
+      "loss": 0.8968,
+      "step": 6508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5894816382504464e-07,
+      "loss": 0.9117,
+      "step": 6509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5790326395351397e-07,
+      "loss": 0.8562,
+      "step": 6510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5686044897950903e-07,
+      "loss": 0.9111,
+      "step": 6511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5581971912620464e-07,
+      "loss": 0.9526,
+      "step": 6512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5478107461632705e-07,
+      "loss": 0.9099,
+      "step": 6513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5374451567215184e-07,
+      "loss": 0.8689,
+      "step": 6514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.527100425155171e-07,
+      "loss": 0.9762,
+      "step": 6515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5167765536780463e-07,
+      "loss": 0.8883,
+      "step": 6516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.506473544499588e-07,
+      "loss": 0.9226,
+      "step": 6517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.496191399824699e-07,
+      "loss": 0.9792,
+      "step": 6518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4859301218538856e-07,
+      "loss": 0.8976,
+      "step": 6519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.475689712783136e-07,
+      "loss": 0.3442,
+      "step": 6520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.465470174803986e-07,
+      "loss": 0.9405,
+      "step": 6521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4552715101035185e-07,
+      "loss": 0.877,
+      "step": 6522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4450937208643443e-07,
+      "loss": 0.8999,
+      "step": 6523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.434936809264599e-07,
+      "loss": 0.856,
+      "step": 6524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4248007774779425e-07,
+      "loss": 0.8976,
+      "step": 6525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4146856276735963e-07,
+      "loss": 0.952,
+      "step": 6526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4045913620162954e-07,
+      "loss": 0.9118,
+      "step": 6527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3945179826662887e-07,
+      "loss": 0.9004,
+      "step": 6528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3844654917793864e-07,
+      "loss": 0.9038,
+      "step": 6529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3744338915068888e-07,
+      "loss": 0.9058,
+      "step": 6530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3644231839956898e-07,
+      "loss": 0.8976,
+      "step": 6531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3544333713881317e-07,
+      "loss": 0.8941,
+      "step": 6532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3444644558221485e-07,
+      "loss": 0.9276,
+      "step": 6533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3345164394311782e-07,
+      "loss": 0.9533,
+      "step": 6534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3245893243441842e-07,
+      "loss": 0.8809,
+      "step": 6535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3146831126856562e-07,
+      "loss": 0.8603,
+      "step": 6536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3047978065756095e-07,
+      "loss": 0.8549,
+      "step": 6537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.294933408129607e-07,
+      "loss": 0.971,
+      "step": 6538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.285089919458694e-07,
+      "loss": 0.9236,
+      "step": 6539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.275267342669496e-07,
+      "loss": 0.8714,
+      "step": 6540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2654656798641094e-07,
+      "loss": 0.8265,
+      "step": 6541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2556849331401898e-07,
+      "loss": 0.8659,
+      "step": 6542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.245925104590918e-07,
+      "loss": 0.8791,
+      "step": 6543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.236186196304957e-07,
+      "loss": 0.3674,
+      "step": 6544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2264682103665613e-07,
+      "loss": 0.8808,
+      "step": 6545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2167711488554234e-07,
+      "loss": 0.9055,
+      "step": 6546
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2070950138468384e-07,
+      "loss": 0.9354,
+      "step": 6547
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1974398074115832e-07,
+      "loss": 0.8378,
+      "step": 6548
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1878055316159385e-07,
+      "loss": 0.9355,
+      "step": 6549
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1781921885217438e-07,
+      "loss": 0.8782,
+      "step": 6550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1685997801863422e-07,
+      "loss": 0.937,
+      "step": 6551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1590283086625918e-07,
+      "loss": 0.3506,
+      "step": 6552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.149477775998865e-07,
+      "loss": 0.9279,
+      "step": 6553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1399481842390934e-07,
+      "loss": 0.9342,
+      "step": 6554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1304395354226681e-07,
+      "loss": 0.9142,
+      "step": 6555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1209518315845278e-07,
+      "loss": 0.9255,
+      "step": 6556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1114850747551487e-07,
+      "loss": 0.9161,
+      "step": 6557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1020392669604873e-07,
+      "loss": 0.906,
+      "step": 6558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.092614410222027e-07,
+      "loss": 0.8439,
+      "step": 6559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0832105065567875e-07,
+      "loss": 0.9278,
+      "step": 6560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0738275579772815e-07,
+      "loss": 0.8951,
+      "step": 6561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.064465566491558e-07,
+      "loss": 0.8833,
+      "step": 6562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0551245341031366e-07,
+      "loss": 0.9293,
+      "step": 6563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0458044628111184e-07,
+      "loss": 0.8899,
+      "step": 6564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0365053546100743e-07,
+      "loss": 0.335,
+      "step": 6565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0272272114900903e-07,
+      "loss": 0.9339,
+      "step": 6566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.017970035436778e-07,
+      "loss": 0.8887,
+      "step": 6567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.008733828431253e-07,
+      "loss": 0.9123,
+      "step": 6568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999518592450167e-07,
+      "loss": 0.9379,
+      "step": 6569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9903243294656317e-07,
+      "loss": 0.9709,
+      "step": 6570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9811510414453284e-07,
+      "loss": 0.9084,
+      "step": 6571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9719987303524092e-07,
+      "loss": 0.9237,
+      "step": 6572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9628673981455738e-07,
+      "loss": 0.9061,
+      "step": 6573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9537570467789924e-07,
+      "loss": 0.9122,
+      "step": 6574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.944667678202361e-07,
+      "loss": 0.94,
+      "step": 6575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9355992943609015e-07,
+      "loss": 0.8977,
+      "step": 6576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9265518971953167e-07,
+      "loss": 0.8506,
+      "step": 6577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9175254886418471e-07,
+      "loss": 0.9661,
+      "step": 6578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9085200706322026e-07,
+      "loss": 0.8955,
+      "step": 6579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8995356450936532e-07,
+      "loss": 0.8544,
+      "step": 6580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.890572213948938e-07,
+      "loss": 0.9166,
+      "step": 6581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8816297791163007e-07,
+      "loss": 0.9,
+      "step": 6582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8727083425095105e-07,
+      "loss": 0.9335,
+      "step": 6583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.86380790603784e-07,
+      "loss": 0.9085,
+      "step": 6584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8549284716060656e-07,
+      "loss": 0.3174,
+      "step": 6585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.846070041114456e-07,
+      "loss": 0.9211,
+      "step": 6586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.837232616458817e-07,
+      "loss": 0.8563,
+      "step": 6587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.828416199530425e-07,
+      "loss": 0.9128,
+      "step": 6588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8196207922160592e-07,
+      "loss": 0.883,
+      "step": 6589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8108463963980472e-07,
+      "loss": 0.3685,
+      "step": 6590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8020930139541648e-07,
+      "loss": 0.8596,
+      "step": 6591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7933606467577247e-07,
+      "loss": 0.849,
+      "step": 6592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.784649296677532e-07,
+      "loss": 0.84,
+      "step": 6593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7759589655779064e-07,
+      "loss": 0.85,
+      "step": 6594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.767289655318638e-07,
+      "loss": 0.8553,
+      "step": 6595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.758641367755054e-07,
+      "loss": 0.8181,
+      "step": 6596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7500141047379516e-07,
+      "loss": 0.9032,
+      "step": 6597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7414078681136536e-07,
+      "loss": 0.9504,
+      "step": 6598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7328226597239873e-07,
+      "loss": 0.7941,
+      "step": 6599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.724258481406238e-07,
+      "loss": 0.9085,
+      "step": 6600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7157153349932399e-07,
+      "loss": 0.8984,
+      "step": 6601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.707193222313286e-07,
+      "loss": 0.895,
+      "step": 6602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.698692145190195e-07,
+      "loss": 0.9033,
+      "step": 6603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6902121054432784e-07,
+      "loss": 0.851,
+      "step": 6604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6817531048873404e-07,
+      "loss": 0.3665,
+      "step": 6605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.673315145332677e-07,
+      "loss": 1.0063,
+      "step": 6606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.664898228585099e-07,
+      "loss": 0.8604,
+      "step": 6607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6565023564458994e-07,
+      "loss": 0.8912,
+      "step": 6608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.648127530711874e-07,
+      "loss": 0.8904,
+      "step": 6609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6397737531753e-07,
+      "loss": 0.8893,
+      "step": 6610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.63144102562397e-07,
+      "loss": 0.9878,
+      "step": 6611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.623129349841157e-07,
+      "loss": 0.9152,
+      "step": 6612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.614838727605661e-07,
+      "loss": 0.9216,
+      "step": 6613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6065691606917177e-07,
+      "loss": 0.8487,
+      "step": 6614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.598320650869112e-07,
+      "loss": 0.8817,
+      "step": 6615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5900931999030866e-07,
+      "loss": 0.8294,
+      "step": 6616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5818868095544005e-07,
+      "loss": 0.9226,
+      "step": 6617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5737014815792929e-07,
+      "loss": 0.8608,
+      "step": 6618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5655372177294957e-07,
+      "loss": 0.9097,
+      "step": 6619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.557394019752234e-07,
+      "loss": 0.9489,
+      "step": 6620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.549271889390236e-07,
+      "loss": 0.3398,
+      "step": 6621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5411708283816995e-07,
+      "loss": 0.9711,
+      "step": 6622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5330908384603382e-07,
+      "loss": 0.3551,
+      "step": 6623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5250319213553356e-07,
+      "loss": 0.8234,
+      "step": 6624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5169940787913783e-07,
+      "loss": 0.8664,
+      "step": 6625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5089773124886353e-07,
+      "loss": 0.8818,
+      "step": 6626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5009816241627673e-07,
+      "loss": 0.9297,
+      "step": 6627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4930070155249277e-07,
+      "loss": 0.9,
+      "step": 6628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4850534882817403e-07,
+      "loss": 0.3173,
+      "step": 6629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4771210441353546e-07,
+      "loss": 0.9169,
+      "step": 6630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.469209684783368e-07,
+      "loss": 0.8969,
+      "step": 6631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4613194119189044e-07,
+      "loss": 0.9,
+      "step": 6632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.453450227230524e-07,
+      "loss": 0.985,
+      "step": 6633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.445602132402335e-07,
+      "loss": 0.9165,
+      "step": 6634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.437775129113883e-07,
+      "loss": 0.921,
+      "step": 6635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.429969219040217e-07,
+      "loss": 0.8648,
+      "step": 6636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4221844038518894e-07,
+      "loss": 0.9263,
+      "step": 6637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.414420685214901e-07,
+      "loss": 0.9386,
+      "step": 6638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4066780647907785e-07,
+      "loss": 0.9298,
+      "step": 6639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.398956544236485e-07,
+      "loss": 0.8716,
+      "step": 6640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.391256125204532e-07,
+      "loss": 0.9266,
+      "step": 6641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.383576809342857e-07,
+      "loss": 0.8633,
+      "step": 6642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3759185982949007e-07,
+      "loss": 0.882,
+      "step": 6643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.368281493699608e-07,
+      "loss": 0.9071,
+      "step": 6644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3606654971913714e-07,
+      "loss": 0.8545,
+      "step": 6645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.353070610400087e-07,
+      "loss": 0.8847,
+      "step": 6646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3454968349511322e-07,
+      "loss": 0.8749,
+      "step": 6647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3379441724653774e-07,
+      "loss": 0.9052,
+      "step": 6648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.330412624559152e-07,
+      "loss": 0.9047,
+      "step": 6649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3229021928442554e-07,
+      "loss": 0.9591,
+      "step": 6650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.315412878928024e-07,
+      "loss": 0.9148,
+      "step": 6651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.307944684413198e-07,
+      "loss": 0.3531,
+      "step": 6652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3004976108980882e-07,
+      "loss": 0.88,
+      "step": 6653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.293071659976386e-07,
+      "loss": 0.3643,
+      "step": 6654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2856668332373533e-07,
+      "loss": 0.8732,
+      "step": 6655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2782831322656786e-07,
+      "loss": 0.8997,
+      "step": 6656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2709205586415196e-07,
+      "loss": 0.858,
+      "step": 6657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2635791139405716e-07,
+      "loss": 0.9616,
+      "step": 6658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2562587997339338e-07,
+      "loss": 0.9445,
+      "step": 6659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2489596175882413e-07,
+      "loss": 0.8751,
+      "step": 6660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.241681569065578e-07,
+      "loss": 0.8753,
+      "step": 6661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2344246557235318e-07,
+      "loss": 0.8726,
+      "step": 6662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2271888791151266e-07,
+      "loss": 0.8992,
+      "step": 6663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2199742407888792e-07,
+      "loss": 0.9225,
+      "step": 6664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2127807422888105e-07,
+      "loss": 0.9083,
+      "step": 6665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2056083851543777e-07,
+      "loss": 0.8477,
+      "step": 6666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1984571709205306e-07,
+      "loss": 0.913,
+      "step": 6667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1913271011177119e-07,
+      "loss": 0.3407,
+      "step": 6668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1842181772718009e-07,
+      "loss": 0.9148,
+      "step": 6669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1771304009041917e-07,
+      "loss": 0.9209,
+      "step": 6670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1700637735317155e-07,
+      "loss": 0.9154,
+      "step": 6671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1630182966666958e-07,
+      "loss": 0.9531,
+      "step": 6672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1559939718169377e-07,
+      "loss": 0.9541,
+      "step": 6673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1489908004857164e-07,
+      "loss": 0.8962,
+      "step": 6674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1420087841717553e-07,
+      "loss": 0.9244,
+      "step": 6675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1350479243692925e-07,
+      "loss": 0.3455,
+      "step": 6676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1281082225679918e-07,
+      "loss": 0.931,
+      "step": 6677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1211896802530319e-07,
+      "loss": 0.9141,
+      "step": 6678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1142922989050397e-07,
+      "loss": 0.9162,
+      "step": 6679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.107416080000101e-07,
+      "loss": 0.9417,
+      "step": 6680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1005610250098054e-07,
+      "loss": 0.8698,
+      "step": 6681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0937271354012013e-07,
+      "loss": 0.9543,
+      "step": 6682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0869144126367859e-07,
+      "loss": 0.9401,
+      "step": 6683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0801228581745703e-07,
+      "loss": 0.922,
+      "step": 6684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0733524734679702e-07,
+      "loss": 0.9408,
+      "step": 6685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0666032599659593e-07,
+      "loss": 0.8822,
+      "step": 6686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0598752191128825e-07,
+      "loss": 0.858,
+      "step": 6687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0531683523486324e-07,
+      "loss": 0.9197,
+      "step": 6688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0464826611085389e-07,
+      "loss": 0.8761,
+      "step": 6689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0398181468233792e-07,
+      "loss": 0.8596,
+      "step": 6690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0331748109194461e-07,
+      "loss": 0.9091,
+      "step": 6691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0265526548184468e-07,
+      "loss": 0.9645,
+      "step": 6692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.019951679937603e-07,
+      "loss": 0.3349,
+      "step": 6693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0133718876895849e-07,
+      "loss": 0.9658,
+      "step": 6694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0068132794825214e-07,
+      "loss": 0.8941,
+      "step": 6695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0002758567200122e-07,
+      "loss": 0.9085,
+      "step": 6696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.937596208011158e-08,
+      "loss": 0.8836,
+      "step": 6697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.872645731203944e-08,
+      "loss": 0.8958,
+      "step": 6698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.807907150678253e-08,
+      "loss": 0.876,
+      "step": 6699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.743380480288778e-08,
+      "loss": 0.893,
+      "step": 6700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.679065733844805e-08,
+      "loss": 0.8888,
+      "step": 6701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.614962925110439e-08,
+      "loss": 0.8652,
+      "step": 6702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.551072067804035e-08,
+      "loss": 0.8839,
+      "step": 6703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.48739317559888e-08,
+      "loss": 0.8955,
+      "step": 6704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.423926262122851e-08,
+      "loss": 0.8717,
+      "step": 6705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.360671340958416e-08,
+      "loss": 0.9037,
+      "step": 6706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.297628425642857e-08,
+      "loss": 0.8689,
+      "step": 6707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.234797529667827e-08,
+      "loss": 0.9678,
+      "step": 6708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.172178666479681e-08,
+      "loss": 0.9352,
+      "step": 6709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.109771849479477e-08,
+      "loss": 0.9379,
+      "step": 6710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.047577092022752e-08,
+      "loss": 0.9019,
+      "step": 6711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.985594407419862e-08,
+      "loss": 0.8971,
+      "step": 6712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.923823808935639e-08,
+      "loss": 0.9266,
+      "step": 6713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.862265309789509e-08,
+      "loss": 0.92,
+      "step": 6714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.800918923155599e-08,
+      "loss": 0.8715,
+      "step": 6715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.739784662162521e-08,
+      "loss": 0.9464,
+      "step": 6716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.678862539893807e-08,
+      "loss": 0.9056,
+      "step": 6717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.618152569386918e-08,
+      "loss": 0.8615,
+      "step": 6718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.557654763634792e-08,
+      "loss": 0.9359,
+      "step": 6719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.497369135584188e-08,
+      "loss": 0.9014,
+      "step": 6720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.437295698136894e-08,
+      "loss": 0.8846,
+      "step": 6721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.377434464149181e-08,
+      "loss": 0.9768,
+      "step": 6722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.317785446431914e-08,
+      "loss": 0.8691,
+      "step": 6723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.258348657750437e-08,
+      "loss": 0.9187,
+      "step": 6724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.199124110825019e-08,
+      "loss": 0.9122,
+      "step": 6725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.140111818329966e-08,
+      "loss": 0.9352,
+      "step": 6726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.081311792894508e-08,
+      "loss": 0.9115,
+      "step": 6727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.02272404710247e-08,
+      "loss": 0.8636,
+      "step": 6728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.964348593492044e-08,
+      "loss": 1.0054,
+      "step": 6729
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.906185444556236e-08,
+      "loss": 0.3293,
+      "step": 6730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.848234612742422e-08,
+      "loss": 0.9384,
+      "step": 6731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.790496110452683e-08,
+      "loss": 0.8951,
+      "step": 6732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.732969950043579e-08,
+      "loss": 0.9276,
+      "step": 6733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.675656143826038e-08,
+      "loss": 0.9427,
+      "step": 6734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.618554704066028e-08,
+      "loss": 0.9302,
+      "step": 6735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.56166564298344e-08,
+      "loss": 0.9578,
+      "step": 6736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.504988972753425e-08,
+      "loss": 0.8319,
+      "step": 6737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.448524705505055e-08,
+      "loss": 0.8876,
+      "step": 6738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.392272853322225e-08,
+      "loss": 0.9114,
+      "step": 6739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.336233428243411e-08,
+      "loss": 0.3865,
+      "step": 6740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.280406442261579e-08,
+      "loss": 0.3377,
+      "step": 6741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.224791907324169e-08,
+      "loss": 0.9457,
+      "step": 6742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.169389835333218e-08,
+      "loss": 0.9112,
+      "step": 6743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.11420023814513e-08,
+      "loss": 0.8616,
+      "step": 6744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.059223127571235e-08,
+      "loss": 0.8738,
+      "step": 6745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.00445851537701e-08,
+      "loss": 0.9083,
+      "step": 6746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.949906413282526e-08,
+      "loss": 0.968,
+      "step": 6747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.895566832962553e-08,
+      "loss": 0.9273,
+      "step": 6748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.841439786046123e-08,
+      "loss": 0.9288,
+      "step": 6749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.78752528411708e-08,
+      "loss": 0.8792,
+      "step": 6750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.733823338713419e-08,
+      "loss": 0.8527,
+      "step": 6751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.680333961327945e-08,
+      "loss": 0.858,
+      "step": 6752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.627057163407946e-08,
+      "loss": 0.9068,
+      "step": 6753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.573992956354969e-08,
+      "loss": 0.8714,
+      "step": 6754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.521141351525484e-08,
+      "loss": 0.9035,
+      "step": 6755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.46850236023e-08,
+      "loss": 0.915,
+      "step": 6756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.416075993733839e-08,
+      "loss": 0.905,
+      "step": 6757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.363862263256693e-08,
+      "loss": 0.906,
+      "step": 6758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.311861179972734e-08,
+      "loss": 0.8728,
+      "step": 6759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.260072755010726e-08,
+      "loss": 0.9134,
+      "step": 6760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.208496999453917e-08,
+      "loss": 0.329,
+      "step": 6761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.15713392433992e-08,
+      "loss": 0.9091,
+      "step": 6762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.105983540660942e-08,
+      "loss": 0.8791,
+      "step": 6763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.055045859363562e-08,
+      "loss": 0.9144,
+      "step": 6764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.004320891348836e-08,
+      "loss": 0.8786,
+      "step": 6765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9538086474726366e-08,
+      "loss": 0.9149,
+      "step": 6766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.903509138544761e-08,
+      "loss": 0.8903,
+      "step": 6767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.853422375329931e-08,
+      "loss": 0.9084,
+      "step": 6768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8035483685471294e-08,
+      "loss": 0.9169,
+      "step": 6769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.753887128869928e-08,
+      "loss": 0.8701,
+      "step": 6770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7044386669260486e-08,
+      "loss": 0.9259,
+      "step": 6771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.655202993298248e-08,
+      "loss": 0.8891,
+      "step": 6772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.606180118523097e-08,
+      "loss": 0.8644,
+      "step": 6773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.557370053092093e-08,
+      "loss": 0.9213,
+      "step": 6774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5087728074509904e-08,
+      "loss": 0.8925,
+      "step": 6775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.460388392000138e-08,
+      "loss": 0.8614,
+      "step": 6776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.41221681709414e-08,
+      "loss": 0.3523,
+      "step": 6777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.364258093042085e-08,
+      "loss": 0.9348,
+      "step": 6778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.316512230107762e-08,
+      "loss": 0.899,
+      "step": 6779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2689792385091086e-08,
+      "loss": 0.9136,
+      "step": 6780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2216591284186544e-08,
+      "loss": 0.8637,
+      "step": 6781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.17455190996341e-08,
+      "loss": 0.9192,
+      "step": 6782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.127657593224533e-08,
+      "loss": 0.8947,
+      "step": 6783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.080976188237996e-08,
+      "loss": 0.86,
+      "step": 6784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.034507704993918e-08,
+      "loss": 0.9529,
+      "step": 6785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.988252153437234e-08,
+      "loss": 0.8914,
+      "step": 6786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9422095434668024e-08,
+      "loss": 0.9208,
+      "step": 6787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.896379884936298e-08,
+      "loss": 0.8717,
+      "step": 6788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.850763187653651e-08,
+      "loss": 0.9088,
+      "step": 6789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.805359461381165e-08,
+      "loss": 0.9074,
+      "step": 6790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.760168715835734e-08,
+      "loss": 0.8964,
+      "step": 6791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.71519096068862e-08,
+      "loss": 0.9342,
+      "step": 6792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6704262055654594e-08,
+      "loss": 0.9353,
+      "step": 6793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.625874460046254e-08,
+      "loss": 0.9105,
+      "step": 6794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5815357336654874e-08,
+      "loss": 0.9523,
+      "step": 6795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.537410035912126e-08,
+      "loss": 0.9068,
+      "step": 6796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4934973762293944e-08,
+      "loss": 0.8765,
+      "step": 6797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.449797764015107e-08,
+      "loss": 0.9347,
+      "step": 6798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4063112086212276e-08,
+      "loss": 0.9002,
+      "step": 6799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3630377193544235e-08,
+      "loss": 0.8783,
+      "step": 6800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.319977305475509e-08,
+      "loss": 0.9217,
+      "step": 6801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.27712997619989e-08,
+      "loss": 0.8666,
+      "step": 6802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.234495740697231e-08,
+      "loss": 0.8467,
+      "step": 6803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1920746080915675e-08,
+      "loss": 0.9066,
+      "step": 6804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.149866587461637e-08,
+      "loss": 0.9257,
+      "step": 6805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1078716878401035e-08,
+      "loss": 0.9076,
+      "step": 6806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.066089918214444e-08,
+      "loss": 0.8743,
+      "step": 6807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.024521287526173e-08,
+      "loss": 0.8734,
+      "step": 6808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.983165804671396e-08,
+      "loss": 0.8604,
+      "step": 6809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.942023478500701e-08,
+      "loss": 0.9081,
+      "step": 6810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.901094317818821e-08,
+      "loss": 0.8711,
+      "step": 6811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.860378331384973e-08,
+      "loss": 0.8857,
+      "step": 6812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.819875527912742e-08,
+      "loss": 0.9636,
+      "step": 6813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.779585916070083e-08,
+      "loss": 0.8357,
+      "step": 6814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.739509504479322e-08,
+      "loss": 0.9534,
+      "step": 6815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6996463017172635e-08,
+      "loss": 0.8117,
+      "step": 6816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.659996316314973e-08,
+      "loss": 0.9096,
+      "step": 6817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.620559556757886e-08,
+      "loss": 0.9369,
+      "step": 6818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.581336031485805e-08,
+      "loss": 0.8931,
+      "step": 6819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5423257488929054e-08,
+      "loss": 0.8836,
+      "step": 6820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.503528717327842e-08,
+      "loss": 0.888,
+      "step": 6821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.464944945093418e-08,
+      "loss": 0.8951,
+      "step": 6822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.426574440446917e-08,
+      "loss": 0.8926,
+      "step": 6823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3884172116002144e-08,
+      "loss": 0.96,
+      "step": 6824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.35047326671889e-08,
+      "loss": 0.8824,
+      "step": 6825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.31274261392367e-08,
+      "loss": 0.9231,
+      "step": 6826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.275225261289095e-08,
+      "loss": 0.8659,
+      "step": 6827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.237921216844186e-08,
+      "loss": 0.8708,
+      "step": 6828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.200830488572448e-08,
+      "loss": 0.3137,
+      "step": 6829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1639530844116414e-08,
+      "loss": 0.9228,
+      "step": 6830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.127289012253676e-08,
+      "loss": 0.9147,
+      "step": 6831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.090838279945163e-08,
+      "loss": 0.9329,
+      "step": 6832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0546008952869744e-08,
+      "loss": 0.9203,
+      "step": 6833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.018576866034018e-08,
+      "loss": 0.909,
+      "step": 6834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9827661998959035e-08,
+      "loss": 0.9806,
+      "step": 6835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9471689045363906e-08,
+      "loss": 0.9096,
+      "step": 6836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.911784987573718e-08,
+      "loss": 0.9329,
+      "step": 6837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8766144565801625e-08,
+      "loss": 0.897,
+      "step": 6838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8416573190828134e-08,
+      "loss": 0.9035,
+      "step": 6839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.806913582562576e-08,
+      "loss": 0.9366,
+      "step": 6840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7723832544550577e-08,
+      "loss": 0.8935,
+      "step": 6841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7380663421500142e-08,
+      "loss": 0.9245,
+      "step": 6842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7039628529915706e-08,
+      "loss": 0.9004,
+      "step": 6843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6700727942782224e-08,
+      "loss": 0.951,
+      "step": 6844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.636396173262834e-08,
+      "loss": 0.9175,
+      "step": 6845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6029329971523077e-08,
+      "loss": 0.8812,
+      "step": 6846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5696832731081367e-08,
+      "loss": 0.8823,
+      "step": 6847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5366470082461848e-08,
+      "loss": 0.9588,
+      "step": 6848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.503824209636463e-08,
+      "loss": 0.8903,
+      "step": 6849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.471214884303241e-08,
+      "loss": 0.9049,
+      "step": 6850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.438819039225271e-08,
+      "loss": 0.8958,
+      "step": 6851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.406636681335561e-08,
+      "loss": 0.9858,
+      "step": 6852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.374667817521492e-08,
+      "loss": 0.9679,
+      "step": 6853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.342912454624702e-08,
+      "loss": 0.9723,
+      "step": 6854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3113705994409764e-08,
+      "loss": 0.8951,
+      "step": 6855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2800422587205827e-08,
+      "loss": 0.8831,
+      "step": 6856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.248927439168269e-08,
+      "loss": 0.93,
+      "step": 6857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.218026147442709e-08,
+      "loss": 0.31,
+      "step": 6858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1873383901571678e-08,
+      "loss": 0.9429,
+      "step": 6859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1568641738790585e-08,
+      "loss": 0.9115,
+      "step": 6860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.126603505130054e-08,
+      "loss": 0.9142,
+      "step": 6861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.096556390386306e-08,
+      "loss": 0.9365,
+      "step": 6862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0667228360782278e-08,
+      "loss": 0.936,
+      "step": 6863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0371028485903775e-08,
+      "loss": 0.8942,
+      "step": 6864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0076964342616857e-08,
+      "loss": 0.9651,
+      "step": 6865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9785035993854508e-08,
+      "loss": 0.937,
+      "step": 6866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.949524350209231e-08,
+      "loss": 0.3376,
+      "step": 6867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.920758692934732e-08,
+      "loss": 0.8844,
+      "step": 6868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.892206633718252e-08,
+      "loss": 0.9509,
+      "step": 6869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.863868178670014e-08,
+      "loss": 0.9858,
+      "step": 6870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8357433338547226e-08,
+      "loss": 0.9099,
+      "step": 6871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8078321052914517e-08,
+      "loss": 0.9386,
+      "step": 6872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7801344989534232e-08,
+      "loss": 0.8237,
+      "step": 6873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7526505207682288e-08,
+      "loss": 0.9532,
+      "step": 6874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7253801766176082e-08,
+      "loss": 0.8817,
+      "step": 6875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6983234723375596e-08,
+      "loss": 0.9255,
+      "step": 6876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6714804137186735e-08,
+      "loss": 0.8846,
+      "step": 6877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6448510065055767e-08,
+      "loss": 0.9063,
+      "step": 6878
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.618435256397044e-08,
+      "loss": 0.8696,
+      "step": 6879
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.592233169046442e-08,
+      "loss": 0.8745,
+      "step": 6880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5662447500612855e-08,
+      "loss": 0.8854,
+      "step": 6881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.540470005003125e-08,
+      "loss": 0.9244,
+      "step": 6882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5149089393882153e-08,
+      "loss": 0.8808,
+      "step": 6883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4895615586867363e-08,
+      "loss": 0.9499,
+      "step": 6884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4644278683232372e-08,
+      "loss": 0.9664,
+      "step": 6885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4395078736766376e-08,
+      "loss": 0.8989,
+      "step": 6886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4148015800800053e-08,
+      "loss": 0.9324,
+      "step": 6887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.390308992820777e-08,
+      "loss": 0.8747,
+      "step": 6888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3660301171405377e-08,
+      "loss": 0.9126,
+      "step": 6889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3419649582351314e-08,
+      "loss": 0.8601,
+      "step": 6890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3181135212547714e-08,
+      "loss": 0.8415,
+      "step": 6891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2944758113038191e-08,
+      "loss": 0.8824,
+      "step": 6892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2710518334411171e-08,
+      "loss": 0.9229,
+      "step": 6893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.247841592679544e-08,
+      "loss": 0.8719,
+      "step": 6894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2248450939862378e-08,
+      "loss": 0.8675,
+      "step": 6895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.202062342282706e-08,
+      "loss": 0.9739,
+      "step": 6896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1794933424446042e-08,
+      "loss": 0.8994,
+      "step": 6897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1571380993019577e-08,
+      "loss": 0.8752,
+      "step": 6898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1349966176390503e-08,
+      "loss": 0.3099,
+      "step": 6899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1130689021943142e-08,
+      "loss": 0.9029,
+      "step": 6900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0913549576604399e-08,
+      "loss": 0.8873,
+      "step": 6901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0698547886843769e-08,
+      "loss": 0.9778,
+      "step": 6902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0485683998675555e-08,
+      "loss": 0.924,
+      "step": 6903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.027495795765221e-08,
+      "loss": 0.9188,
+      "step": 6904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.00663698088721e-08,
+      "loss": 0.9455,
+      "step": 6905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.859919596975076e-09,
+      "loss": 0.8405,
+      "step": 6906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.655607366142461e-09,
+      "loss": 0.9298,
+      "step": 6907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.45343316010039e-09,
+      "loss": 0.8906,
+      "step": 6908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.253397022115363e-09,
+      "loss": 0.3357,
+      "step": 6909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.05549899499758e-09,
+      "loss": 0.9491,
+      "step": 6910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.85973912109761e-09,
+      "loss": 0.9628,
+      "step": 6911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.666117442309718e-09,
+      "loss": 0.8965,
+      "step": 6912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.47463400007187e-09,
+      "loss": 0.9016,
+      "step": 6913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.285288835363504e-09,
+      "loss": 0.8913,
+      "step": 6914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.098081988705542e-09,
+      "loss": 0.9239,
+      "step": 6915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.913013500161493e-09,
+      "loss": 0.8957,
+      "step": 6916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.730083409338562e-09,
+      "loss": 0.8582,
+      "step": 6917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.549291755385435e-09,
+      "loss": 0.8927,
+      "step": 6918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3706385769922724e-09,
+      "loss": 0.7912,
+      "step": 6919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.194123912394046e-09,
+      "loss": 0.9281,
+      "step": 6920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.019747799364984e-09,
+      "loss": 0.8778,
+      "step": 6921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.847510275223013e-09,
+      "loss": 0.8545,
+      "step": 6922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.677411376830867e-09,
+      "loss": 0.9637,
+      "step": 6923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.509451140588319e-09,
+      "loss": 0.3666,
+      "step": 6924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3436296024421695e-09,
+      "loss": 0.8725,
+      "step": 6925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.179946797879588e-09,
+      "loss": 0.8633,
+      "step": 6926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.018402761929221e-09,
+      "loss": 0.9153,
+      "step": 6927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.858997529164523e-09,
+      "loss": 0.909,
+      "step": 6928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.701731133697097e-09,
+      "loss": 0.9511,
+      "step": 6929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.546603609186685e-09,
+      "loss": 0.8783,
+      "step": 6930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.393614988830065e-09,
+      "loss": 0.859,
+      "step": 6931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.242765305367714e-09,
+      "loss": 0.8942,
+      "step": 6932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.094054591084918e-09,
+      "loss": 0.8476,
+      "step": 6933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.94748287780511e-09,
+      "loss": 0.3755,
+      "step": 6934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.80305019689653e-09,
+      "loss": 0.9153,
+      "step": 6935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.660756579270009e-09,
+      "loss": 0.8672,
+      "step": 6936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5206020553756335e-09,
+      "loss": 0.9318,
+      "step": 6937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.382586655210519e-09,
+      "loss": 0.9279,
+      "step": 6938
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2467104083088186e-09,
+      "loss": 0.905,
+      "step": 6939
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.112973343750603e-09,
+      "loss": 0.8732,
+      "step": 6940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.981375490156314e-09,
+      "loss": 0.9339,
+      "step": 6941
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.851916875688977e-09,
+      "loss": 0.9121,
+      "step": 6942
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.724597528054208e-09,
+      "loss": 0.9559,
+      "step": 6943
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5994174744991008e-09,
+      "loss": 0.937,
+      "step": 6944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4763767418144466e-09,
+      "loss": 0.8915,
+      "step": 6945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3554753563314057e-09,
+      "loss": 0.9284,
+      "step": 6946
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.236713343922615e-09,
+      "loss": 0.9186,
+      "step": 6947
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.120090730006631e-09,
+      "loss": 0.9043,
+      "step": 6948
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0056075395412664e-09,
+      "loss": 0.8926,
+      "step": 6949
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8932637970247037e-09,
+      "loss": 0.9084,
+      "step": 6950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.783059526502152e-09,
+      "loss": 0.9415,
+      "step": 6951
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6749947515569698e-09,
+      "loss": 0.9184,
+      "step": 6952
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.569069495316212e-09,
+      "loss": 0.9711,
+      "step": 6953
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4652837804495233e-09,
+      "loss": 0.9164,
+      "step": 6954
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.363637629166915e-09,
+      "loss": 0.9244,
+      "step": 6955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.264131063223207e-09,
+      "loss": 0.9027,
+      "step": 6956
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1667641039113675e-09,
+      "loss": 0.9271,
+      "step": 6957
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0715367720713917e-09,
+      "loss": 0.8926,
+      "step": 6958
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.978449088080314e-09,
+      "loss": 0.952,
+      "step": 6959
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8875010718610864e-09,
+      "loss": 0.898,
+      "step": 6960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7986927428781387e-09,
+      "loss": 0.9013,
+      "step": 6961
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7120241201351584e-09,
+      "loss": 0.9212,
+      "step": 6962
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.627495222181752e-09,
+      "loss": 0.9246,
+      "step": 6963
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5451060671078932e-09,
+      "loss": 0.968,
+      "step": 6964
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4648566725450343e-09,
+      "loss": 0.9218,
+      "step": 6965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3867470556661044e-09,
+      "loss": 0.8776,
+      "step": 6966
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3107772331899527e-09,
+      "loss": 0.8207,
+      "step": 6967
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2369472213724643e-09,
+      "loss": 0.9052,
+      "step": 6968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1652570360154435e-09,
+      "loss": 0.9395,
+      "step": 6969
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0957066924599525e-09,
+      "loss": 0.9175,
+      "step": 6970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.028296205591861e-09,
+      "loss": 0.8931,
+      "step": 6971
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.630255898362972e-10,
+      "loss": 0.8915,
+      "step": 6972
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.998948591631973e-10,
+      "loss": 0.912,
+      "step": 6973
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.389040270806447e-10,
+      "loss": 0.8687,
+      "step": 6974
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.800531066437522e-10,
+      "loss": 0.891,
+      "step": 6975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.233421104468896e-10,
+      "loss": 0.9442,
+      "step": 6976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.687710506247946e-10,
+      "loss": 0.8212,
+      "step": 6977
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.163399388581237e-10,
+      "loss": 0.912,
+      "step": 6978
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.660487863667907e-10,
+      "loss": 0.9412,
+      "step": 6979
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.178976039144079e-10,
+      "loss": 0.9197,
+      "step": 6980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.71886401803845e-10,
+      "loss": 0.9097,
+      "step": 6981
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.280151898838902e-10,
+      "loss": 0.8737,
+      "step": 6982
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8628397754258974e-10,
+      "loss": 0.9981,
+      "step": 6983
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.466927737105774e-10,
+      "loss": 0.9563,
+      "step": 6984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.092415868610754e-10,
+      "loss": 0.8917,
+      "step": 6985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7393042500878376e-10,
+      "loss": 0.8653,
+      "step": 6986
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4075929571099055e-10,
+      "loss": 0.8342,
+      "step": 6987
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0972820606646184e-10,
+      "loss": 0.8949,
+      "step": 6988
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8083716271544148e-10,
+      "loss": 0.9579,
+      "step": 6989
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.540861718418718e-10,
+      "loss": 0.8742,
+      "step": 6990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2947523917006267e-10,
+      "loss": 0.9165,
+      "step": 6991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0700436996802233e-10,
+      "loss": 0.9046,
+      "step": 6992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.667356904301649e-11,
+      "loss": 0.921,
+      "step": 6993
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.848284074711942e-11,
+      "loss": 0.9011,
+      "step": 6994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2432188972773024e-11,
+      "loss": 0.9254,
+      "step": 6995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8521617156117574e-11,
+      "loss": 0.9536,
+      "step": 6996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6751128272550774e-11,
+      "loss": 0.91,
+      "step": 6997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.712072484227889e-11,
+      "loss": 0.8754,
+      "step": 6998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.630408925875856e-12,
+      "loss": 0.9539,
+      "step": 6999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.280182126503718e-12,
+      "loss": 0.9219,
+      "step": 7000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0700455888024152e-12,
+      "loss": 0.4884,
+      "step": 7001
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.3682,
+      "step": 7002
+    },
+    {
+      "epoch": 1.0,
+      "step": 7002,
+      "total_flos": 2202206777835520.0,
+      "train_loss": 0.9426911605516184,
+      "train_runtime": 48317.5494,
+      "train_samples_per_second": 18.547,
+      "train_steps_per_second": 0.145
+    }
+  ],
+  "max_steps": 7002,
+  "num_train_epochs": 1,
+  "total_flos": 2202206777835520.0,
+  "trial_name": null,
+  "trial_params": null
+}