diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,39019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5112072355485647,
+  "eval_steps": 10000,
+  "global_step": 6500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999706972585e-05,
+      "loss": 1.5795,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999998827890357e-05,
+      "loss": 1.754,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999997362753348e-05,
+      "loss": 1.5956,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999995311561605e-05,
+      "loss": 0.6651,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999992674315227e-05,
+      "loss": 1.1238,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99999894510143e-05,
+      "loss": 2.3481,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999998564165896e-05,
+      "loss": 1.6445,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999981246249356e-05,
+      "loss": 1.1852,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999976264785652e-05,
+      "loss": 1.114,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999970697268054e-05,
+      "loss": 0.8444,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999964543696766e-05,
+      "loss": 1.2454,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999995780407204e-05,
+      "loss": 2.2296,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999950478394133e-05,
+      "loss": 1.1768,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999994256666333e-05,
+      "loss": 2.0126,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999934068879947e-05,
+      "loss": 1.387,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999924985044314e-05,
+      "loss": 1.6745,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999991531515678e-05,
+      "loss": 1.274,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999905059217724e-05,
+      "loss": 1.5859,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999894217227556e-05,
+      "loss": 1.0141,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999882789186694e-05,
+      "loss": 1.686,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999870775095576e-05,
+      "loss": 1.6247,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999985817495468e-05,
+      "loss": 1.4209,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999844988764504e-05,
+      "loss": 2.0931,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999983121652555e-05,
+      "loss": 1.9927,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999816858238364e-05,
+      "loss": 2.1526,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999801913903508e-05,
+      "loss": 1.2677,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999786383521565e-05,
+      "loss": 1.5658,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999770267093136e-05,
+      "loss": 1.5756,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999753564618856e-05,
+      "loss": 0.9316,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999736276099376e-05,
+      "loss": 1.6823,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999718401535375e-05,
+      "loss": 1.7528,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999699940927546e-05,
+      "loss": 1.986,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999968089427661e-05,
+      "loss": 1.5908,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999661261583317e-05,
+      "loss": 1.9048,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999641042848427e-05,
+      "loss": 1.8603,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999620238072736e-05,
+      "loss": 1.3522,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999598847257056e-05,
+      "loss": 1.2503,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999957687040222e-05,
+      "loss": 1.4977,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999554307509088e-05,
+      "loss": 2.402,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999953115857854e-05,
+      "loss": 1.6138,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999507423611482e-05,
+      "loss": 1.4886,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999483102608846e-05,
+      "loss": 1.4037,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999945819557157e-05,
+      "loss": 1.8236,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999432702500643e-05,
+      "loss": 2.2239,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999406623397048e-05,
+      "loss": 2.1971,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999937995826181e-05,
+      "loss": 1.0433,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999352707095967e-05,
+      "loss": 1.7063,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999932486990059e-05,
+      "loss": 1.8747,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999296446676758e-05,
+      "loss": 1.8582,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999267437425587e-05,
+      "loss": 1.6876,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999923784214821e-05,
+      "loss": 0.624,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999207660845783e-05,
+      "loss": 0.7974,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999176893519488e-05,
+      "loss": 1.3107,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999914554017052e-05,
+      "loss": 0.7121,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999113600800108e-05,
+      "loss": 0.6714,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999081075409504e-05,
+      "loss": 1.3787,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999904796399997e-05,
+      "loss": 0.1954,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999014266572812e-05,
+      "loss": 0.6864,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9998979983129334e-05,
+      "loss": 0.852,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9998945113670878e-05,
+      "loss": 0.5063,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999890965819881e-05,
+      "loss": 0.5452,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9998873616714516e-05,
+      "loss": 0.7472,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99988369892194e-05,
+      "loss": 0.82,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99987997757149e-05,
+      "loss": 1.3238,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999876197620246e-05,
+      "loss": 0.6635,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998723590683566e-05,
+      "loss": 1.5179,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999868461915971e-05,
+      "loss": 0.8199,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998645061632425e-05,
+      "loss": 1.8805,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998604918103242e-05,
+      "loss": 1.9692,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998564188573745e-05,
+      "loss": 0.8651,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998522873045513e-05,
+      "loss": 0.8848,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998480971520166e-05,
+      "loss": 1.5316,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998438483999337e-05,
+      "loss": 0.7601,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998395410484696e-05,
+      "loss": 0.8665,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998351750977913e-05,
+      "loss": 0.5487,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99983075054807e-05,
+      "loss": 0.7243,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999826267399479e-05,
+      "loss": 1.1582,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998217256521926e-05,
+      "loss": 1.0457,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998171253063887e-05,
+      "loss": 0.9357,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999812466362247e-05,
+      "loss": 1.1158,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998077488199495e-05,
+      "loss": 1.1547,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9998029726796806e-05,
+      "loss": 1.4705,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997981379416267e-05,
+      "loss": 1.0294,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999793244605977e-05,
+      "loss": 1.2379,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997882926729224e-05,
+      "loss": 1.6083,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997832821426565e-05,
+      "loss": 1.2068,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999778213015375e-05,
+      "loss": 1.3878,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997730852912763e-05,
+      "loss": 1.2022,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99976789897056e-05,
+      "loss": 1.2121,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997626540534296e-05,
+      "loss": 1.156,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997573505400895e-05,
+      "loss": 1.6848,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997519884307466e-05,
+      "loss": 1.1989,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999746567725611e-05,
+      "loss": 1.3165,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997410884248945e-05,
+      "loss": 1.6889,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997355505288108e-05,
+      "loss": 1.4829,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997299540375764e-05,
+      "loss": 1.355,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997242989514104e-05,
+      "loss": 1.3433,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997185852705327e-05,
+      "loss": 1.5048,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997128129951677e-05,
+      "loss": 1.5641,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997069821255403e-05,
+      "loss": 1.1493,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9997010926618786e-05,
+      "loss": 1.3701,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996951446044118e-05,
+      "loss": 0.6782,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999689137953373e-05,
+      "loss": 1.7044,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996830727089978e-05,
+      "loss": 1.6288,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996769488715218e-05,
+      "loss": 1.6771,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996707664411844e-05,
+      "loss": 1.9044,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999664525418228e-05,
+      "loss": 1.2813,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999658225802896e-05,
+      "loss": 2.107,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999651867595434e-05,
+      "loss": 0.7144,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999645450796091e-05,
+      "loss": 0.7223,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996389754051176e-05,
+      "loss": 1.6784,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999632441422766e-05,
+      "loss": 1.3566,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996258488492935e-05,
+      "loss": 1.9514,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996191976849557e-05,
+      "loss": 1.0533,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996124879300137e-05,
+      "loss": 1.5727,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9996057195847287e-05,
+      "loss": 2.1826,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995988926493656e-05,
+      "loss": 2.101,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995920071241908e-05,
+      "loss": 1.6808,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995850630094743e-05,
+      "loss": 1.188,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999578060305486e-05,
+      "loss": 0.986,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999570999012501e-05,
+      "loss": 1.7659,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995638791307942e-05,
+      "loss": 0.9505,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995567006606438e-05,
+      "loss": 1.2433,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995494636023306e-05,
+      "loss": 1.933,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999542167956137e-05,
+      "loss": 1.6114,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995348137223483e-05,
+      "loss": 1.5785,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999527400901252e-05,
+      "loss": 1.0934,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999519929493137e-05,
+      "loss": 1.8557,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995123994982962e-05,
+      "loss": 0.796,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9995048109170236e-05,
+      "loss": 1.0683,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999497163749615e-05,
+      "loss": 1.5434,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994894579963693e-05,
+      "loss": 1.4502,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994816936575883e-05,
+      "loss": 1.0679,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999473870733575e-05,
+      "loss": 1.5122,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994659892246347e-05,
+      "loss": 1.678,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994580491310756e-05,
+      "loss": 0.9277,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994500504532077e-05,
+      "loss": 1.3676,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994419931913442e-05,
+      "loss": 1.5379,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994338773457987e-05,
+      "loss": 1.0827,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99942570291689e-05,
+      "loss": 1.1223,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994174699049356e-05,
+      "loss": 1.3263,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994091783102587e-05,
+      "loss": 1.2968,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9994008281331822e-05,
+      "loss": 1.8248,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993924193740325e-05,
+      "loss": 1.0321,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993839520331388e-05,
+      "loss": 1.2999,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993754261108316e-05,
+      "loss": 1.1893,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999366841607444e-05,
+      "loss": 1.0665,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993581985233106e-05,
+      "loss": 0.9301,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993494968587705e-05,
+      "loss": 1.3674,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999340736614163e-05,
+      "loss": 0.9896,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99933191778983e-05,
+      "loss": 1.4181,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999323040386116e-05,
+      "loss": 0.4366,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993141044033687e-05,
+      "loss": 0.6926,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9993051098419372e-05,
+      "loss": 0.6361,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999296056702172e-05,
+      "loss": 1.635,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999286944984427e-05,
+      "loss": 0.4548,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992777746890592e-05,
+      "loss": 1.7004,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999268545816426e-05,
+      "loss": 0.3577,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999259258366888e-05,
+      "loss": 0.8682,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999249912340808e-05,
+      "loss": 0.742,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992405077385518e-05,
+      "loss": 0.8362,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992310445604865e-05,
+      "loss": 1.94,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992215228069816e-05,
+      "loss": 0.5628,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999211942478409e-05,
+      "loss": 0.8799,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9992023035751434e-05,
+      "loss": 0.7484,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991926060975613e-05,
+      "loss": 1.8482,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999182850046041e-05,
+      "loss": 1.3333,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999173035420965e-05,
+      "loss": 0.8427,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991631622227158e-05,
+      "loss": 0.9399,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991532304516794e-05,
+      "loss": 2.1583,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991432401082436e-05,
+      "loss": 1.941,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999133191192799e-05,
+      "loss": 1.0219,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9991230837057383e-05,
+      "loss": 1.0447,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999112917647456e-05,
+      "loss": 1.0022,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999102693018349e-05,
+      "loss": 0.9463,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990924098188178e-05,
+      "loss": 1.2173,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990820680492638e-05,
+      "loss": 1.3955,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990716677100907e-05,
+      "loss": 0.5988,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990612088017054e-05,
+      "loss": 0.9637,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990506913245154e-05,
+      "loss": 1.2098,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999040115278933e-05,
+      "loss": 1.8376,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999029480665371e-05,
+      "loss": 1.4246,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990187874842446e-05,
+      "loss": 0.594,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9990080357359712e-05,
+      "loss": 0.5691,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9989972254209718e-05,
+      "loss": 0.5711,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.998986356539668e-05,
+      "loss": 0.4687,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.998975429092485e-05,
+      "loss": 1.196,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9989644430798496e-05,
+      "loss": 1.2568,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9989533985021907e-05,
+      "loss": 1.4138,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9989422953599407e-05,
+      "loss": 0.9982,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9989311336535322e-05,
+      "loss": 0.7231,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998919913383402e-05,
+      "loss": 1.0025,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9989086345499883e-05,
+      "loss": 1.2351,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988972971537318e-05,
+      "loss": 1.5933,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988859011950752e-05,
+      "loss": 0.9698,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988744466744648e-05,
+      "loss": 1.0171,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988629335923467e-05,
+      "loss": 1.1541,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988513619491713e-05,
+      "loss": 0.8304,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988397317453912e-05,
+      "loss": 0.6023,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.99882804298146e-05,
+      "loss": 0.7358,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998816295657835e-05,
+      "loss": 1.6539,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9988044897749748e-05,
+      "loss": 1.806,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987926253333408e-05,
+      "loss": 0.5928,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987807023333966e-05,
+      "loss": 1.5635,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998768720775608e-05,
+      "loss": 0.7318,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987566806604428e-05,
+      "loss": 2.0159,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998744581988372e-05,
+      "loss": 0.6869,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998732424759868e-05,
+      "loss": 1.2176,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987202089754053e-05,
+      "loss": 0.6517,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9987079346354624e-05,
+      "loss": 0.4527,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986956017405178e-05,
+      "loss": 0.8132,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986832102910535e-05,
+      "loss": 0.5817,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986707602875535e-05,
+      "loss": 0.5389,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986582517305053e-05,
+      "loss": 2.2074,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998645684620396e-05,
+      "loss": 0.5065,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998633058957718e-05,
+      "loss": 1.2276,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986203747429644e-05,
+      "loss": 0.9436,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9986076319766303e-05,
+      "loss": 0.5685,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985948306592132e-05,
+      "loss": 1.1761,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985819707912138e-05,
+      "loss": 0.6521,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985690523731343e-05,
+      "loss": 0.8172,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985560754054798e-05,
+      "loss": 0.8829,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985430398887576e-05,
+      "loss": 1.2145,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985299458234758e-05,
+      "loss": 1.4001,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985167932101474e-05,
+      "loss": 1.0731,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9985035820492848e-05,
+      "loss": 0.9106,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9984903123414055e-05,
+      "loss": 0.4713,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9984769840870274e-05,
+      "loss": 1.4578,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998463597286671e-05,
+      "loss": 1.1789,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9984501519408596e-05,
+      "loss": 1.5964,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9984366480501182e-05,
+      "loss": 0.8166,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998423085614975e-05,
+      "loss": 0.8751,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.99840946463596e-05,
+      "loss": 0.8399,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998395785113604e-05,
+      "loss": 0.9901,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983820470484427e-05,
+      "loss": 1.3762,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998368250441013e-05,
+      "loss": 0.8362,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998354395291853e-05,
+      "loss": 0.8501,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983404816015047e-05,
+      "loss": 1.2197,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983265093705116e-05,
+      "loss": 1.0967,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9983124785994198e-05,
+      "loss": 1.1119,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982983892887765e-05,
+      "loss": 1.0417,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982842414391335e-05,
+      "loss": 1.0049,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998270035051043e-05,
+      "loss": 1.0261,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982557701250597e-05,
+      "loss": 0.8795,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998241446661741e-05,
+      "loss": 0.9432,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982270646616472e-05,
+      "loss": 0.8908,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9982126241253394e-05,
+      "loss": 0.7052,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9981981250533823e-05,
+      "loss": 0.6339,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9981835674463426e-05,
+      "loss": 0.4177,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998168951304788e-05,
+      "loss": 0.5685,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9981542766292905e-05,
+      "loss": 1.5354,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9981395434204237e-05,
+      "loss": 0.4832,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9981247516787624e-05,
+      "loss": 0.4892,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998109901404885e-05,
+      "loss": 0.6251,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980949925993715e-05,
+      "loss": 0.6186,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980800252628047e-05,
+      "loss": 1.4416,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980649993957686e-05,
+      "loss": 1.2439,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998049914998851e-05,
+      "loss": 0.7046,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980347720726417e-05,
+      "loss": 1.7189,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.998019570617731e-05,
+      "loss": 1.4698,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9980043106347133e-05,
+      "loss": 1.5325,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9979889921241853e-05,
+      "loss": 1.2784,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9979736150867456e-05,
+      "loss": 0.5592,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9979581795229942e-05,
+      "loss": 0.5754,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9979426854335348e-05,
+      "loss": 0.6586,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9979271328189726e-05,
+      "loss": 0.5556,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997911521679915e-05,
+      "loss": 0.6429,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978958520169727e-05,
+      "loss": 1.0417,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978801238307564e-05,
+      "loss": 0.8961,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978643371218822e-05,
+      "loss": 1.6161,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978484918909663e-05,
+      "loss": 0.6454,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978325881386275e-05,
+      "loss": 0.7034,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978166258654874e-05,
+      "loss": 0.5087,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9978006050721698e-05,
+      "loss": 0.727,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9977845257593e-05,
+      "loss": 2.3222,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9977683879275073e-05,
+      "loss": 0.7978,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9977521915774214e-05,
+      "loss": 0.9344,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997735936709675e-05,
+      "loss": 1.1998,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997719623324904e-05,
+      "loss": 0.954,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997703251423745e-05,
+      "loss": 0.8552,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9976868210068382e-05,
+      "loss": 0.9604,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997670332074825e-05,
+      "loss": 1.1759,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9976537846283497e-05,
+      "loss": 1.3332,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9976371786680593e-05,
+      "loss": 1.2434,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997620514194602e-05,
+      "loss": 1.2519,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9976037912086294e-05,
+      "loss": 0.946,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9975870097107947e-05,
+      "loss": 0.3742,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9975701697017535e-05,
+      "loss": 1.0043,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997553271182164e-05,
+      "loss": 1.0249,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9975363141526856e-05,
+      "loss": 0.2836,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997519298613982e-05,
+      "loss": 0.965,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997502224566717e-05,
+      "loss": 0.8841,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997485092011558e-05,
+      "loss": 0.8292,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9974679009491748e-05,
+      "loss": 0.8796,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9974506513802385e-05,
+      "loss": 0.3918,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9974333433054236e-05,
+      "loss": 0.4315,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9974159767254058e-05,
+      "loss": 0.6528,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9973985516408636e-05,
+      "loss": 0.8137,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9973810680524788e-05,
+      "loss": 0.4873,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997363525960933e-05,
+      "loss": 0.5699,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9973459253669124e-05,
+      "loss": 1.3434,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9973282662711047e-05,
+      "loss": 1.6767,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9973105486741997e-05,
+      "loss": 0.3152,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972927725768898e-05,
+      "loss": 0.6934,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972749379798693e-05,
+      "loss": 1.5048,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997257044883835e-05,
+      "loss": 0.6025,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972390932894858e-05,
+      "loss": 1.2076,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972210831975234e-05,
+      "loss": 1.192,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9972030146086518e-05,
+      "loss": 0.7083,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997184887523576e-05,
+      "loss": 0.7175,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9971667019430054e-05,
+      "loss": 0.497,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9971484578676493e-05,
+      "loss": 0.4727,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9971301552982212e-05,
+      "loss": 0.4923,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.997111794235436e-05,
+      "loss": 1.8059,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9970933746800115e-05,
+      "loss": 0.4714,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9970748966326668e-05,
+      "loss": 0.7034,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9970563600941242e-05,
+      "loss": 1.4454,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9970377650651076e-05,
+      "loss": 0.6268,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9970191115463434e-05,
+      "loss": 0.6552,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.997000399538561e-05,
+      "loss": 0.9575,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9969816290424913e-05,
+      "loss": 1.0113,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9969628000588673e-05,
+      "loss": 1.0923,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9969439125884246e-05,
+      "loss": 0.4388,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996924966631902e-05,
+      "loss": 1.0628,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9969059621900387e-05,
+      "loss": 0.5837,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996886899263578e-05,
+      "loss": 0.8959,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9968677778532644e-05,
+      "loss": 1.5586,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9968485979598447e-05,
+      "loss": 0.6831,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9968293595840688e-05,
+      "loss": 1.394,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9968100627266876e-05,
+      "loss": 1.0136,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9967907073884557e-05,
+      "loss": 0.9018,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9967712935701292e-05,
+      "loss": 0.9662,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9967518212724667e-05,
+      "loss": 1.1656,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9967322904962286e-05,
+      "loss": 2.1614,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996712701242178e-05,
+      "loss": 0.7729,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9966930535110803e-05,
+      "loss": 0.6784,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9966733473037038e-05,
+      "loss": 0.9859,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9966535826208176e-05,
+      "loss": 0.7989,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996633759463194e-05,
+      "loss": 0.9438,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9966138778316083e-05,
+      "loss": 0.666,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9965939377268363e-05,
+      "loss": 1.1277,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9965739391496575e-05,
+      "loss": 0.6542,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996553882100853e-05,
+      "loss": 0.6955,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9965337665812068e-05,
+      "loss": 0.7619,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9965135925915047e-05,
+      "loss": 0.542,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9964933601325354e-05,
+      "loss": 0.6837,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9964730692050885e-05,
+      "loss": 0.7209,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9964527198099568e-05,
+      "loss": 0.4562,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9964323119479357e-05,
+      "loss": 0.5815,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996411845619823e-05,
+      "loss": 0.7732,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9963913208264173e-05,
+      "loss": 0.5919,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9963707375685216e-05,
+      "loss": 0.6463,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9963500958469393e-05,
+      "loss": 0.3936,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9963293956624773e-05,
+      "loss": 0.3152,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9963086370159435e-05,
+      "loss": 2.0719,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996287819908151e-05,
+      "loss": 0.6227,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9962669443399107e-05,
+      "loss": 0.4547,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9962460103120394e-05,
+      "loss": 0.6693,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9962250178253553e-05,
+      "loss": 0.4912,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996203966880678e-05,
+      "loss": 0.6594,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99618285747883e-05,
+      "loss": 1.0379,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9961616896206363e-05,
+      "loss": 0.5169,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9961404633069235e-05,
+      "loss": 1.4968,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9961191785385218e-05,
+      "loss": 0.8975,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996097835316262e-05,
+      "loss": 0.5281,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996076433640978e-05,
+      "loss": 1.8229,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996054973513507e-05,
+      "loss": 0.8307,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.996033454934686e-05,
+      "loss": 0.4163,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9960118779053563e-05,
+      "loss": 0.5719,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9959902424263614e-05,
+      "loss": 0.6948,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9959685484985468e-05,
+      "loss": 0.7483,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9959467961227584e-05,
+      "loss": 0.6536,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9959249852998475e-05,
+      "loss": 0.6228,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995903116030666e-05,
+      "loss": 0.7778,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9958811883160687e-05,
+      "loss": 1.4674,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9958592021569114e-05,
+      "loss": 0.5596,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995837157554054e-05,
+      "loss": 1.1504,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995815054508357e-05,
+      "loss": 0.7398,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995792893020685e-05,
+      "loss": 0.9343,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9957706730919024e-05,
+      "loss": 0.8313,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9957483947228786e-05,
+      "loss": 0.6505,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9957260579144838e-05,
+      "loss": 1.2285,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99570366266759e-05,
+      "loss": 0.745,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995681208983073e-05,
+      "loss": 1.0312,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99565869686181e-05,
+      "loss": 0.5782,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99563612630468e-05,
+      "loss": 0.7239,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9956134973125652e-05,
+      "loss": 1.4854,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9955908098863498e-05,
+      "loss": 0.3932,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99556806402692e-05,
+      "loss": 1.4432,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9955452597351646e-05,
+      "loss": 1.0165,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9955223970119744e-05,
+      "loss": 1.4397,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995499475858243e-05,
+      "loss": 1.2083,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9954764962748657e-05,
+      "loss": 1.4136,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9954534582627404e-05,
+      "loss": 0.8764,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995430361822767e-05,
+      "loss": 0.888,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995407206955848e-05,
+      "loss": 0.9855,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995383993662888e-05,
+      "loss": 0.8684,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9953607219447947e-05,
+      "loss": 0.865,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9953373918024764e-05,
+      "loss": 1.0291,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9953140032368448e-05,
+      "loss": 0.8182,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9952905562488135e-05,
+      "loss": 1.1834,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9952670508392993e-05,
+      "loss": 0.776,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9952434870092204e-05,
+      "loss": 1.013,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9952198647594966e-05,
+      "loss": 0.6898,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995196184091052e-05,
+      "loss": 0.4478,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995172445004811e-05,
+      "loss": 0.3897,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995148647501701e-05,
+      "loss": 1.3706,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9951247915826528e-05,
+      "loss": 0.8476,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995100877248597e-05,
+      "loss": 0.4815,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995076904500469e-05,
+      "loss": 0.5357,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9950528733392057e-05,
+      "loss": 0.634,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.995028783765745e-05,
+      "loss": 1.0657,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9950046357810283e-05,
+      "loss": 1.3857,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9949804293859992e-05,
+      "loss": 0.5647,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994956164581604e-05,
+      "loss": 0.3962,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9949318413687904e-05,
+      "loss": 2.4865,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994907459748508e-05,
+      "loss": 2.1423,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9948830197217106e-05,
+      "loss": 2.1811,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9948585212893524e-05,
+      "loss": 0.6191,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9948339644523906e-05,
+      "loss": 0.8452,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994809349211784e-05,
+      "loss": 0.8377,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9947846755684958e-05,
+      "loss": 0.7201,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9947599435234888e-05,
+      "loss": 2.2258,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99473515307773e-05,
+      "loss": 1.1178,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9947103042321874e-05,
+      "loss": 1.2007,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9946853969878326e-05,
+      "loss": 0.8652,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9946604313456377e-05,
+      "loss": 0.6479,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994635407306579e-05,
+      "loss": 1.0756,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9946103248716337e-05,
+      "loss": 1.1533,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9945851840417823e-05,
+      "loss": 0.9687,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9945599848180064e-05,
+      "loss": 0.6074,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9945347272012906e-05,
+      "loss": 0.9488,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9945094111926225e-05,
+      "loss": 0.8962,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9944840367929904e-05,
+      "loss": 1.0089,
+      "step": 434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994458604003386e-05,
+      "loss": 0.9512,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994433112824803e-05,
+      "loss": 0.7305,
+      "step": 436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9944075632582374e-05,
+      "loss": 2.0012,
+      "step": 437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9943819553046874e-05,
+      "loss": 0.9199,
+      "step": 438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994356288965153e-05,
+      "loss": 0.6743,
+      "step": 439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9943305642406378e-05,
+      "loss": 0.6948,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9943047811321462e-05,
+      "loss": 0.7123,
+      "step": 441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.994278939640686e-05,
+      "loss": 0.6621,
+      "step": 442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9942530397672664e-05,
+      "loss": 0.9399,
+      "step": 443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9942270815128998e-05,
+      "loss": 0.8196,
+      "step": 444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9942010648786004e-05,
+      "loss": 0.6354,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.994174989865384e-05,
+      "loss": 1.0374,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9941488564742702e-05,
+      "loss": 1.1116,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99412266470628e-05,
+      "loss": 0.4563,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.994096414562436e-05,
+      "loss": 0.4312,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.994070106043764e-05,
+      "loss": 0.6014,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9940437391512923e-05,
+      "loss": 1.4282,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.994017313886051e-05,
+      "loss": 0.5244,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9939908302490722e-05,
+      "loss": 0.4331,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9939642882413905e-05,
+      "loss": 0.4933,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9939376878640437e-05,
+      "loss": 0.6209,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99391102911807e-05,
+      "loss": 0.6337,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993884312004512e-05,
+      "loss": 0.5614,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9938575365244134e-05,
+      "loss": 0.7013,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9938307026788196e-05,
+      "loss": 0.7677,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9938038104687795e-05,
+      "loss": 0.345,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993776859895344e-05,
+      "loss": 1.0035,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993749850959566e-05,
+      "loss": 0.7061,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9937227836625e-05,
+      "loss": 1.6106,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9936956580052036e-05,
+      "loss": 1.6806,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993668473988738e-05,
+      "loss": 1.3948,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993641231614164e-05,
+      "loss": 0.8046,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9936139308825464e-05,
+      "loss": 1.8274,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9935865717949517e-05,
+      "loss": 0.9786,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9935591543524492e-05,
+      "loss": 0.7645,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9935316785561098e-05,
+      "loss": 2.2112,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9935041444070063e-05,
+      "loss": 1.4942,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993476551906216e-05,
+      "loss": 0.7551,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9934489010548155e-05,
+      "loss": 1.6196,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9934211918538864e-05,
+      "loss": 0.8211,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9933934243045104e-05,
+      "loss": 0.805,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9933655984077726e-05,
+      "loss": 0.7882,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9933377141647603e-05,
+      "loss": 1.2728,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9933097715765627e-05,
+      "loss": 0.7287,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993281770644272e-05,
+      "loss": 0.5,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9932537113689812e-05,
+      "loss": 0.3758,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993225593751788e-05,
+      "loss": 0.6484,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9931974177937898e-05,
+      "loss": 1.0228,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993169183496088e-05,
+      "loss": 0.7513,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993140890859786e-05,
+      "loss": 0.6959,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9931125398859884e-05,
+      "loss": 0.6079,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9930841305758034e-05,
+      "loss": 0.2486,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.993055662930341e-05,
+      "loss": 0.2645,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9930271369507133e-05,
+      "loss": 0.5093,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9929985526380347e-05,
+      "loss": 1.4629,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9929699099934217e-05,
+      "loss": 1.0136,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9929412090179943e-05,
+      "loss": 0.8585,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9929124497128736e-05,
+      "loss": 1.0361,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992883632079182e-05,
+      "loss": 0.6975,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992854756118047e-05,
+      "loss": 1.2529,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9928258218305965e-05,
+      "loss": 1.3943,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9927968292179605e-05,
+      "loss": 0.7662,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9927677782812717e-05,
+      "loss": 0.9204,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992738669021665e-05,
+      "loss": 0.454,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992709501440279e-05,
+      "loss": 0.7117,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9926802755382517e-05,
+      "loss": 0.8462,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9926509913167257e-05,
+      "loss": 1.3573,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9926216487768448e-05,
+      "loss": 1.3715,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9925922479197558e-05,
+      "loss": 1.2366,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9925627887466076e-05,
+      "loss": 0.8451,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9925332712585505e-05,
+      "loss": 0.7138,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9925036954567382e-05,
+      "loss": 0.5336,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9924740613423264e-05,
+      "loss": 0.7848,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9924443689164724e-05,
+      "loss": 0.697,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9924146181803363e-05,
+      "loss": 0.533,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9923848091350814e-05,
+      "loss": 0.6276,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992354941781871e-05,
+      "loss": 1.0411,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992325016121873e-05,
+      "loss": 1.7481,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992295032156256e-05,
+      "loss": 0.5855,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9922649898861922e-05,
+      "loss": 0.9673,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992234889312855e-05,
+      "loss": 0.579,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99220473043742e-05,
+      "loss": 0.4671,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9921745132610663e-05,
+      "loss": 1.8876,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992144237784974e-05,
+      "loss": 0.5153,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9921139040103257e-05,
+      "loss": 0.4073,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.992083511938307e-05,
+      "loss": 0.5955,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9920530615701054e-05,
+      "loss": 0.3187,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9920225529069105e-05,
+      "loss": 1.6439,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9919919859499144e-05,
+      "loss": 0.669,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9919613607003112e-05,
+      "loss": 1.3484,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991930677159297e-05,
+      "loss": 1.7646,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991899935328071e-05,
+      "loss": 2.359,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991869135207835e-05,
+      "loss": 1.2353,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9918382767997914e-05,
+      "loss": 1.8271,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991807360105146e-05,
+      "loss": 1.3019,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9917763851251072e-05,
+      "loss": 1.2683,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9917453518608845e-05,
+      "loss": 0.3887,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991714260313691e-05,
+      "loss": 1.323,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991683110484741e-05,
+      "loss": 0.509,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991651902375252e-05,
+      "loss": 0.8729,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9916206359864434e-05,
+      "loss": 0.385,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9915893113195356e-05,
+      "loss": 1.0441,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991557928375754e-05,
+      "loss": 0.2471,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9915264871563238e-05,
+      "loss": 0.9545,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991494987662474e-05,
+      "loss": 0.8913,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9914634298954346e-05,
+      "loss": 0.9647,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991431813856439e-05,
+      "loss": 0.7956,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9914001395467226e-05,
+      "loss": 0.7014,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9913684069675227e-05,
+      "loss": 0.8964,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991336616120079e-05,
+      "loss": 0.5545,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9913047670056338e-05,
+      "loss": 0.7459,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9912728596254314e-05,
+      "loss": 0.8983,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9912408939807182e-05,
+      "loss": 0.6721,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9912088700727435e-05,
+      "loss": 0.5629,
+      "step": 548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991176787902758e-05,
+      "loss": 0.5394,
+      "step": 549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991144647472016e-05,
+      "loss": 0.8497,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991112448781772e-05,
+      "loss": 1.5589,
+      "step": 551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991080191833285e-05,
+      "loss": 0.3598,
+      "step": 552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991047876627815e-05,
+      "loss": 1.7842,
+      "step": 553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.991015503166625e-05,
+      "loss": 0.6,
+      "step": 554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9909830714509786e-05,
+      "loss": 0.5258,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9909505814821442e-05,
+      "loss": 0.7464,
+      "step": 556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9909180332613904e-05,
+      "loss": 0.5002,
+      "step": 557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9908854267899897e-05,
+      "loss": 0.4183,
+      "step": 558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990852762069215e-05,
+      "loss": 1.0157,
+      "step": 559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9908200391003432e-05,
+      "loss": 0.4675,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990787257884653e-05,
+      "loss": 0.326,
+      "step": 561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990754418423424e-05,
+      "loss": 0.1573,
+      "step": 562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990721520717941e-05,
+      "loss": 0.7617,
+      "step": 563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9906885647694875e-05,
+      "loss": 1.2678,
+      "step": 564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9906555505793527e-05,
+      "loss": 0.5196,
+      "step": 565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9906224781488254e-05,
+      "loss": 0.9255,
+      "step": 566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.990589347479198e-05,
+      "loss": 0.4447,
+      "step": 567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9905561585717652e-05,
+      "loss": 0.6399,
+      "step": 568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9905229114278235e-05,
+      "loss": 0.635,
+      "step": 569
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9904896060486722e-05,
+      "loss": 1.7407,
+      "step": 570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9904562424356118e-05,
+      "loss": 0.6132,
+      "step": 571
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9904228205899466e-05,
+      "loss": 1.2094,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9903893405129822e-05,
+      "loss": 0.8097,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.990355802206026e-05,
+      "loss": 0.7666,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9903222056703898e-05,
+      "loss": 0.5268,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9902885509073845e-05,
+      "loss": 2.2642,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9902548379183262e-05,
+      "loss": 1.0031,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9902210667045318e-05,
+      "loss": 0.3674,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9901872372673205e-05,
+      "loss": 0.3887,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9901533496080145e-05,
+      "loss": 0.4521,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9901194037279368e-05,
+      "loss": 0.6965,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9900853996284152e-05,
+      "loss": 1.3967,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.990051337310777e-05,
+      "loss": 0.8525,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9900172167763537e-05,
+      "loss": 1.2399,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9899830380264783e-05,
+      "loss": 0.308,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9899488010624857e-05,
+      "loss": 1.2806,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9899145058857138e-05,
+      "loss": 0.7203,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989880152497503e-05,
+      "loss": 0.4809,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989845740899195e-05,
+      "loss": 0.8846,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9898112710921343e-05,
+      "loss": 0.2855,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9897767430776674e-05,
+      "loss": 0.7223,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9897421568571442e-05,
+      "loss": 0.9001,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989707512431915e-05,
+      "loss": 0.9306,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9896728098033345e-05,
+      "loss": 0.5997,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989638048972757e-05,
+      "loss": 1.0173,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989603229941542e-05,
+      "loss": 0.6624,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989568352711049e-05,
+      "loss": 1.0158,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9895334172826415e-05,
+      "loss": 0.6602,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9894984236576838e-05,
+      "loss": 0.5566,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9894633718375433e-05,
+      "loss": 0.7845,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9894282618235894e-05,
+      "loss": 0.7186,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9893930936171937e-05,
+      "loss": 1.0431,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9893578672197305e-05,
+      "loss": 1.6958,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9893225826325765e-05,
+      "loss": 1.359,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9892872398571096e-05,
+      "loss": 0.5175,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989251838894711e-05,
+      "loss": 0.7264,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9892163797467636e-05,
+      "loss": 0.4438,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.989180862414653e-05,
+      "loss": 1.6727,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9891452868997662e-05,
+      "loss": 1.6486,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9891096532034945e-05,
+      "loss": 0.2046,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9890739613272293e-05,
+      "loss": 1.2469,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9890382112723648e-05,
+      "loss": 0.1026,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9890024030402983e-05,
+      "loss": 0.7062,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9889665366324288e-05,
+      "loss": 1.0852,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9889306120501575e-05,
+      "loss": 0.5904,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9888946292948873e-05,
+      "loss": 0.6595,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9888585883680254e-05,
+      "loss": 0.3239,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988822489270979e-05,
+      "loss": 0.7006,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988786332005159e-05,
+      "loss": 0.6156,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9887501165719778e-05,
+      "loss": 0.5077,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.98871384297285e-05,
+      "loss": 1.4623,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9886775112091937e-05,
+      "loss": 1.7288,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9886411212824274e-05,
+      "loss": 1.4099,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9886046731939737e-05,
+      "loss": 1.4417,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988568166945256e-05,
+      "loss": 2.0503,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9885316025377013e-05,
+      "loss": 1.4548,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9884949799727373e-05,
+      "loss": 0.3641,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9884582992517957e-05,
+      "loss": 1.0049,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9884215603763095e-05,
+      "loss": 1.4502,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9883847633477136e-05,
+      "loss": 0.1728,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988347908167446e-05,
+      "loss": 1.7016,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9883109948369462e-05,
+      "loss": 0.7355,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9882740233576573e-05,
+      "loss": 0.9046,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988236993731023e-05,
+      "loss": 0.8667,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9881999059584905e-05,
+      "loss": 1.1116,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9881627600415086e-05,
+      "loss": 0.8021,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9881255559815288e-05,
+      "loss": 1.1957,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9880882937800044e-05,
+      "loss": 0.7746,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9880509734383916e-05,
+      "loss": 0.2391,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.988013594958148e-05,
+      "loss": 1.0846,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9879761583407347e-05,
+      "loss": 0.8564,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9879386635876137e-05,
+      "loss": 1.0347,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9879011107002504e-05,
+      "loss": 0.6615,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9878634996801114e-05,
+      "loss": 0.742,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9878258305286673e-05,
+      "loss": 0.7098,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9877881032473887e-05,
+      "loss": 0.5812,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.98775031783775e-05,
+      "loss": 0.7159,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9877124743012278e-05,
+      "loss": 1.2813,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9876745726393e-05,
+      "loss": 0.4308,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9876366128534483e-05,
+      "loss": 0.3918,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987598594945155e-05,
+      "loss": 0.5597,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987560518915906e-05,
+      "loss": 0.4578,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9875223847671886e-05,
+      "loss": 1.7367,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9874841925004928e-05,
+      "loss": 1.786,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9874459421173107e-05,
+      "loss": 1.8256,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987407633619137e-05,
+      "loss": 1.8965,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9873692670074686e-05,
+      "loss": 0.4684,
+      "step": 657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987330842283804e-05,
+      "loss": 0.4181,
+      "step": 658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9872923594496447e-05,
+      "loss": 0.3227,
+      "step": 659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9872538185064942e-05,
+      "loss": 0.4891,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9872152194558583e-05,
+      "loss": 1.5391,
+      "step": 661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9871765622992453e-05,
+      "loss": 0.3398,
+      "step": 662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9871378470381654e-05,
+      "loss": 0.5134,
+      "step": 663
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9870990736741307e-05,
+      "loss": 1.5999,
+      "step": 664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987060242208657e-05,
+      "loss": 0.5861,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.987021352643261e-05,
+      "loss": 0.6661,
+      "step": 666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9869824049794616e-05,
+      "loss": 1.3571,
+      "step": 667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9869433992187817e-05,
+      "loss": 0.6852,
+      "step": 668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9869043353627442e-05,
+      "loss": 1.0393,
+      "step": 669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986865213412876e-05,
+      "loss": 0.6089,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9868260333707055e-05,
+      "loss": 0.563,
+      "step": 671
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9867867952377624e-05,
+      "loss": 1.4144,
+      "step": 672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9867474990155817e-05,
+      "loss": 1.4321,
+      "step": 673
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9867081447056975e-05,
+      "loss": 0.9185,
+      "step": 674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9866687323096473e-05,
+      "loss": 1.323,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986629261828971e-05,
+      "loss": 0.7928,
+      "step": 676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9865897332652112e-05,
+      "loss": 0.8524,
+      "step": 677
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986550146619912e-05,
+      "loss": 1.077,
+      "step": 678
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9865105018946206e-05,
+      "loss": 1.4092,
+      "step": 679
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9864707990908847e-05,
+      "loss": 1.7962,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986431038210257e-05,
+      "loss": 0.7166,
+      "step": 681
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9863912192542898e-05,
+      "loss": 1.0479,
+      "step": 682
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9863513422245394e-05,
+      "loss": 1.5032,
+      "step": 683
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9863114071225636e-05,
+      "loss": 1.1514,
+      "step": 684
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9862714139499224e-05,
+      "loss": 1.0627,
+      "step": 685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.986231362708179e-05,
+      "loss": 0.9127,
+      "step": 686
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9861912533988983e-05,
+      "loss": 1.1849,
+      "step": 687
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9861510860236462e-05,
+      "loss": 1.0263,
+      "step": 688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9861108605839932e-05,
+      "loss": 1.0371,
+      "step": 689
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9860705770815107e-05,
+      "loss": 0.5711,
+      "step": 690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9860302355177725e-05,
+      "loss": 1.4121,
+      "step": 691
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9859898358943543e-05,
+      "loss": 0.548,
+      "step": 692
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.985949378212835e-05,
+      "loss": 0.6833,
+      "step": 693
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9859088624747955e-05,
+      "loss": 1.225,
+      "step": 694
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9858682886818183e-05,
+      "loss": 0.9258,
+      "step": 695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.985827656835489e-05,
+      "loss": 0.7998,
+      "step": 696
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.985786966937395e-05,
+      "loss": 0.5323,
+      "step": 697
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9857462189891257e-05,
+      "loss": 0.4087,
+      "step": 698
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9857054129922735e-05,
+      "loss": 0.6691,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9856645489484326e-05,
+      "loss": 0.668,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9856236268591995e-05,
+      "loss": 0.634,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9855826467261735e-05,
+      "loss": 0.2726,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.985541608550955e-05,
+      "loss": 0.6595,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.985500512335148e-05,
+      "loss": 0.4793,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9854593580803578e-05,
+      "loss": 0.5189,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.985418145788192e-05,
+      "loss": 0.472,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9853768754602613e-05,
+      "loss": 0.3937,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9853355470981782e-05,
+      "loss": 0.479,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9852941607035573e-05,
+      "loss": 0.7048,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9852527162780152e-05,
+      "loss": 0.4148,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9852112138231716e-05,
+      "loss": 2.2502,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9851696533406476e-05,
+      "loss": 0.5901,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9851280348320675e-05,
+      "loss": 0.3577,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.985086358299057e-05,
+      "loss": 1.4286,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9850446237432446e-05,
+      "loss": 0.2868,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9850028311662605e-05,
+      "loss": 0.9149,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984960980569738e-05,
+      "loss": 0.2816,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984919071955312e-05,
+      "loss": 0.7468,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.98487710532462e-05,
+      "loss": 0.7477,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9848350806793014e-05,
+      "loss": 0.8761,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9847929980209985e-05,
+      "loss": 0.7187,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9847508573513554e-05,
+      "loss": 0.9171,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9847086586720182e-05,
+      "loss": 1.2125,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984666401984636e-05,
+      "loss": 0.7155,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984624087290859e-05,
+      "loss": 1.2762,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984581714592342e-05,
+      "loss": 1.3035,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984539283890739e-05,
+      "loss": 1.543,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9844967951877082e-05,
+      "loss": 0.233,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9844542484849103e-05,
+      "loss": 1.2084,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9844116437840067e-05,
+      "loss": 1.2515,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984368981086663e-05,
+      "loss": 0.3182,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9843262603945447e-05,
+      "loss": 0.3854,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984283481709322e-05,
+      "loss": 0.6602,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984240645032666e-05,
+      "loss": 0.8864,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9841977503662504e-05,
+      "loss": 0.3788,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.984154797711751e-05,
+      "loss": 1.7247,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9841117870708458e-05,
+      "loss": 0.199,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9840687184452153e-05,
+      "loss": 0.5846,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9840255918365424e-05,
+      "loss": 0.6338,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9839824072465123e-05,
+      "loss": 0.8534,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9839391646768114e-05,
+      "loss": 0.8796,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.98389586412913e-05,
+      "loss": 0.9679,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9838525056051596e-05,
+      "loss": 1.5252,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983809089106594e-05,
+      "loss": 0.8063,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9837656146351298e-05,
+      "loss": 1.1196,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983722082192466e-05,
+      "loss": 0.8219,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983678491780302e-05,
+      "loss": 0.7,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983634843400342e-05,
+      "loss": 0.9573,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9835911370542918e-05,
+      "loss": 0.7963,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983547372743858e-05,
+      "loss": 0.594,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9835035504707506e-05,
+      "loss": 1.2713,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9834596702366824e-05,
+      "loss": 0.7728,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9834157320433674e-05,
+      "loss": 1.5683,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983371735892522e-05,
+      "loss": 1.5749,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9833276817858655e-05,
+      "loss": 0.3943,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983283569725119e-05,
+      "loss": 0.4681,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9832393997120064e-05,
+      "loss": 0.3198,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9831951717482522e-05,
+      "loss": 0.5249,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983150885835586e-05,
+      "loss": 0.3546,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.983106541975737e-05,
+      "loss": 0.7309,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9830621401704385e-05,
+      "loss": 0.6192,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9830176804214245e-05,
+      "loss": 0.4876,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9829731627304325e-05,
+      "loss": 0.544,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9829285870992018e-05,
+      "loss": 0.9635,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9828839535294738e-05,
+      "loss": 0.8108,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982839262022992e-05,
+      "loss": 1.5943,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9827945125815038e-05,
+      "loss": 0.592,
+      "step": 767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9827497052067563e-05,
+      "loss": 0.8206,
+      "step": 768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9827048399005003e-05,
+      "loss": 0.4024,
+      "step": 769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9826599166644897e-05,
+      "loss": 0.3235,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9826149355004783e-05,
+      "loss": 2.3684,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9825698964102248e-05,
+      "loss": 0.797,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982524799395488e-05,
+      "loss": 1.3842,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.98247964445803e-05,
+      "loss": 0.8443,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982434431599615e-05,
+      "loss": 0.6233,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9823891608220098e-05,
+      "loss": 0.783,
+      "step": 776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982343832126983e-05,
+      "loss": 0.6199,
+      "step": 777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9822984455163056e-05,
+      "loss": 1.1252,
+      "step": 778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982253000991751e-05,
+      "loss": 0.6941,
+      "step": 779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982207498555094e-05,
+      "loss": 0.881,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9821619382081134e-05,
+      "loss": 1.3315,
+      "step": 781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9821163199525886e-05,
+      "loss": 0.703,
+      "step": 782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9820706437903024e-05,
+      "loss": 0.9335,
+      "step": 783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982024909723039e-05,
+      "loss": 1.71,
+      "step": 784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9819791177525855e-05,
+      "loss": 1.4178,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981933267880731e-05,
+      "loss": 0.7543,
+      "step": 786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9818873601092665e-05,
+      "loss": 0.6392,
+      "step": 787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9818413944399857e-05,
+      "loss": 0.6653,
+      "step": 788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981795370874685e-05,
+      "loss": 0.6664,
+      "step": 789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981749289415162e-05,
+      "loss": 0.7722,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9817031500632177e-05,
+      "loss": 0.5506,
+      "step": 791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981656952820654e-05,
+      "loss": 0.9201,
+      "step": 792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9816106976892765e-05,
+      "loss": 1.5196,
+      "step": 793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981564384670892e-05,
+      "loss": 1.3355,
+      "step": 794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9815180137673107e-05,
+      "loss": 0.3653,
+      "step": 795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9814715849803428e-05,
+      "loss": 0.6236,
+      "step": 796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981425098311804e-05,
+      "loss": 1.1041,
+      "step": 797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9813785537635093e-05,
+      "loss": 0.9457,
+      "step": 798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981331951337278e-05,
+      "loss": 0.6899,
+      "step": 799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.981285291034931e-05,
+      "loss": 0.619,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9812385728582903e-05,
+      "loss": 1.5847,
+      "step": 801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9811917968091823e-05,
+      "loss": 0.4447,
+      "step": 802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9811449628894334e-05,
+      "loss": 0.5579,
+      "step": 803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9810980711008744e-05,
+      "loss": 0.3273,
+      "step": 804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9810511214453372e-05,
+      "loss": 1.262,
+      "step": 805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9810041139246558e-05,
+      "loss": 0.4803,
+      "step": 806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980957048540667e-05,
+      "loss": 0.4497,
+      "step": 807
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9809099252952098e-05,
+      "loss": 0.5847,
+      "step": 808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980862744190125e-05,
+      "loss": 0.3942,
+      "step": 809
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9808155052272562e-05,
+      "loss": 0.6023,
+      "step": 810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9807682084084497e-05,
+      "loss": 0.4318,
+      "step": 811
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980720853735552e-05,
+      "loss": 0.7647,
+      "step": 812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980673441210414e-05,
+      "loss": 0.618,
+      "step": 813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9806259708348877e-05,
+      "loss": 0.8,
+      "step": 814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9805784426108287e-05,
+      "loss": 0.6529,
+      "step": 815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9805308565400932e-05,
+      "loss": 0.6403,
+      "step": 816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980483212624541e-05,
+      "loss": 0.4375,
+      "step": 817
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9804355108660326e-05,
+      "loss": 1.1409,
+      "step": 818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9803877512664328e-05,
+      "loss": 1.1392,
+      "step": 819
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9803399338276066e-05,
+      "loss": 1.6683,
+      "step": 820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980292058551423e-05,
+      "loss": 0.7568,
+      "step": 821
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9802441254397523e-05,
+      "loss": 1.0071,
+      "step": 822
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.980196134494467e-05,
+      "loss": 0.7413,
+      "step": 823
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9801480857174423e-05,
+      "loss": 0.813,
+      "step": 824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9800999791105556e-05,
+      "loss": 1.7055,
+      "step": 825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9800518146756864e-05,
+      "loss": 1.2648,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9800035924147167e-05,
+      "loss": 1.0113,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9799553123295295e-05,
+      "loss": 0.5148,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9799069744220125e-05,
+      "loss": 1.2144,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.979858578694054e-05,
+      "loss": 1.1318,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9798101251475437e-05,
+      "loss": 0.5631,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9797616137843758e-05,
+      "loss": 1.8622,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9797130446064452e-05,
+      "loss": 0.404,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97966441761565e-05,
+      "loss": 0.6511,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97961573281389e-05,
+      "loss": 1.2265,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9795669902030666e-05,
+      "loss": 0.8539,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9795181897850852e-05,
+      "loss": 0.6207,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9794693315618512e-05,
+      "loss": 0.6527,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9794204155352745e-05,
+      "loss": 1.6232,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9793714417072664e-05,
+      "loss": 0.8124,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9793224100797395e-05,
+      "loss": 1.0858,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9792733206546105e-05,
+      "loss": 0.6503,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9792241734337963e-05,
+      "loss": 0.6626,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9791749684192176e-05,
+      "loss": 0.8628,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9791257056127967e-05,
+      "loss": 0.7054,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9790763850164586e-05,
+      "loss": 0.7555,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9790270066321297e-05,
+      "loss": 1.295,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9789775704617397e-05,
+      "loss": 0.4315,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97892807650722e-05,
+      "loss": 0.5092,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978878524770505e-05,
+      "loss": 0.4967,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9788289152535293e-05,
+      "loss": 0.362,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978779247958232e-05,
+      "loss": 1.68,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978729522886554e-05,
+      "loss": 0.3965,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978679740040437e-05,
+      "loss": 1.3394,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9786298994218272e-05,
+      "loss": 1.5597,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978580001032671e-05,
+      "loss": 1.6328,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9785300448749183e-05,
+      "loss": 0.4359,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978480030950521e-05,
+      "loss": 0.2888,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978429959261433e-05,
+      "loss": 0.4472,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97837982980961e-05,
+      "loss": 0.4026,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978329642597012e-05,
+      "loss": 0.2959,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9782793976255985e-05,
+      "loss": 1.7475,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978229094897334e-05,
+      "loss": 0.4577,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9781787344141824e-05,
+      "loss": 0.6763,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9781283161781117e-05,
+      "loss": 1.3219,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.978077840191092e-05,
+      "loss": 0.6436,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9780273064550954e-05,
+      "loss": 0.6054,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9779767149720967e-05,
+      "loss": 0.6869,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9779260657440718e-05,
+      "loss": 0.8125,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9778753587729997e-05,
+      "loss": 0.6636,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9778245940608613e-05,
+      "loss": 1.3119,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9777737716096408e-05,
+      "loss": 0.7573,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9777228914213234e-05,
+      "loss": 0.9714,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9776719534978967e-05,
+      "loss": 1.4042,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9776209578413517e-05,
+      "loss": 0.9159,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9775699044536797e-05,
+      "loss": 1.691,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9775187933368766e-05,
+      "loss": 0.5379,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977467624492938e-05,
+      "loss": 0.8546,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9774163979238645e-05,
+      "loss": 0.4428,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9773651136316565e-05,
+      "loss": 1.3104,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9773137716183175e-05,
+      "loss": 0.5066,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9772623718858546e-05,
+      "loss": 1.0935,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9772109144362753e-05,
+      "loss": 1.2162,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97715939927159e-05,
+      "loss": 0.6259,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9771078263938113e-05,
+      "loss": 1.488,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9770561958049547e-05,
+      "loss": 0.9851,
+      "step": 886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.977004507507037e-05,
+      "loss": 0.3197,
+      "step": 887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976952761502078e-05,
+      "loss": 0.2379,
+      "step": 888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976900957792099e-05,
+      "loss": 1.0112,
+      "step": 889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9768490963791242e-05,
+      "loss": 0.7181,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97679717726518e-05,
+      "loss": 0.7368,
+      "step": 891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9767452004522947e-05,
+      "loss": 0.7365,
+      "step": 892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976693165942499e-05,
+      "loss": 1.0988,
+      "step": 893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976641073737826e-05,
+      "loss": 0.6229,
+      "step": 894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976588923840311e-05,
+      "loss": 0.8845,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976536716251992e-05,
+      "loss": 0.4048,
+      "step": 896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976484450974908e-05,
+      "loss": 0.6141,
+      "step": 897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9764321280111014e-05,
+      "loss": 0.669,
+      "step": 898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976379747362616e-05,
+      "loss": 0.3173,
+      "step": 899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976327309031499e-05,
+      "loss": 0.7908,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9762748130197987e-05,
+      "loss": 0.5781,
+      "step": 901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.976222259329566e-05,
+      "loss": 1.3542,
+      "step": 902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9761696479628546e-05,
+      "loss": 1.2742,
+      "step": 903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9761169789217203e-05,
+      "loss": 1.2507,
+      "step": 904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9760642522082204e-05,
+      "loss": 0.345,
+      "step": 905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9760114678244148e-05,
+      "loss": 1.1334,
+      "step": 906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9759586257723663e-05,
+      "loss": 0.2558,
+      "step": 907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9759057260541395e-05,
+      "loss": 0.6338,
+      "step": 908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9758527686718007e-05,
+      "loss": 1.5195,
+      "step": 909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975799753627419e-05,
+      "loss": 1.1326,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9757466809230664e-05,
+      "loss": 0.6455,
+      "step": 911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9756935505608154e-05,
+      "loss": 0.5076,
+      "step": 912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9756403625427427e-05,
+      "loss": 0.65,
+      "step": 913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9755871168709266e-05,
+      "loss": 0.3626,
+      "step": 914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975533813547446e-05,
+      "loss": 2.1881,
+      "step": 915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9754804525743853e-05,
+      "loss": 0.5765,
+      "step": 916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9754270339538275e-05,
+      "loss": 0.9097,
+      "step": 917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975373557687861e-05,
+      "loss": 1.8667,
+      "step": 918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9753200237785752e-05,
+      "loss": 1.8226,
+      "step": 919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9752664322280606e-05,
+      "loss": 0.8978,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9752127830384118e-05,
+      "loss": 0.4416,
+      "step": 921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975159076211725e-05,
+      "loss": 1.0086,
+      "step": 922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975105311750098e-05,
+      "loss": 1.1869,
+      "step": 923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.975051489655632e-05,
+      "loss": 0.8989,
+      "step": 924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9749976099304292e-05,
+      "loss": 0.4997,
+      "step": 925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9749436725765953e-05,
+      "loss": 0.9735,
+      "step": 926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9748896775962375e-05,
+      "loss": 1.2392,
+      "step": 927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974835624991465e-05,
+      "loss": 0.7634,
+      "step": 928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97478151476439e-05,
+      "loss": 0.4242,
+      "step": 929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9747273469171267e-05,
+      "loss": 1.2992,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974673121451791e-05,
+      "loss": 0.6901,
+      "step": 931
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974618838370502e-05,
+      "loss": 1.2557,
+      "step": 932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9745644976753803e-05,
+      "loss": 1.2332,
+      "step": 933
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9745100993685495e-05,
+      "loss": 0.8965,
+      "step": 934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9744556434521342e-05,
+      "loss": 0.7779,
+      "step": 935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9744011299282617e-05,
+      "loss": 0.882,
+      "step": 936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9743465587990636e-05,
+      "loss": 0.6179,
+      "step": 937
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9742919300666703e-05,
+      "loss": 0.3224,
+      "step": 938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974237243733217e-05,
+      "loss": 1.0528,
+      "step": 939
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9741824998008397e-05,
+      "loss": 0.3327,
+      "step": 940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.974127698271678e-05,
+      "loss": 0.5338,
+      "step": 941
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9740728391478728e-05,
+      "loss": 0.8662,
+      "step": 942
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9740179224315673e-05,
+      "loss": 0.4942,
+      "step": 943
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.973962948124907e-05,
+      "loss": 1.2835,
+      "step": 944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9739079162300397e-05,
+      "loss": 0.8988,
+      "step": 945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.973852826749116e-05,
+      "loss": 0.8749,
+      "step": 946
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.973797679684288e-05,
+      "loss": 0.4562,
+      "step": 947
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.97374247503771e-05,
+      "loss": 1.0236,
+      "step": 948
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9736872128115397e-05,
+      "loss": 1.2376,
+      "step": 949
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.973631893007935e-05,
+      "loss": 0.9201,
+      "step": 950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9735765156290584e-05,
+      "loss": 0.5472,
+      "step": 951
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.973521080677073e-05,
+      "loss": 0.3462,
+      "step": 952
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9734655881541444e-05,
+      "loss": 2.0933,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9734100380624413e-05,
+      "loss": 0.486,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9733544304041334e-05,
+      "loss": 0.3259,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9732987651813943e-05,
+      "loss": 0.5677,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.973243042396398e-05,
+      "loss": 0.6158,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.973187262051322e-05,
+      "loss": 1.1308,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.973131424148345e-05,
+      "loss": 0.6829,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9730755286896497e-05,
+      "loss": 0.358,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9730195756774196e-05,
+      "loss": 1.0626,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9729635651138397e-05,
+      "loss": 0.4229,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9729074970010996e-05,
+      "loss": 0.3389,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9728513713413897e-05,
+      "loss": 0.6345,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9727951881369024e-05,
+      "loss": 1.9377,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9727389473898335e-05,
+      "loss": 1.8096,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9726826491023793e-05,
+      "loss": 0.3543,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9726262932767403e-05,
+      "loss": 2.2821,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972569879915118e-05,
+      "loss": 0.541,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972513409019717e-05,
+      "loss": 1.5056,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9724568805927423e-05,
+      "loss": 1.3669,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972400294636404e-05,
+      "loss": 1.396,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9723436511529122e-05,
+      "loss": 0.8665,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.97228695014448e-05,
+      "loss": 0.7107,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9722301916133225e-05,
+      "loss": 1.1807,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9721733755616578e-05,
+      "loss": 0.639,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9721165019917055e-05,
+      "loss": 1.4205,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.972059570905688e-05,
+      "loss": 0.4109,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9720025823058288e-05,
+      "loss": 0.6681,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971945536194355e-05,
+      "loss": 1.6604,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9718884325734954e-05,
+      "loss": 1.2697,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971831271445481e-05,
+      "loss": 2.0165,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9717740528125454e-05,
+      "loss": 1.2419,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971716776676923e-05,
+      "loss": 1.1515,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971659443040853e-05,
+      "loss": 0.4748,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971602051906575e-05,
+      "loss": 0.9662,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9715446032763308e-05,
+      "loss": 0.5095,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971487097152366e-05,
+      "loss": 0.2575,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971429533536926e-05,
+      "loss": 1.3578,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971371912432261e-05,
+      "loss": 1.4004,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9713142338406214e-05,
+      "loss": 0.904,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9712564977642614e-05,
+      "loss": 1.1616,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9711987042054366e-05,
+      "loss": 0.8187,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9711408531664048e-05,
+      "loss": 0.6015,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.971082944649426e-05,
+      "loss": 0.7121,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9710249786567635e-05,
+      "loss": 1.1438,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9709669551906817e-05,
+      "loss": 0.5096,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9709088742534473e-05,
+      "loss": 0.8163,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.97085073584733e-05,
+      "loss": 0.7102,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9707925399746004e-05,
+      "loss": 0.5073,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9707342866375335e-05,
+      "loss": 0.4154,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9706759758384047e-05,
+      "loss": 1.2107,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970617607579492e-05,
+      "loss": 1.4705,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9705591818630755e-05,
+      "loss": 0.6147,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970500698691439e-05,
+      "loss": 0.3169,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970442158066867e-05,
+      "loss": 0.5675,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9703835599916463e-05,
+      "loss": 1.6907,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9703249044680667e-05,
+      "loss": 1.1817,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.97026619149842e-05,
+      "loss": 0.3653,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9702074210849998e-05,
+      "loss": 1.6681,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970148593230103e-05,
+      "loss": 1.0009,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970089707936027e-05,
+      "loss": 1.7078,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.970030765205073e-05,
+      "loss": 1.7713,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9699717650395437e-05,
+      "loss": 0.4053,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9699127074417447e-05,
+      "loss": 1.1484,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9698535924139826e-05,
+      "loss": 1.542,
+      "step": 1016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9697944199585685e-05,
+      "loss": 0.6734,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9697351900778123e-05,
+      "loss": 1.668,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.96967590277403e-05,
+      "loss": 0.35,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9696165580495368e-05,
+      "loss": 0.3562,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9695571559066517e-05,
+      "loss": 0.5572,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9694976963476956e-05,
+      "loss": 0.4967,
+      "step": 1022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9694381793749912e-05,
+      "loss": 1.1416,
+      "step": 1023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9693786049908643e-05,
+      "loss": 0.5963,
+      "step": 1024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.969318973197643e-05,
+      "loss": 0.6696,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.969259283997656e-05,
+      "loss": 0.605,
+      "step": 1026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9691995373932357e-05,
+      "loss": 0.5898,
+      "step": 1027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.969139733386717e-05,
+      "loss": 0.2714,
+      "step": 1028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9690798719804358e-05,
+      "loss": 1.0202,
+      "step": 1029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.969019953176731e-05,
+      "loss": 0.7574,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968959976977944e-05,
+      "loss": 1.8513,
+      "step": 1031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968899943386418e-05,
+      "loss": 1.1988,
+      "step": 1032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9688398524044983e-05,
+      "loss": 1.1998,
+      "step": 1033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968779704034533e-05,
+      "loss": 0.7656,
+      "step": 1034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9687194982788716e-05,
+      "loss": 0.6949,
+      "step": 1035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968659235139867e-05,
+      "loss": 1.5652,
+      "step": 1036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968598914619873e-05,
+      "loss": 1.804,
+      "step": 1037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968538536721247e-05,
+      "loss": 0.5333,
+      "step": 1038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968478101446348e-05,
+      "loss": 1.0028,
+      "step": 1039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9684176087975363e-05,
+      "loss": 0.2949,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9683570587771767e-05,
+      "loss": 1.1036,
+      "step": 1041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9682964513876337e-05,
+      "loss": 1.049,
+      "step": 1042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968235786631276e-05,
+      "loss": 0.8665,
+      "step": 1043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9681750645104732e-05,
+      "loss": 0.7265,
+      "step": 1044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9681142850275983e-05,
+      "loss": 0.7417,
+      "step": 1045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.968053448185026e-05,
+      "loss": 0.5648,
+      "step": 1046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9679925539851325e-05,
+      "loss": 0.6168,
+      "step": 1047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9679316024302977e-05,
+      "loss": 0.4255,
+      "step": 1048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9678705935229028e-05,
+      "loss": 0.7113,
+      "step": 1049
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.967809527265331e-05,
+      "loss": 0.4342,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9677484036599692e-05,
+      "loss": 1.2927,
+      "step": 1051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.967687222709204e-05,
+      "loss": 0.6717,
+      "step": 1052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9676259844154272e-05,
+      "loss": 1.7756,
+      "step": 1053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.967564688781031e-05,
+      "loss": 0.3523,
+      "step": 1054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9675033358084097e-05,
+      "loss": 0.3135,
+      "step": 1055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9674419254999608e-05,
+      "loss": 0.8578,
+      "step": 1056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9673804578580835e-05,
+      "loss": 1.4994,
+      "step": 1057
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9673189328851793e-05,
+      "loss": 0.3568,
+      "step": 1058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9672573505836523e-05,
+      "loss": 0.3992,
+      "step": 1059
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9671957109559086e-05,
+      "loss": 0.5025,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.967134014004356e-05,
+      "loss": 1.4415,
+      "step": 1061
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9670722597314055e-05,
+      "loss": 1.9866,
+      "step": 1062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9670104481394692e-05,
+      "loss": 0.7966,
+      "step": 1063
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9669485792309625e-05,
+      "loss": 0.97,
+      "step": 1064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966886653008303e-05,
+      "loss": 0.6182,
+      "step": 1065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9668246694739097e-05,
+      "loss": 0.2732,
+      "step": 1066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9667626286302044e-05,
+      "loss": 0.6499,
+      "step": 1067
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966700530479611e-05,
+      "loss": 0.9649,
+      "step": 1068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9666383750245563e-05,
+      "loss": 0.6422,
+      "step": 1069
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966576162267468e-05,
+      "loss": 0.8698,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966513892210777e-05,
+      "loss": 2.5466,
+      "step": 1071
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966451564856916e-05,
+      "loss": 1.7299,
+      "step": 1072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9663891802083204e-05,
+      "loss": 0.6469,
+      "step": 1073
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.966326738267428e-05,
+      "loss": 1.1788,
+      "step": 1074
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9662642390366773e-05,
+      "loss": 0.3199,
+      "step": 1075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9662016825185113e-05,
+      "loss": 0.7521,
+      "step": 1076
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9661390687153737e-05,
+      "loss": 0.4732,
+      "step": 1077
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9660763976297105e-05,
+      "loss": 0.8171,
+      "step": 1078
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9660136692639704e-05,
+      "loss": 1.0517,
+      "step": 1079
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.965950883620605e-05,
+      "loss": 0.8345,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9658880407020666e-05,
+      "loss": 0.5156,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9658251405108103e-05,
+      "loss": 0.6967,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.965762183049294e-05,
+      "loss": 0.863,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.965699168319978e-05,
+      "loss": 0.8213,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9656360963253234e-05,
+      "loss": 1.065,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9655729670677947e-05,
+      "loss": 0.9767,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9655097805498587e-05,
+      "loss": 0.5648,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9654465367739835e-05,
+      "loss": 0.8036,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9653832357426408e-05,
+      "loss": 0.9657,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.965319877458303e-05,
+      "loss": 0.8643,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9652564619234467e-05,
+      "loss": 2.1452,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9651929891405482e-05,
+      "loss": 1.162,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9651294591120884e-05,
+      "loss": 0.5128,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9650658718405487e-05,
+      "loss": 0.7721,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9650022273284138e-05,
+      "loss": 1.2618,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9649385255781707e-05,
+      "loss": 1.3351,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9648747665923074e-05,
+      "loss": 0.8243,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.964810950373316e-05,
+      "loss": 0.9235,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9647470769236893e-05,
+      "loss": 0.5806,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9646831462459227e-05,
+      "loss": 0.4169,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9646191583425143e-05,
+      "loss": 1.5669,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9645551132159636e-05,
+      "loss": 1.0958,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9644910108687736e-05,
+      "loss": 1.617,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9644268513034482e-05,
+      "loss": 0.7934,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9643626345224947e-05,
+      "loss": 0.4627,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9642983605284215e-05,
+      "loss": 1.0924,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96423402932374e-05,
+      "loss": 0.5411,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.964169640910964e-05,
+      "loss": 0.727,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9641051952926088e-05,
+      "loss": 0.6668,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.964040692471192e-05,
+      "loss": 2.16,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9639761324492346e-05,
+      "loss": 0.5242,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9639115152292583e-05,
+      "loss": 1.8692,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963846840813788e-05,
+      "loss": 1.6905,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9637821092053505e-05,
+      "loss": 0.5168,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9637173204064746e-05,
+      "loss": 0.7483,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9636524744196922e-05,
+      "loss": 0.7127,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9635875712475364e-05,
+      "loss": 0.6951,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963522610892543e-05,
+      "loss": 1.3558,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9634575933572505e-05,
+      "loss": 1.9952,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9633925186441987e-05,
+      "loss": 0.8256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96332738675593e-05,
+      "loss": 2.2108,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963262197694989e-05,
+      "loss": 0.5868,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9631969514639237e-05,
+      "loss": 1.9505,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.963131648065282e-05,
+      "loss": 0.5286,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9630662875016158e-05,
+      "loss": 1.2769,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9630008697754796e-05,
+      "loss": 0.7354,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962935394889428e-05,
+      "loss": 1.4173,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96286986284602e-05,
+      "loss": 1.5885,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9628042736478153e-05,
+      "loss": 0.9529,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9627386272973764e-05,
+      "loss": 1.2641,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962672923797269e-05,
+      "loss": 0.6902,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96260716315006e-05,
+      "loss": 0.9052,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962541345358318e-05,
+      "loss": 1.8353,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962475470424615e-05,
+      "loss": 1.3783,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9624095383515247e-05,
+      "loss": 0.8578,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962343549141623e-05,
+      "loss": 0.5528,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962277502797488e-05,
+      "loss": 0.5385,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9622113993217008e-05,
+      "loss": 0.8393,
+      "step": 1138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.962145238716843e-05,
+      "loss": 1.0703,
+      "step": 1139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9620790209855007e-05,
+      "loss": 0.9921,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96201274613026e-05,
+      "loss": 1.2148,
+      "step": 1141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9619464141537106e-05,
+      "loss": 0.8847,
+      "step": 1142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9618800250584447e-05,
+      "loss": 0.7373,
+      "step": 1143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9618135788470556e-05,
+      "loss": 1.2354,
+      "step": 1144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9617470755221393e-05,
+      "loss": 0.6791,
+      "step": 1145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9616805150862945e-05,
+      "loss": 0.7588,
+      "step": 1146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9616138975421214e-05,
+      "loss": 0.6413,
+      "step": 1147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9615472228922233e-05,
+      "loss": 0.7169,
+      "step": 1148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9614804911392043e-05,
+      "loss": 0.6246,
+      "step": 1149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9614137022856723e-05,
+      "loss": 0.4508,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.961346856334236e-05,
+      "loss": 1.0149,
+      "step": 1151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9612799532875088e-05,
+      "loss": 1.2356,
+      "step": 1152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9612129931481028e-05,
+      "loss": 1.782,
+      "step": 1153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9611459759186348e-05,
+      "loss": 0.2636,
+      "step": 1154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9610789016017235e-05,
+      "loss": 1.375,
+      "step": 1155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.961011770199989e-05,
+      "loss": 0.6662,
+      "step": 1156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9609445817160546e-05,
+      "loss": 0.354,
+      "step": 1157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9608773361525452e-05,
+      "loss": 0.6458,
+      "step": 1158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9608100335120883e-05,
+      "loss": 0.3883,
+      "step": 1159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960742673797313e-05,
+      "loss": 1.755,
+      "step": 1160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960675257010851e-05,
+      "loss": 0.3166,
+      "step": 1161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9606077831553367e-05,
+      "loss": 1.0506,
+      "step": 1162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9605402522334063e-05,
+      "loss": 0.3107,
+      "step": 1163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960472664247698e-05,
+      "loss": 0.3944,
+      "step": 1164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960405019200853e-05,
+      "loss": 1.1968,
+      "step": 1165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9603373170955133e-05,
+      "loss": 1.2399,
+      "step": 1166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9602695579343248e-05,
+      "loss": 0.5626,
+      "step": 1167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960201741719935e-05,
+      "loss": 0.7878,
+      "step": 1168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.960133868454993e-05,
+      "loss": 0.465,
+      "step": 1169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9600659381421508e-05,
+      "loss": 0.589,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9599979507840623e-05,
+      "loss": 0.6733,
+      "step": 1171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9599299063833843e-05,
+      "loss": 1.5168,
+      "step": 1172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9598618049427748e-05,
+      "loss": 0.4928,
+      "step": 1173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9597936464648944e-05,
+      "loss": 1.771,
+      "step": 1174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9597254309524067e-05,
+      "loss": 0.9649,
+      "step": 1175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9596571584079766e-05,
+      "loss": 0.8248,
+      "step": 1176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9595888288342713e-05,
+      "loss": 0.4776,
+      "step": 1177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9595204422339608e-05,
+      "loss": 0.76,
+      "step": 1178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.959451998609717e-05,
+      "loss": 0.5787,
+      "step": 1179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.959383497964214e-05,
+      "loss": 0.5342,
+      "step": 1180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9593149403001275e-05,
+      "loss": 0.9981,
+      "step": 1181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9592463256201368e-05,
+      "loss": 0.9645,
+      "step": 1182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9591776539269226e-05,
+      "loss": 0.6016,
+      "step": 1183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.959108925223168e-05,
+      "loss": 0.6713,
+      "step": 1184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9590401395115577e-05,
+      "loss": 0.8833,
+      "step": 1185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9589712967947797e-05,
+      "loss": 0.6516,
+      "step": 1186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9589023970755232e-05,
+      "loss": 0.6889,
+      "step": 1187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958833440356481e-05,
+      "loss": 0.5332,
+      "step": 1188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9587644266403468e-05,
+      "loss": 0.8818,
+      "step": 1189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9586953559298167e-05,
+      "loss": 0.2779,
+      "step": 1190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9586262282275892e-05,
+      "loss": 0.7304,
+      "step": 1191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9585570435363657e-05,
+      "loss": 0.4294,
+      "step": 1192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9584878018588496e-05,
+      "loss": 0.6925,
+      "step": 1193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958418503197745e-05,
+      "loss": 0.4897,
+      "step": 1194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9583491475557606e-05,
+      "loss": 0.6344,
+      "step": 1195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958279734935605e-05,
+      "loss": 0.3937,
+      "step": 1196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9582102653399912e-05,
+      "loss": 0.5848,
+      "step": 1197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9581407387716328e-05,
+      "loss": 1.0005,
+      "step": 1198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9580711552332468e-05,
+      "loss": 0.5019,
+      "step": 1199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.958001514727551e-05,
+      "loss": 0.4152,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.957931817257267e-05,
+      "loss": 0.417,
+      "step": 1201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9578620628251177e-05,
+      "loss": 1.8595,
+      "step": 1202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9577922514338284e-05,
+      "loss": 0.3722,
+      "step": 1203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9577223830861267e-05,
+      "loss": 0.4582,
+      "step": 1204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9576524577847418e-05,
+      "loss": 0.7204,
+      "step": 1205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9575824755324068e-05,
+      "loss": 0.711,
+      "step": 1206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.957512436331855e-05,
+      "loss": 0.383,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9574423401858233e-05,
+      "loss": 0.8472,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.95737218709705e-05,
+      "loss": 0.5598,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9573019770682764e-05,
+      "loss": 0.3186,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.957231710102246e-05,
+      "loss": 0.3496,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9571613862017033e-05,
+      "loss": 0.543,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9570910053693963e-05,
+      "loss": 0.7284,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9570205676080747e-05,
+      "loss": 1.5101,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9569500729204907e-05,
+      "loss": 0.6011,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956879521309398e-05,
+      "loss": 0.5083,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9568089127775542e-05,
+      "loss": 0.7422,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9567382473277164e-05,
+      "loss": 0.5539,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956667524962647e-05,
+      "loss": 0.7566,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9565967456851087e-05,
+      "loss": 1.2861,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9565259094978662e-05,
+      "loss": 0.2661,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956455016403688e-05,
+      "loss": 1.6929,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9563840664053437e-05,
+      "loss": 0.8437,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9563130595056045e-05,
+      "loss": 0.5078,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956241995707246e-05,
+      "loss": 0.3183,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9561708750130437e-05,
+      "loss": 0.586,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9560996974257766e-05,
+      "loss": 0.6784,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956028462948226e-05,
+      "loss": 0.6976,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9559571715831746e-05,
+      "loss": 1.5885,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9558858233334078e-05,
+      "loss": 0.381,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955814418201713e-05,
+      "loss": 1.0055,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9557429561908808e-05,
+      "loss": 1.3315,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9556714373037023e-05,
+      "loss": 0.5883,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9555998615429725e-05,
+      "loss": 1.4516,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9555282289114875e-05,
+      "loss": 1.0419,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955456539412046e-05,
+      "loss": 0.2968,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9553847930474495e-05,
+      "loss": 1.2235,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9553129898205002e-05,
+      "loss": 0.6151,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9552411297340042e-05,
+      "loss": 0.9976,
+      "step": 1239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955169212790769e-05,
+      "loss": 1.0928,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955097238993604e-05,
+      "loss": 0.9453,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9550252083453222e-05,
+      "loss": 1.0134,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9549531208487365e-05,
+      "loss": 0.4681,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9548809765066645e-05,
+      "loss": 0.7143,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.954808775321924e-05,
+      "loss": 0.7748,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.954736517297337e-05,
+      "loss": 0.4162,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9546642024357257e-05,
+      "loss": 0.5954,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.954591830739916e-05,
+      "loss": 0.7301,
+      "step": 1248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.954519402212735e-05,
+      "loss": 0.5922,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9544469168570127e-05,
+      "loss": 0.5446,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9543743746755817e-05,
+      "loss": 1.778,
+      "step": 1251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9543017756712758e-05,
+      "loss": 0.482,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.954229119846931e-05,
+      "loss": 0.9307,
+      "step": 1253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9541564072053867e-05,
+      "loss": 0.9294,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9540836377494834e-05,
+      "loss": 1.3481,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9540108114820643e-05,
+      "loss": 1.3782,
+      "step": 1256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9539379284059753e-05,
+      "loss": 0.4584,
+      "step": 1257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9538649885240628e-05,
+      "loss": 0.8017,
+      "step": 1258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9537919918391776e-05,
+      "loss": 1.2064,
+      "step": 1259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9537189383541715e-05,
+      "loss": 0.4929,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9536458280718986e-05,
+      "loss": 1.1475,
+      "step": 1261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9535726609952153e-05,
+      "loss": 0.568,
+      "step": 1262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9534994371269795e-05,
+      "loss": 0.586,
+      "step": 1263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9534261564700538e-05,
+      "loss": 1.3627,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9533528190273e-05,
+      "loss": 0.6593,
+      "step": 1265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.953279424801584e-05,
+      "loss": 0.3195,
+      "step": 1266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9532059737957732e-05,
+      "loss": 1.2726,
+      "step": 1267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9531324660127367e-05,
+      "loss": 0.1843,
+      "step": 1268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9530589014553478e-05,
+      "loss": 0.317,
+      "step": 1269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952985280126479e-05,
+      "loss": 0.4769,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9529116020290087e-05,
+      "loss": 0.6372,
+      "step": 1271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952837867165814e-05,
+      "loss": 0.6854,
+      "step": 1272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952764075539776e-05,
+      "loss": 0.6515,
+      "step": 1273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9526902271537783e-05,
+      "loss": 1.0995,
+      "step": 1274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9526163220107055e-05,
+      "loss": 1.2649,
+      "step": 1275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9525423601134458e-05,
+      "loss": 0.5883,
+      "step": 1276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9524683414648885e-05,
+      "loss": 0.5272,
+      "step": 1277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952394266067926e-05,
+      "loss": 0.4794,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9523201339254515e-05,
+      "loss": 0.6579,
+      "step": 1279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952245945040362e-05,
+      "loss": 0.2071,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952171699415556e-05,
+      "loss": 0.5371,
+      "step": 1281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9520973970539347e-05,
+      "loss": 0.8454,
+      "step": 1282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9520230379584003e-05,
+      "loss": 1.4241,
+      "step": 1283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9519486221318585e-05,
+      "loss": 0.6224,
+      "step": 1284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.951874149577217e-05,
+      "loss": 1.0195,
+      "step": 1285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9517996202973848e-05,
+      "loss": 0.9216,
+      "step": 1286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9517250342952743e-05,
+      "loss": 0.4575,
+      "step": 1287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9516503915737995e-05,
+      "loss": 0.3958,
+      "step": 1288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9515756921358766e-05,
+      "loss": 0.2415,
+      "step": 1289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9515009359844244e-05,
+      "loss": 1.1445,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9514261231223633e-05,
+      "loss": 0.8088,
+      "step": 1291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9513512535526163e-05,
+      "loss": 1.1884,
+      "step": 1292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.951276327278109e-05,
+      "loss": 0.7106,
+      "step": 1293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.951201344301768e-05,
+      "loss": 0.8076,
+      "step": 1294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9511263046265236e-05,
+      "loss": 0.7623,
+      "step": 1295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.951051208255308e-05,
+      "loss": 0.586,
+      "step": 1296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950976055191054e-05,
+      "loss": 0.6868,
+      "step": 1297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950900845436699e-05,
+      "loss": 1.1144,
+      "step": 1298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9508255789951804e-05,
+      "loss": 0.8376,
+      "step": 1299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9507502558694398e-05,
+      "loss": 0.552,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9506748760624197e-05,
+      "loss": 1.9408,
+      "step": 1301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9505994395770654e-05,
+      "loss": 0.6013,
+      "step": 1302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950523946416324e-05,
+      "loss": 1.6026,
+      "step": 1303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950448396583145e-05,
+      "loss": 0.8581,
+      "step": 1304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9503727900804805e-05,
+      "loss": 0.3194,
+      "step": 1305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9502971269112843e-05,
+      "loss": 0.4894,
+      "step": 1306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950221407078513e-05,
+      "loss": 0.712,
+      "step": 1307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9501456305851236e-05,
+      "loss": 1.8312,
+      "step": 1308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950069797434078e-05,
+      "loss": 0.4037,
+      "step": 1309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949993907628339e-05,
+      "loss": 0.6554,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949917961170871e-05,
+      "loss": 0.4087,
+      "step": 1311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9498419580646416e-05,
+      "loss": 0.6605,
+      "step": 1312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9497658983126208e-05,
+      "loss": 0.5463,
+      "step": 1313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949689781917779e-05,
+      "loss": 1.4506,
+      "step": 1314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9496136088830916e-05,
+      "loss": 0.836,
+      "step": 1315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949537379211533e-05,
+      "loss": 0.3085,
+      "step": 1316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949461092906083e-05,
+      "loss": 0.6161,
+      "step": 1317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9493847499697218e-05,
+      "loss": 0.8223,
+      "step": 1318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9493083504054315e-05,
+      "loss": 2.0476,
+      "step": 1319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949231894216198e-05,
+      "loss": 1.5277,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.949155381405007e-05,
+      "loss": 0.5173,
+      "step": 1321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9490788119748493e-05,
+      "loss": 1.4032,
+      "step": 1322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9490021859287164e-05,
+      "loss": 1.0612,
+      "step": 1323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9489255032696013e-05,
+      "loss": 0.5816,
+      "step": 1324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9488487640005005e-05,
+      "loss": 0.719,
+      "step": 1325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9487719681244125e-05,
+      "loss": 1.2932,
+      "step": 1326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.948695115644337e-05,
+      "loss": 0.5244,
+      "step": 1327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.948618206563277e-05,
+      "loss": 1.1173,
+      "step": 1328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9485412408842377e-05,
+      "loss": 0.7744,
+      "step": 1329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.948464218610226e-05,
+      "loss": 0.7935,
+      "step": 1330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9483871397442507e-05,
+      "loss": 1.3032,
+      "step": 1331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9483100042893236e-05,
+      "loss": 0.8015,
+      "step": 1332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9482328122484587e-05,
+      "loss": 0.921,
+      "step": 1333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9481555636246717e-05,
+      "loss": 0.1837,
+      "step": 1334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9480782584209806e-05,
+      "loss": 0.4446,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9480008966404065e-05,
+      "loss": 1.0314,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9479234782859705e-05,
+      "loss": 1.2373,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9478460033606987e-05,
+      "loss": 0.5974,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9477684718676173e-05,
+      "loss": 0.8127,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9476908838097557e-05,
+      "loss": 1.2623,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9476132391901455e-05,
+      "loss": 0.7918,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9475355380118198e-05,
+      "loss": 0.7069,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.947457780277815e-05,
+      "loss": 0.4991,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9473799659911686e-05,
+      "loss": 0.9927,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9473020951549215e-05,
+      "loss": 0.9312,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9472241677721155e-05,
+      "loss": 0.6167,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9471461838457956e-05,
+      "loss": 0.297,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9470681433790082e-05,
+      "loss": 0.8748,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.946990046374803e-05,
+      "loss": 0.5157,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9469118928362304e-05,
+      "loss": 0.4372,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.946833682766345e-05,
+      "loss": 1.2572,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9467554161682018e-05,
+      "loss": 1.3855,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9466770930448588e-05,
+      "loss": 0.923,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.946598713399376e-05,
+      "loss": 1.5483,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.946520277234816e-05,
+      "loss": 0.5076,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9464417845542432e-05,
+      "loss": 0.2716,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9463632353607244e-05,
+      "loss": 1.9882,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9462846296573287e-05,
+      "loss": 0.728,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9462059674471263e-05,
+      "loss": 1.5421,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.946127248733192e-05,
+      "loss": 1.523,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9460484735186004e-05,
+      "loss": 0.8143,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9459696418064296e-05,
+      "loss": 0.8846,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9458907535997593e-05,
+      "loss": 0.5695,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.945811808901672e-05,
+      "loss": 1.0482,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9457328077152518e-05,
+      "loss": 0.5423,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9456537500435857e-05,
+      "loss": 0.4522,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9455746358897623e-05,
+      "loss": 0.9508,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9454954652568724e-05,
+      "loss": 0.4086,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9454162381480098e-05,
+      "loss": 0.6008,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.945336954566269e-05,
+      "loss": 1.5018,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9452576145147485e-05,
+      "loss": 0.7999,
+      "step": 1371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9451782179965477e-05,
+      "loss": 0.5242,
+      "step": 1372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.945098765014769e-05,
+      "loss": 1.2109,
+      "step": 1373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9450192555725167e-05,
+      "loss": 1.3797,
+      "step": 1374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9449396896728965e-05,
+      "loss": 1.0927,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9448600673190176e-05,
+      "loss": 1.3311,
+      "step": 1376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944780388513991e-05,
+      "loss": 0.6536,
+      "step": 1377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9447006532609294e-05,
+      "loss": 0.2838,
+      "step": 1378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9446208615629483e-05,
+      "loss": 1.2497,
+      "step": 1379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944541013423165e-05,
+      "loss": 0.6417,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9444611088447e-05,
+      "loss": 1.5591,
+      "step": 1381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944381147830674e-05,
+      "loss": 0.9842,
+      "step": 1382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944301130384212e-05,
+      "loss": 1.2233,
+      "step": 1383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.94422105650844e-05,
+      "loss": 0.9635,
+      "step": 1384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.944140926206486e-05,
+      "loss": 0.9751,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9440607394814816e-05,
+      "loss": 0.8252,
+      "step": 1386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9439804963365594e-05,
+      "loss": 0.2414,
+      "step": 1387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.943900196774854e-05,
+      "loss": 1.2538,
+      "step": 1388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9438198407995035e-05,
+      "loss": 0.9333,
+      "step": 1389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9437394284136474e-05,
+      "loss": 0.8328,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9436589596204267e-05,
+      "loss": 0.9895,
+      "step": 1391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9435784344229863e-05,
+      "loss": 0.5559,
+      "step": 1392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.943497852824471e-05,
+      "loss": 0.5549,
+      "step": 1393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9434172148280307e-05,
+      "loss": 1.2594,
+      "step": 1394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.943336520436815e-05,
+      "loss": 0.4606,
+      "step": 1395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9432557696539766e-05,
+      "loss": 0.9253,
+      "step": 1396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9431749624826714e-05,
+      "loss": 0.5128,
+      "step": 1397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9430940989260556e-05,
+      "loss": 0.4263,
+      "step": 1398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9430131789872887e-05,
+      "loss": 0.3989,
+      "step": 1399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9429322026695327e-05,
+      "loss": 0.6705,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.942851169975951e-05,
+      "loss": 0.4344,
+      "step": 1401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.94277008090971e-05,
+      "loss": 1.3959,
+      "step": 1402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.942688935473977e-05,
+      "loss": 2.1608,
+      "step": 1403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9426077336719232e-05,
+      "loss": 1.4637,
+      "step": 1404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9425264755067208e-05,
+      "loss": 0.5725,
+      "step": 1405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9424451609815446e-05,
+      "loss": 0.2943,
+      "step": 1406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9423637900995724e-05,
+      "loss": 0.4109,
+      "step": 1407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9422823628639822e-05,
+      "loss": 1.0794,
+      "step": 1408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9422008792779558e-05,
+      "loss": 0.7394,
+      "step": 1409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9421193393446766e-05,
+      "loss": 0.5915,
+      "step": 1410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9420377430673313e-05,
+      "loss": 0.5043,
+      "step": 1411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941956090449107e-05,
+      "loss": 0.7773,
+      "step": 1412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941874381493194e-05,
+      "loss": 0.4839,
+      "step": 1413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941792616202785e-05,
+      "loss": 0.5498,
+      "step": 1414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9417107945810747e-05,
+      "loss": 0.8234,
+      "step": 1415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9416289166312593e-05,
+      "loss": 0.3611,
+      "step": 1416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941546982356538e-05,
+      "loss": 0.5162,
+      "step": 1417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941464991760112e-05,
+      "loss": 0.6935,
+      "step": 1418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9413829448451855e-05,
+      "loss": 0.3754,
+      "step": 1419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9413008416149628e-05,
+      "loss": 0.3827,
+      "step": 1420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9412186820726528e-05,
+      "loss": 2.1373,
+      "step": 1421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9411364662214643e-05,
+      "loss": 0.3203,
+      "step": 1422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.941054194064611e-05,
+      "loss": 0.2166,
+      "step": 1423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.940971865605306e-05,
+      "loss": 0.5579,
+      "step": 1424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.940889480846767e-05,
+      "loss": 1.2799,
+      "step": 1425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9408070397922118e-05,
+      "loss": 0.3621,
+      "step": 1426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9407245424448618e-05,
+      "loss": 1.5365,
+      "step": 1427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9406419888079404e-05,
+      "loss": 1.0011,
+      "step": 1428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9405593788846726e-05,
+      "loss": 1.4861,
+      "step": 1429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9404767126782865e-05,
+      "loss": 0.6627,
+      "step": 1430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.940393990192011e-05,
+      "loss": 1.5898,
+      "step": 1431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9403112114290792e-05,
+      "loss": 0.4038,
+      "step": 1432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9402283763927246e-05,
+      "loss": 0.8141,
+      "step": 1433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9401454850861838e-05,
+      "loss": 1.6562,
+      "step": 1434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9400625375126958e-05,
+      "loss": 1.2068,
+      "step": 1435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9399795336755003e-05,
+      "loss": 0.6543,
+      "step": 1436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939896473577841e-05,
+      "loss": 0.8095,
+      "step": 1437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9398133572229637e-05,
+      "loss": 0.7689,
+      "step": 1438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9397301846141147e-05,
+      "loss": 0.7035,
+      "step": 1439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939646955754544e-05,
+      "loss": 0.8459,
+      "step": 1440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939563670647503e-05,
+      "loss": 1.2301,
+      "step": 1441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939480329296247e-05,
+      "loss": 0.5539,
+      "step": 1442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.93939693170403e-05,
+      "loss": 0.6721,
+      "step": 1443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9393134778741124e-05,
+      "loss": 0.4063,
+      "step": 1444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939229967809754e-05,
+      "loss": 0.1876,
+      "step": 1445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.939146401514217e-05,
+      "loss": 1.0806,
+      "step": 1446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9390627789907673e-05,
+      "loss": 0.3278,
+      "step": 1447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938979100242672e-05,
+      "loss": 0.6855,
+      "step": 1448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9388953652731996e-05,
+      "loss": 0.7775,
+      "step": 1449
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938811574085622e-05,
+      "loss": 0.6516,
+      "step": 1450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9387277266832135e-05,
+      "loss": 1.904,
+      "step": 1451
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9386438230692497e-05,
+      "loss": 0.3755,
+      "step": 1452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9385598632470083e-05,
+      "loss": 0.855,
+      "step": 1453
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9384758472197702e-05,
+      "loss": 0.6784,
+      "step": 1454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9383917749908177e-05,
+      "loss": 0.7558,
+      "step": 1455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938307646563436e-05,
+      "loss": 0.782,
+      "step": 1456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938223461940911e-05,
+      "loss": 0.7342,
+      "step": 1457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938139221126533e-05,
+      "loss": 0.2389,
+      "step": 1458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.938054924123592e-05,
+      "loss": 0.3569,
+      "step": 1459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.937970570935383e-05,
+      "loss": 0.347,
+      "step": 1460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9378861615652007e-05,
+      "loss": 0.8844,
+      "step": 1461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9378016960163433e-05,
+      "loss": 0.544,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9377171742921107e-05,
+      "loss": 0.5957,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9376325963958057e-05,
+      "loss": 0.4126,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.937547962330732e-05,
+      "loss": 0.3758,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9374632721001972e-05,
+      "loss": 0.1735,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9373785257075095e-05,
+      "loss": 0.4674,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9372937231559796e-05,
+      "loss": 0.9027,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9372088644489223e-05,
+      "loss": 1.6022,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9371239495896517e-05,
+      "loss": 0.2155,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9370389785814856e-05,
+      "loss": 0.7435,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9369539514277444e-05,
+      "loss": 1.7817,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9368688681317496e-05,
+      "loss": 1.3732,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.936783728696826e-05,
+      "loss": 0.4613,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9366985331262992e-05,
+      "loss": 1.1934,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.936613281423499e-05,
+      "loss": 1.0418,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9365279735917546e-05,
+      "loss": 1.1877,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9364426096344e-05,
+      "loss": 1.1614,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.936357189554771e-05,
+      "loss": 0.8272,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9362717133562038e-05,
+      "loss": 0.7149,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9361861810420386e-05,
+      "loss": 1.3427,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.936100592615617e-05,
+      "loss": 0.8209,
+      "step": 1482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9360149480802834e-05,
+      "loss": 0.8256,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9359292474393833e-05,
+      "loss": 0.3497,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9358434906962656e-05,
+      "loss": 0.5297,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9357576778542798e-05,
+      "loss": 1.1756,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.93567180891678e-05,
+      "loss": 1.1187,
+      "step": 1487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9355858838871205e-05,
+      "loss": 0.3029,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9354999027686586e-05,
+      "loss": 0.672,
+      "step": 1489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.935413865564753e-05,
+      "loss": 0.2909,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.935327772278766e-05,
+      "loss": 0.613,
+      "step": 1491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9352416229140614e-05,
+      "loss": 0.7843,
+      "step": 1492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9351554174740037e-05,
+      "loss": 0.63,
+      "step": 1493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9350691559619627e-05,
+      "loss": 0.5619,
+      "step": 1494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9349828383813073e-05,
+      "loss": 0.4128,
+      "step": 1495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.934896464735411e-05,
+      "loss": 0.444,
+      "step": 1496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9348100350276477e-05,
+      "loss": 0.5204,
+      "step": 1497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.934723549261395e-05,
+      "loss": 0.7344,
+      "step": 1498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9346370074400307e-05,
+      "loss": 0.3454,
+      "step": 1499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9345504095669373e-05,
+      "loss": 0.5521,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9344637556454976e-05,
+      "loss": 1.4547,
+      "step": 1501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9343770456790973e-05,
+      "loss": 1.1776,
+      "step": 1502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9342902796711237e-05,
+      "loss": 1.6714,
+      "step": 1503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9342034576249678e-05,
+      "loss": 1.2732,
+      "step": 1504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.934116579544021e-05,
+      "loss": 0.2487,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9340296454316777e-05,
+      "loss": 0.2313,
+      "step": 1506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9339426552913347e-05,
+      "loss": 1.3946,
+      "step": 1507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9338556091263907e-05,
+      "loss": 0.7265,
+      "step": 1508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9337685069402465e-05,
+      "loss": 0.4674,
+      "step": 1509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.933681348736305e-05,
+      "loss": 0.4042,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9335941345179726e-05,
+      "loss": 0.5377,
+      "step": 1511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.933506864288655e-05,
+      "loss": 0.3498,
+      "step": 1512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9334195380517636e-05,
+      "loss": 0.9842,
+      "step": 1513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.933332155810709e-05,
+      "loss": 0.4525,
+      "step": 1514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.933244717568906e-05,
+      "loss": 1.5202,
+      "step": 1515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.93315722332977e-05,
+      "loss": 0.5967,
+      "step": 1516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9330696730967207e-05,
+      "loss": 1.7009,
+      "step": 1517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932982066873178e-05,
+      "loss": 0.1711,
+      "step": 1518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932894404662564e-05,
+      "loss": 0.8955,
+      "step": 1519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9328066864683052e-05,
+      "loss": 0.6467,
+      "step": 1520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932718912293828e-05,
+      "loss": 0.6948,
+      "step": 1521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932631082142561e-05,
+      "loss": 0.3498,
+      "step": 1522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9325431960179375e-05,
+      "loss": 0.9443,
+      "step": 1523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9324552539233896e-05,
+      "loss": 1.1648,
+      "step": 1524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932367255862354e-05,
+      "loss": 1.8978,
+      "step": 1525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9322792018382686e-05,
+      "loss": 1.2057,
+      "step": 1526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932191091854574e-05,
+      "loss": 0.7518,
+      "step": 1527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.932102925914712e-05,
+      "loss": 2.1795,
+      "step": 1528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9320147040221284e-05,
+      "loss": 1.051,
+      "step": 1529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.931926426180269e-05,
+      "loss": 1.6328,
+      "step": 1530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9318380923925835e-05,
+      "loss": 0.7197,
+      "step": 1531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9317497026625227e-05,
+      "loss": 1.3052,
+      "step": 1532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9316612569935405e-05,
+      "loss": 1.039,
+      "step": 1533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9315727553890916e-05,
+      "loss": 0.7749,
+      "step": 1534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.931484197852635e-05,
+      "loss": 1.7071,
+      "step": 1535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.93139558438763e-05,
+      "loss": 0.739,
+      "step": 1536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9313069149975383e-05,
+      "loss": 0.5203,
+      "step": 1537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9312181896858253e-05,
+      "loss": 0.808,
+      "step": 1538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.931129408455957e-05,
+      "loss": 0.4546,
+      "step": 1539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9310405713114016e-05,
+      "loss": 0.5263,
+      "step": 1540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9309516782556306e-05,
+      "loss": 1.2369,
+      "step": 1541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930862729292117e-05,
+      "loss": 0.9875,
+      "step": 1542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9307737244243365e-05,
+      "loss": 0.9596,
+      "step": 1543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930684663655765e-05,
+      "loss": 1.4418,
+      "step": 1544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9305955469898843e-05,
+      "loss": 0.5455,
+      "step": 1545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930506374430175e-05,
+      "loss": 1.2233,
+      "step": 1546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9304171459801207e-05,
+      "loss": 0.9447,
+      "step": 1547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9303278616432084e-05,
+      "loss": 0.9189,
+      "step": 1548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930238521422926e-05,
+      "loss": 1.2325,
+      "step": 1549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9301491253227644e-05,
+      "loss": 0.3373,
+      "step": 1550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.930059673346216e-05,
+      "loss": 0.4399,
+      "step": 1551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9299701654967762e-05,
+      "loss": 0.4775,
+      "step": 1552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.929880601777942e-05,
+      "loss": 0.4315,
+      "step": 1553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.929790982193212e-05,
+      "loss": 2.1431,
+      "step": 1554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.929701306746088e-05,
+      "loss": 0.6981,
+      "step": 1555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.929611575440074e-05,
+      "loss": 0.4426,
+      "step": 1556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9295217882786757e-05,
+      "loss": 1.4492,
+      "step": 1557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9294319452654008e-05,
+      "loss": 1.3103,
+      "step": 1558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.92934204640376e-05,
+      "loss": 0.3053,
+      "step": 1559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9292520916972653e-05,
+      "loss": 0.38,
+      "step": 1560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9291620811494312e-05,
+      "loss": 1.6256,
+      "step": 1561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.929072014763775e-05,
+      "loss": 0.7241,
+      "step": 1562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9289818925438147e-05,
+      "loss": 0.3682,
+      "step": 1563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9288917144930722e-05,
+      "loss": 0.6419,
+      "step": 1564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9288014806150705e-05,
+      "loss": 0.643,
+      "step": 1565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9287111909133354e-05,
+      "loss": 0.8515,
+      "step": 1566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9286208453913938e-05,
+      "loss": 2.0354,
+      "step": 1567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.928530444052776e-05,
+      "loss": 0.5237,
+      "step": 1568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9284399869010136e-05,
+      "loss": 0.5543,
+      "step": 1569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9283494739396417e-05,
+      "loss": 0.917,
+      "step": 1570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9282589051721963e-05,
+      "loss": 0.877,
+      "step": 1571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9281682806022155e-05,
+      "loss": 0.6741,
+      "step": 1572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9280776002332404e-05,
+      "loss": 0.7703,
+      "step": 1573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9279868640688135e-05,
+      "loss": 0.5131,
+      "step": 1574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9278960721124806e-05,
+      "loss": 1.1659,
+      "step": 1575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9278052243677883e-05,
+      "loss": 0.9072,
+      "step": 1576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9277143208382864e-05,
+      "loss": 1.2189,
+      "step": 1577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9276233615275266e-05,
+      "loss": 1.3536,
+      "step": 1578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9275323464390625e-05,
+      "loss": 0.6727,
+      "step": 1579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.92744127557645e-05,
+      "loss": 1.0744,
+      "step": 1580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9273501489432476e-05,
+      "loss": 0.1515,
+      "step": 1581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9272589665430152e-05,
+      "loss": 1.2543,
+      "step": 1582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9271677283793158e-05,
+      "loss": 1.08,
+      "step": 1583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9270764344557142e-05,
+      "loss": 0.9306,
+      "step": 1584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9269850847757766e-05,
+      "loss": 1.3884,
+      "step": 1585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9268936793430722e-05,
+      "loss": 1.0161,
+      "step": 1586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9268022181611733e-05,
+      "loss": 0.3634,
+      "step": 1587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.926710701233652e-05,
+      "loss": 0.9774,
+      "step": 1588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.926619128564084e-05,
+      "loss": 1.4043,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9265275001560486e-05,
+      "loss": 1.3792,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.926435816013124e-05,
+      "loss": 0.8191,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.926344076138893e-05,
+      "loss": 0.8703,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9262522805369396e-05,
+      "loss": 1.0296,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.926160429210851e-05,
+      "loss": 0.606,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.926068522164215e-05,
+      "loss": 0.356,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9259765594006227e-05,
+      "loss": 0.989,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9258845409236675e-05,
+      "loss": 0.5326,
+      "step": 1597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9257924667369445e-05,
+      "loss": 0.4963,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.925700336844051e-05,
+      "loss": 0.5321,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.925608151248586e-05,
+      "loss": 0.7608,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.925515909954152e-05,
+      "loss": 1.5622,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.925423612964353e-05,
+      "loss": 1.5541,
+      "step": 1602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.925331260282794e-05,
+      "loss": 1.2718,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9252388519130842e-05,
+      "loss": 1.1547,
+      "step": 1604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9251463878588335e-05,
+      "loss": 0.4023,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9250538681236546e-05,
+      "loss": 2.0138,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.924961292711163e-05,
+      "loss": 1.5842,
+      "step": 1607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9248686616249746e-05,
+      "loss": 1.6685,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9247759748687092e-05,
+      "loss": 0.2593,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9246832324459877e-05,
+      "loss": 0.455,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.924590434360434e-05,
+      "loss": 1.3365,
+      "step": 1611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9244975806156737e-05,
+      "loss": 0.4408,
+      "step": 1612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9244046712153337e-05,
+      "loss": 1.8445,
+      "step": 1613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9243117061630454e-05,
+      "loss": 0.5833,
+      "step": 1614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9242186854624396e-05,
+      "loss": 1.3325,
+      "step": 1615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.924125609117152e-05,
+      "loss": 0.5808,
+      "step": 1616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9240324771308186e-05,
+      "loss": 1.8232,
+      "step": 1617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9239392895070776e-05,
+      "loss": 0.4675,
+      "step": 1618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9238460462495705e-05,
+      "loss": 0.4035,
+      "step": 1619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9237527473619394e-05,
+      "loss": 0.4148,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.923659392847831e-05,
+      "loss": 0.532,
+      "step": 1621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9235659827108916e-05,
+      "loss": 1.4805,
+      "step": 1622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9234725169547713e-05,
+      "loss": 1.064,
+      "step": 1623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9233789955831213e-05,
+      "loss": 1.4584,
+      "step": 1624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.923285418599596e-05,
+      "loss": 0.7205,
+      "step": 1625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9231917860078515e-05,
+      "loss": 0.8693,
+      "step": 1626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.923098097811545e-05,
+      "loss": 0.784,
+      "step": 1627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9230043540143387e-05,
+      "loss": 0.3651,
+      "step": 1628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9229105546198938e-05,
+      "loss": 1.1105,
+      "step": 1629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.922816699631876e-05,
+      "loss": 0.349,
+      "step": 1630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.922722789053951e-05,
+      "loss": 0.3715,
+      "step": 1631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9226288228897897e-05,
+      "loss": 0.5657,
+      "step": 1632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.922534801143062e-05,
+      "loss": 0.7245,
+      "step": 1633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9224407238174416e-05,
+      "loss": 0.1955,
+      "step": 1634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.922346590916604e-05,
+      "loss": 0.6607,
+      "step": 1635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.922252402444228e-05,
+      "loss": 0.5452,
+      "step": 1636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9221581584039927e-05,
+      "loss": 1.2556,
+      "step": 1637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.92206385879958e-05,
+      "loss": 0.8704,
+      "step": 1638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.921969503634675e-05,
+      "loss": 0.7247,
+      "step": 1639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.921875092912964e-05,
+      "loss": 0.8505,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9217806266381348e-05,
+      "loss": 0.5459,
+      "step": 1641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9216861048138795e-05,
+      "loss": 0.4671,
+      "step": 1642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.92159152744389e-05,
+      "loss": 0.5029,
+      "step": 1643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9214968945318625e-05,
+      "loss": 0.3789,
+      "step": 1644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9214022060814935e-05,
+      "loss": 0.6308,
+      "step": 1645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.921307462096483e-05,
+      "loss": 0.5479,
+      "step": 1646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.921212662580532e-05,
+      "loss": 0.4842,
+      "step": 1647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9211178075373456e-05,
+      "loss": 0.6394,
+      "step": 1648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9210228969706286e-05,
+      "loss": 0.3513,
+      "step": 1649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.92092793088409e-05,
+      "loss": 0.6064,
+      "step": 1650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9208329092814393e-05,
+      "loss": 1.5495,
+      "step": 1651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9207378321663896e-05,
+      "loss": 1.6659,
+      "step": 1652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9206426995426557e-05,
+      "loss": 0.422,
+      "step": 1653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9205475114139545e-05,
+      "loss": 1.5845,
+      "step": 1654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9204522677840045e-05,
+      "loss": 0.6746,
+      "step": 1655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9203569686565274e-05,
+      "loss": 0.6387,
+      "step": 1656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.920261614035246e-05,
+      "loss": 0.5407,
+      "step": 1657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9201662039238867e-05,
+      "loss": 0.9373,
+      "step": 1658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9200707383261767e-05,
+      "loss": 1.9767,
+      "step": 1659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9199752172458452e-05,
+      "loss": 0.4067,
+      "step": 1660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9198796406866257e-05,
+      "loss": 1.1071,
+      "step": 1661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9197840086522514e-05,
+      "loss": 0.994,
+      "step": 1662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919688321146459e-05,
+      "loss": 0.6324,
+      "step": 1663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9195925781729866e-05,
+      "loss": 0.5433,
+      "step": 1664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919496779735576e-05,
+      "loss": 0.632,
+      "step": 1665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9194009258379686e-05,
+      "loss": 0.4476,
+      "step": 1666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9193050164839103e-05,
+      "loss": 0.7816,
+      "step": 1667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919209051677148e-05,
+      "loss": 0.7496,
+      "step": 1668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.919113031421432e-05,
+      "loss": 0.5448,
+      "step": 1669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9190169557205126e-05,
+      "loss": 0.3415,
+      "step": 1670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9189208245781443e-05,
+      "loss": 0.5013,
+      "step": 1671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9188246379980822e-05,
+      "loss": 0.5231,
+      "step": 1672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.918728395984085e-05,
+      "loss": 0.7253,
+      "step": 1673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.918632098539913e-05,
+      "loss": 0.8313,
+      "step": 1674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9185357456693283e-05,
+      "loss": 0.8532,
+      "step": 1675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.918439337376095e-05,
+      "loss": 0.7608,
+      "step": 1676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9183428736639806e-05,
+      "loss": 1.7501,
+      "step": 1677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9182463545367536e-05,
+      "loss": 1.5643,
+      "step": 1678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9181497799981853e-05,
+      "loss": 0.6144,
+      "step": 1679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9180531500520484e-05,
+      "loss": 0.8509,
+      "step": 1680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9179564647021183e-05,
+      "loss": 0.3924,
+      "step": 1681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9178597239521728e-05,
+      "loss": 1.6538,
+      "step": 1682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.917762927805992e-05,
+      "loss": 0.9123,
+      "step": 1683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9176660762673567e-05,
+      "loss": 0.8493,
+      "step": 1684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9175691693400518e-05,
+      "loss": 0.4034,
+      "step": 1685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9174722070278636e-05,
+      "loss": 0.3223,
+      "step": 1686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.917375189334579e-05,
+      "loss": 1.1236,
+      "step": 1687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9172781162639905e-05,
+      "loss": 0.9368,
+      "step": 1688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9171809878198896e-05,
+      "loss": 0.5476,
+      "step": 1689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9170838040060714e-05,
+      "loss": 0.7464,
+      "step": 1690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9169865648263325e-05,
+      "loss": 0.7965,
+      "step": 1691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9168892702844723e-05,
+      "loss": 1.3268,
+      "step": 1692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9167919203842932e-05,
+      "loss": 0.8198,
+      "step": 1693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.916694515129597e-05,
+      "loss": 1.0909,
+      "step": 1694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9165970545241904e-05,
+      "loss": 0.7133,
+      "step": 1695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.916499538571881e-05,
+      "loss": 0.4522,
+      "step": 1696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9164019672764784e-05,
+      "loss": 0.7695,
+      "step": 1697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9163043406417954e-05,
+      "loss": 0.5431,
+      "step": 1698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9162066586716456e-05,
+      "loss": 0.5902,
+      "step": 1699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.916108921369846e-05,
+      "loss": 0.6785,
+      "step": 1700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9160111287402154e-05,
+      "loss": 1.4647,
+      "step": 1701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9159132807865738e-05,
+      "loss": 0.8728,
+      "step": 1702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9158153775127446e-05,
+      "loss": 0.419,
+      "step": 1703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9157174189225532e-05,
+      "loss": 1.8242,
+      "step": 1704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9156194050198263e-05,
+      "loss": 1.9443,
+      "step": 1705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.915521335808394e-05,
+      "loss": 0.7198,
+      "step": 1706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.915423211292087e-05,
+      "loss": 0.9009,
+      "step": 1707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9153250314747398e-05,
+      "loss": 1.6686,
+      "step": 1708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.915226796360188e-05,
+      "loss": 0.5212,
+      "step": 1709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.91512850595227e-05,
+      "loss": 0.4748,
+      "step": 1710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9150301602548258e-05,
+      "loss": 1.0451,
+      "step": 1711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9149317592716975e-05,
+      "loss": 0.6763,
+      "step": 1712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9148333030067305e-05,
+      "loss": 1.4774,
+      "step": 1713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.91473479146377e-05,
+      "loss": 0.8971,
+      "step": 1714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9146362246466668e-05,
+      "loss": 0.657,
+      "step": 1715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9145376025592706e-05,
+      "loss": 0.9955,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.914438925205435e-05,
+      "loss": 0.1983,
+      "step": 1717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.914340192589015e-05,
+      "loss": 0.3439,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.914241404713869e-05,
+      "loss": 1.0338,
+      "step": 1719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.914142561583856e-05,
+      "loss": 0.4461,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9140436632028384e-05,
+      "loss": 1.0849,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9139447095746788e-05,
+      "loss": 0.7484,
+      "step": 1722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.913845700703245e-05,
+      "loss": 0.6772,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.913746636592404e-05,
+      "loss": 0.4587,
+      "step": 1724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9136475172460275e-05,
+      "loss": 1.9503,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9135483426679873e-05,
+      "loss": 0.3415,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9134491128621583e-05,
+      "loss": 1.4709,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9133498278324177e-05,
+      "loss": 0.875,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9132504875826443e-05,
+      "loss": 1.2029,
+      "step": 1729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9131510921167196e-05,
+      "loss": 0.8119,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9130516414385267e-05,
+      "loss": 0.7752,
+      "step": 1731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.912952135551951e-05,
+      "loss": 0.9993,
+      "step": 1732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9128525744608818e-05,
+      "loss": 0.7127,
+      "step": 1733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9127529581692068e-05,
+      "loss": 1.0197,
+      "step": 1734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9126532866808192e-05,
+      "loss": 0.6469,
+      "step": 1735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.912553559999613e-05,
+      "loss": 0.3304,
+      "step": 1736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9124537781294846e-05,
+      "loss": 0.8467,
+      "step": 1737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9123539410743325e-05,
+      "loss": 0.7398,
+      "step": 1738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9122540488380577e-05,
+      "loss": 1.9291,
+      "step": 1739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.912154101424562e-05,
+      "loss": 0.7697,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9120540988377514e-05,
+      "loss": 0.9228,
+      "step": 1741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9119540410815327e-05,
+      "loss": 0.6617,
+      "step": 1742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.911853928159815e-05,
+      "loss": 0.9941,
+      "step": 1743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9117537600765102e-05,
+      "loss": 0.351,
+      "step": 1744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9116535368355314e-05,
+      "loss": 0.7982,
+      "step": 1745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9115532584407946e-05,
+      "loss": 0.4805,
+      "step": 1746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9114529248962173e-05,
+      "loss": 0.4331,
+      "step": 1747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9113525362057203e-05,
+      "loss": 0.5848,
+      "step": 1748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9112520923732256e-05,
+      "loss": 1.072,
+      "step": 1749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9111515934026573e-05,
+      "loss": 0.6487,
+      "step": 1750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.911051039297942e-05,
+      "loss": 0.2748,
+      "step": 1751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9109504300630083e-05,
+      "loss": 0.548,
+      "step": 1752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9108497657017873e-05,
+      "loss": 0.3011,
+      "step": 1753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.910749046218212e-05,
+      "loss": 1.7103,
+      "step": 1754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9106482716162173e-05,
+      "loss": 0.211,
+      "step": 1755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9105474418997407e-05,
+      "loss": 0.6974,
+      "step": 1756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.910446557072721e-05,
+      "loss": 0.6403,
+      "step": 1757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.910345617139101e-05,
+      "loss": 1.6336,
+      "step": 1758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9102446221028237e-05,
+      "loss": 0.2733,
+      "step": 1759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.910143571967835e-05,
+      "loss": 0.5237,
+      "step": 1760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.910042466738083e-05,
+      "loss": 0.2364,
+      "step": 1761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.909941306417518e-05,
+      "loss": 0.6382,
+      "step": 1762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9098400910100926e-05,
+      "loss": 0.4913,
+      "step": 1763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9097388205197608e-05,
+      "loss": 1.1959,
+      "step": 1764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9096374949504798e-05,
+      "loss": 0.493,
+      "step": 1765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.909536114306208e-05,
+      "loss": 0.5351,
+      "step": 1766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9094346785909064e-05,
+      "loss": 0.5497,
+      "step": 1767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9093331878085385e-05,
+      "loss": 1.2362,
+      "step": 1768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9092316419630693e-05,
+      "loss": 0.5754,
+      "step": 1769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9091300410584663e-05,
+      "loss": 1.6947,
+      "step": 1770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.909028385098699e-05,
+      "loss": 1.5239,
+      "step": 1771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.908926674087739e-05,
+      "loss": 0.854,
+      "step": 1772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9088249080295604e-05,
+      "loss": 0.3208,
+      "step": 1773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9087230869281393e-05,
+      "loss": 0.4226,
+      "step": 1774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.908621210787454e-05,
+      "loss": 1.1057,
+      "step": 1775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9085192796114844e-05,
+      "loss": 0.4815,
+      "step": 1776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9084172934042135e-05,
+      "loss": 1.3165,
+      "step": 1777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.908315252169625e-05,
+      "loss": 0.8157,
+      "step": 1778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.908213155911707e-05,
+      "loss": 0.6755,
+      "step": 1779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9081110046344472e-05,
+      "loss": 0.8274,
+      "step": 1780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9080087983418375e-05,
+      "loss": 0.4968,
+      "step": 1781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.907906537037871e-05,
+      "loss": 1.0035,
+      "step": 1782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9078042207265425e-05,
+      "loss": 1.1818,
+      "step": 1783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9077018494118505e-05,
+      "loss": 0.9961,
+      "step": 1784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.907599423097794e-05,
+      "loss": 0.6884,
+      "step": 1785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.907496941788375e-05,
+      "loss": 0.6608,
+      "step": 1786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.907394405487597e-05,
+      "loss": 0.9696,
+      "step": 1787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9072918141994672e-05,
+      "loss": 0.8251,
+      "step": 1788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.907189167927993e-05,
+      "loss": 1.2558,
+      "step": 1789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9070864666771852e-05,
+      "loss": 0.7029,
+      "step": 1790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9069837104510562e-05,
+      "loss": 0.502,
+      "step": 1791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.906880899253621e-05,
+      "loss": 0.6217,
+      "step": 1792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.906778033088896e-05,
+      "loss": 0.9502,
+      "step": 1793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9066751119609004e-05,
+      "loss": 0.1776,
+      "step": 1794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9065721358736555e-05,
+      "loss": 0.6228,
+      "step": 1795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9064691048311844e-05,
+      "loss": 0.5603,
+      "step": 1796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.906366018837513e-05,
+      "loss": 0.7158,
+      "step": 1797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9062628778966686e-05,
+      "loss": 0.6839,
+      "step": 1798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.906159682012681e-05,
+      "loss": 0.2964,
+      "step": 1799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9060564311895816e-05,
+      "loss": 0.3226,
+      "step": 1800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.905953125431405e-05,
+      "loss": 1.8769,
+      "step": 1801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9058497647421876e-05,
+      "loss": 0.237,
+      "step": 1802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.905746349125967e-05,
+      "loss": 1.2477,
+      "step": 1803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9056428785867843e-05,
+      "loss": 0.5204,
+      "step": 1804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.905539353128682e-05,
+      "loss": 0.5648,
+      "step": 1805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.905435772755705e-05,
+      "loss": 0.4298,
+      "step": 1806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9053321374718996e-05,
+      "loss": 0.7793,
+      "step": 1807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.905228447281315e-05,
+      "loss": 1.0253,
+      "step": 1808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.905124702188003e-05,
+      "loss": 0.7611,
+      "step": 1809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9050209021960166e-05,
+      "loss": 1.5764,
+      "step": 1810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9049170473094113e-05,
+      "loss": 0.4015,
+      "step": 1811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9048131375322452e-05,
+      "loss": 0.6462,
+      "step": 1812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9047091728685772e-05,
+      "loss": 0.5089,
+      "step": 1813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.90460515332247e-05,
+      "loss": 1.7983,
+      "step": 1814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9045010788979872e-05,
+      "loss": 0.4113,
+      "step": 1815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9043969495991954e-05,
+      "loss": 0.4108,
+      "step": 1816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.904292765430162e-05,
+      "loss": 1.6178,
+      "step": 1817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.904188526394959e-05,
+      "loss": 0.826,
+      "step": 1818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.904084232497658e-05,
+      "loss": 0.4328,
+      "step": 1819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9039798837423342e-05,
+      "loss": 1.072,
+      "step": 1820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9038754801330652e-05,
+      "loss": 1.1236,
+      "step": 1821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9037710216739287e-05,
+      "loss": 1.0735,
+      "step": 1822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.903666508369006e-05,
+      "loss": 1.5684,
+      "step": 1823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.903561940222382e-05,
+      "loss": 1.8356,
+      "step": 1824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9034573172381404e-05,
+      "loss": 1.6643,
+      "step": 1825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9033526394203703e-05,
+      "loss": 1.4871,
+      "step": 1826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9032479067731607e-05,
+      "loss": 1.2321,
+      "step": 1827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9031431193006038e-05,
+      "loss": 2.1949,
+      "step": 1828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9030382770067935e-05,
+      "loss": 0.264,
+      "step": 1829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.902933379895826e-05,
+      "loss": 0.7691,
+      "step": 1830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9028284279718005e-05,
+      "loss": 0.8888,
+      "step": 1831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.902723421238816e-05,
+      "loss": 1.0137,
+      "step": 1832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9026183597009763e-05,
+      "loss": 0.9497,
+      "step": 1833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.902513243362386e-05,
+      "loss": 1.4285,
+      "step": 1834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9024080722271517e-05,
+      "loss": 0.8187,
+      "step": 1835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9023028462993823e-05,
+      "loss": 0.925,
+      "step": 1836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.90219756558319e-05,
+      "loss": 0.8166,
+      "step": 1837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.902092230082687e-05,
+      "loss": 0.878,
+      "step": 1838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9019868398019896e-05,
+      "loss": 0.7399,
+      "step": 1839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.901881394745215e-05,
+      "loss": 1.3231,
+      "step": 1840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.901775894916483e-05,
+      "loss": 0.9537,
+      "step": 1841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9016703403199156e-05,
+      "loss": 0.9292,
+      "step": 1842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.901564730959637e-05,
+      "loss": 0.7606,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.901459066839773e-05,
+      "loss": 0.2931,
+      "step": 1844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9013533479644525e-05,
+      "loss": 0.7264,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9012475743378055e-05,
+      "loss": 1.3348,
+      "step": 1846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.901141745963965e-05,
+      "loss": 0.4043,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9010358628470655e-05,
+      "loss": 0.4467,
+      "step": 1848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9009299249912437e-05,
+      "loss": 0.533,
+      "step": 1849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.900823932400639e-05,
+      "loss": 0.9161,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9007178850793918e-05,
+      "loss": 0.6485,
+      "step": 1851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9006117830316467e-05,
+      "loss": 1.0911,
+      "step": 1852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9005056262615485e-05,
+      "loss": 0.396,
+      "step": 1853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.900399414773244e-05,
+      "loss": 0.88,
+      "step": 1854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9002931485708842e-05,
+      "loss": 1.5883,
+      "step": 1855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.90018682765862e-05,
+      "loss": 0.7329,
+      "step": 1856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.900080452040606e-05,
+      "loss": 0.5883,
+      "step": 1857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8999740217209984e-05,
+      "loss": 1.9372,
+      "step": 1858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8998675367039548e-05,
+      "loss": 0.6258,
+      "step": 1859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.899760996993636e-05,
+      "loss": 1.1773,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8996544025942047e-05,
+      "loss": 1.3506,
+      "step": 1861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8995477535098252e-05,
+      "loss": 0.5758,
+      "step": 1862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8994410497446646e-05,
+      "loss": 0.4269,
+      "step": 1863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8993342913028917e-05,
+      "loss": 0.4451,
+      "step": 1864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.899227478188678e-05,
+      "loss": 1.7032,
+      "step": 1865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8991206104061963e-05,
+      "loss": 0.456,
+      "step": 1866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8990136879596215e-05,
+      "loss": 0.838,
+      "step": 1867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.898906710853132e-05,
+      "loss": 0.9097,
+      "step": 1868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.898799679090907e-05,
+      "loss": 0.5769,
+      "step": 1869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.898692592677128e-05,
+      "loss": 0.6603,
+      "step": 1870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8985854516159798e-05,
+      "loss": 0.9988,
+      "step": 1871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.898478255911648e-05,
+      "loss": 0.2847,
+      "step": 1872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.89837100556832e-05,
+      "loss": 0.1424,
+      "step": 1873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.898263700590187e-05,
+      "loss": 1.8344,
+      "step": 1874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.898156340981441e-05,
+      "loss": 0.8153,
+      "step": 1875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.898048926746277e-05,
+      "loss": 0.7937,
+      "step": 1876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.897941457888891e-05,
+      "loss": 0.5095,
+      "step": 1877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8978339344134833e-05,
+      "loss": 0.371,
+      "step": 1878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.897726356324253e-05,
+      "loss": 1.0404,
+      "step": 1879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.897618723625404e-05,
+      "loss": 0.7796,
+      "step": 1880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8975110363211425e-05,
+      "loss": 0.9569,
+      "step": 1881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8974032944156742e-05,
+      "loss": 0.713,
+      "step": 1882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8972954979132095e-05,
+      "loss": 0.5272,
+      "step": 1883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.89718764681796e-05,
+      "loss": 1.2854,
+      "step": 1884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8970797411341397e-05,
+      "loss": 0.4211,
+      "step": 1885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8969717808659642e-05,
+      "loss": 0.8649,
+      "step": 1886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8968637660176517e-05,
+      "loss": 0.5504,
+      "step": 1887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.896755696593422e-05,
+      "loss": 1.2041,
+      "step": 1888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8966475725974978e-05,
+      "loss": 1.9616,
+      "step": 1889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8965393940341036e-05,
+      "loss": 0.7416,
+      "step": 1890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8964311609074653e-05,
+      "loss": 0.739,
+      "step": 1891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8963228732218123e-05,
+      "loss": 0.4276,
+      "step": 1892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8962145309813754e-05,
+      "loss": 0.2939,
+      "step": 1893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8961061341903873e-05,
+      "loss": 0.3961,
+      "step": 1894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8959976828530832e-05,
+      "loss": 0.9151,
+      "step": 1895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8958891769737003e-05,
+      "loss": 0.5056,
+      "step": 1896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.895780616556478e-05,
+      "loss": 0.4054,
+      "step": 1897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8956720016056575e-05,
+      "loss": 0.7573,
+      "step": 1898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8955633321254828e-05,
+      "loss": 0.2609,
+      "step": 1899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8954546081202e-05,
+      "loss": 0.7856,
+      "step": 1900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.895345829594056e-05,
+      "loss": 0.723,
+      "step": 1901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8952369965513015e-05,
+      "loss": 1.1284,
+      "step": 1902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8951281089961885e-05,
+      "loss": 1.4966,
+      "step": 1903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8950191669329712e-05,
+      "loss": 0.3516,
+      "step": 1904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.894910170365906e-05,
+      "loss": 0.2772,
+      "step": 1905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8948011192992515e-05,
+      "loss": 1.2739,
+      "step": 1906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8946920137372685e-05,
+      "loss": 0.1903,
+      "step": 1907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8945828536842194e-05,
+      "loss": 0.3169,
+      "step": 1908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8944736391443696e-05,
+      "loss": 0.8829,
+      "step": 1909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8943643701219857e-05,
+      "loss": 1.1648,
+      "step": 1910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.894255046621337e-05,
+      "loss": 0.9978,
+      "step": 1911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8941456686466953e-05,
+      "loss": 1.0756,
+      "step": 1912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8940362362023333e-05,
+      "loss": 1.2036,
+      "step": 1913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8939267492925266e-05,
+      "loss": 1.7906,
+      "step": 1914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8938172079215535e-05,
+      "loss": 1.8343,
+      "step": 1915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8937076120936937e-05,
+      "loss": 1.7136,
+      "step": 1916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8935979618132283e-05,
+      "loss": 0.8952,
+      "step": 1917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8934882570844425e-05,
+      "loss": 0.7422,
+      "step": 1918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.893378497911622e-05,
+      "loss": 0.7132,
+      "step": 1919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8932686842990547e-05,
+      "loss": 1.2606,
+      "step": 1920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8931588162510318e-05,
+      "loss": 1.1943,
+      "step": 1921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8930488937718454e-05,
+      "loss": 1.0039,
+      "step": 1922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8929389168657905e-05,
+      "loss": 0.4559,
+      "step": 1923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8928288855371634e-05,
+      "loss": 0.8463,
+      "step": 1924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8927187997902638e-05,
+      "loss": 1.1739,
+      "step": 1925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8926086596293925e-05,
+      "loss": 1.8525,
+      "step": 1926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8924984650588525e-05,
+      "loss": 2.0235,
+      "step": 1927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.892388216082949e-05,
+      "loss": 0.7087,
+      "step": 1928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8922779127059902e-05,
+      "loss": 0.78,
+      "step": 1929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.892167554932285e-05,
+      "loss": 0.3748,
+      "step": 1930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8920571427661456e-05,
+      "loss": 0.8628,
+      "step": 1931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8919466762118857e-05,
+      "loss": 0.6322,
+      "step": 1932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8918361552738205e-05,
+      "loss": 0.754,
+      "step": 1933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8917255799562692e-05,
+      "loss": 0.5369,
+      "step": 1934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.891614950263551e-05,
+      "loss": 0.5324,
+      "step": 1935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8915042661999893e-05,
+      "loss": 0.6586,
+      "step": 1936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.891393527769908e-05,
+      "loss": 1.5023,
+      "step": 1937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.891282734977634e-05,
+      "loss": 1.4697,
+      "step": 1938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8911718878274957e-05,
+      "loss": 0.434,
+      "step": 1939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8910609863238234e-05,
+      "loss": 0.2524,
+      "step": 1940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.890950030470951e-05,
+      "loss": 0.7323,
+      "step": 1941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8908390202732134e-05,
+      "loss": 0.3673,
+      "step": 1942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.890727955734947e-05,
+      "loss": 1.0349,
+      "step": 1943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8906168368604927e-05,
+      "loss": 0.53,
+      "step": 1944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.89050566365419e-05,
+      "loss": 0.899,
+      "step": 1945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.890394436120384e-05,
+      "loss": 0.7952,
+      "step": 1946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8902831542634197e-05,
+      "loss": 0.6327,
+      "step": 1947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8901718180876453e-05,
+      "loss": 0.595,
+      "step": 1948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.890060427597411e-05,
+      "loss": 0.6942,
+      "step": 1949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8899489827970674e-05,
+      "loss": 0.4676,
+      "step": 1950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.88983748369097e-05,
+      "loss": 1.337,
+      "step": 1951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.889725930283475e-05,
+      "loss": 1.5635,
+      "step": 1952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8896143225789406e-05,
+      "loss": 1.3814,
+      "step": 1953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.889502660581727e-05,
+      "loss": 0.4564,
+      "step": 1954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8893909442961974e-05,
+      "loss": 0.513,
+      "step": 1955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.889279173726717e-05,
+      "loss": 1.0071,
+      "step": 1956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8891673488776513e-05,
+      "loss": 0.5545,
+      "step": 1957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.88905546975337e-05,
+      "loss": 0.5763,
+      "step": 1958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8889435363582456e-05,
+      "loss": 0.465,
+      "step": 1959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8888315486966496e-05,
+      "loss": 2.0168,
+      "step": 1960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8887195067729574e-05,
+      "loss": 0.2383,
+      "step": 1961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8886074105915473e-05,
+      "loss": 0.6511,
+      "step": 1962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.888495260156799e-05,
+      "loss": 0.3741,
+      "step": 1963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8883830554730942e-05,
+      "loss": 0.7767,
+      "step": 1964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8882707965448163e-05,
+      "loss": 1.7411,
+      "step": 1965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8881584833763516e-05,
+      "loss": 1.7649,
+      "step": 1966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8880461159720882e-05,
+      "loss": 0.1597,
+      "step": 1967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8879336943364163e-05,
+      "loss": 2.2086,
+      "step": 1968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8878212184737286e-05,
+      "loss": 1.6481,
+      "step": 1969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8877086883884186e-05,
+      "loss": 1.0095,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8875961040848837e-05,
+      "loss": 1.0808,
+      "step": 1971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8874834655675224e-05,
+      "loss": 0.6613,
+      "step": 1972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8873707728407352e-05,
+      "loss": 0.5443,
+      "step": 1973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.887258025908926e-05,
+      "loss": 0.6199,
+      "step": 1974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.887145224776499e-05,
+      "loss": 0.8581,
+      "step": 1975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8870323694478614e-05,
+      "loss": 0.5937,
+      "step": 1976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8869194599274226e-05,
+      "loss": 0.6399,
+      "step": 1977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8868064962195944e-05,
+      "loss": 0.4268,
+      "step": 1978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.88669347832879e-05,
+      "loss": 1.6349,
+      "step": 1979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8865804062594252e-05,
+      "loss": 0.7763,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8864672800159176e-05,
+      "loss": 0.8562,
+      "step": 1981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8863540996026868e-05,
+      "loss": 0.2611,
+      "step": 1982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.886240865024155e-05,
+      "loss": 1.4132,
+      "step": 1983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.886127576284747e-05,
+      "loss": 1.1584,
+      "step": 1984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.886014233388888e-05,
+      "loss": 0.3247,
+      "step": 1985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.885900836341007e-05,
+      "loss": 0.9541,
+      "step": 1986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8857873851455347e-05,
+      "loss": 1.3016,
+      "step": 1987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.885673879806903e-05,
+      "loss": 0.6479,
+      "step": 1988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8855603203295463e-05,
+      "loss": 0.2565,
+      "step": 1989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8854467067179023e-05,
+      "loss": 0.6173,
+      "step": 1990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8853330389764094e-05,
+      "loss": 0.7667,
+      "step": 1991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.885219317109509e-05,
+      "loss": 1.1582,
+      "step": 1992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.885105541121644e-05,
+      "loss": 1.0107,
+      "step": 1993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8849917110172596e-05,
+      "loss": 1.2036,
+      "step": 1994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8848778268008035e-05,
+      "loss": 0.5162,
+      "step": 1995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8847638884767244e-05,
+      "loss": 0.4891,
+      "step": 1996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.884649896049475e-05,
+      "loss": 0.5056,
+      "step": 1997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8845358495235087e-05,
+      "loss": 0.2164,
+      "step": 1998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8844217489032806e-05,
+      "loss": 0.4989,
+      "step": 1999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.884307594193249e-05,
+      "loss": 0.3501,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.884193385397875e-05,
+      "loss": 0.434,
+      "step": 2001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8840791225216193e-05,
+      "loss": 0.3611,
+      "step": 2002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8839648055689467e-05,
+      "loss": 1.3938,
+      "step": 2003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.883850434544324e-05,
+      "loss": 2.0429,
+      "step": 2004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8837360094522195e-05,
+      "loss": 0.2999,
+      "step": 2005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8836215302971035e-05,
+      "loss": 1.4805,
+      "step": 2006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8835069970834493e-05,
+      "loss": 0.2836,
+      "step": 2007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8833924098157313e-05,
+      "loss": 0.5886,
+      "step": 2008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8832777684984262e-05,
+      "loss": 0.8321,
+      "step": 2009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8831630731360138e-05,
+      "loss": 0.3867,
+      "step": 2010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8830483237329746e-05,
+      "loss": 0.7058,
+      "step": 2011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.882933520293793e-05,
+      "loss": 0.3836,
+      "step": 2012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8828186628229532e-05,
+      "loss": 0.749,
+      "step": 2013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.882703751324943e-05,
+      "loss": 2.2593,
+      "step": 2014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8825887858042525e-05,
+      "loss": 0.6216,
+      "step": 2015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8824737662653732e-05,
+      "loss": 0.7718,
+      "step": 2016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8823586927127984e-05,
+      "loss": 0.4724,
+      "step": 2017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.882243565151025e-05,
+      "loss": 0.8759,
+      "step": 2018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8821283835845504e-05,
+      "loss": 1.3733,
+      "step": 2019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8820131480178753e-05,
+      "loss": 0.6032,
+      "step": 2020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8818978584555014e-05,
+      "loss": 0.5135,
+      "step": 2021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8817825149019332e-05,
+      "loss": 0.4881,
+      "step": 2022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.881667117361678e-05,
+      "loss": 1.1485,
+      "step": 2023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8815516658392434e-05,
+      "loss": 1.183,
+      "step": 2024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8814361603391408e-05,
+      "loss": 0.6182,
+      "step": 2025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.881320600865882e-05,
+      "loss": 1.951,
+      "step": 2026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8812049874239837e-05,
+      "loss": 1.1785,
+      "step": 2027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8810893200179618e-05,
+      "loss": 0.8258,
+      "step": 2028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8809735986523355e-05,
+      "loss": 0.2129,
+      "step": 2029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.880857823331626e-05,
+      "loss": 0.5941,
+      "step": 2030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8807419940603572e-05,
+      "loss": 0.3346,
+      "step": 2031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8806261108430543e-05,
+      "loss": 0.6603,
+      "step": 2032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.880510173684245e-05,
+      "loss": 1.661,
+      "step": 2033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8803941825884587e-05,
+      "loss": 1.2083,
+      "step": 2034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8802781375602276e-05,
+      "loss": 0.2664,
+      "step": 2035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.880162038604085e-05,
+      "loss": 1.0154,
+      "step": 2036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8800458857245673e-05,
+      "loss": 1.2455,
+      "step": 2037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.879929678926213e-05,
+      "loss": 0.4275,
+      "step": 2038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8798134182135615e-05,
+      "loss": 0.4066,
+      "step": 2039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8796971035911567e-05,
+      "loss": 0.6561,
+      "step": 2040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.879580735063541e-05,
+      "loss": 0.5349,
+      "step": 2041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8794643126352624e-05,
+      "loss": 0.8072,
+      "step": 2042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8793478363108687e-05,
+      "loss": 0.3939,
+      "step": 2043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8792313060949114e-05,
+      "loss": 0.8694,
+      "step": 2044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8791147219919428e-05,
+      "loss": 1.2247,
+      "step": 2045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.878998084006519e-05,
+      "loss": 0.7335,
+      "step": 2046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8788813921431954e-05,
+      "loss": 0.1794,
+      "step": 2047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.878764646406532e-05,
+      "loss": 0.5562,
+      "step": 2048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8786478468010904e-05,
+      "loss": 0.5022,
+      "step": 2049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8785309933314334e-05,
+      "loss": 0.3462,
+      "step": 2050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.878414086002127e-05,
+      "loss": 1.4873,
+      "step": 2051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8782971248177387e-05,
+      "loss": 1.3718,
+      "step": 2052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8781801097828377e-05,
+      "loss": 0.8473,
+      "step": 2053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8780630409019968e-05,
+      "loss": 2.0525,
+      "step": 2054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8779459181797888e-05,
+      "loss": 1.258,
+      "step": 2055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.877828741620791e-05,
+      "loss": 1.0839,
+      "step": 2056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8777115112295797e-05,
+      "loss": 1.8545,
+      "step": 2057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.877594227010737e-05,
+      "loss": 1.8161,
+      "step": 2058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8774768889688446e-05,
+      "loss": 0.4896,
+      "step": 2059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.877359497108486e-05,
+      "loss": 0.5773,
+      "step": 2060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8772420514342492e-05,
+      "loss": 0.7322,
+      "step": 2061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.877124551950722e-05,
+      "loss": 0.5422,
+      "step": 2062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8770069986624956e-05,
+      "loss": 0.9956,
+      "step": 2063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8768893915741623e-05,
+      "loss": 0.0845,
+      "step": 2064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8767717306903173e-05,
+      "loss": 2.1217,
+      "step": 2065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8766540160155576e-05,
+      "loss": 1.6621,
+      "step": 2066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8765362475544822e-05,
+      "loss": 0.8548,
+      "step": 2067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8764184253116932e-05,
+      "loss": 0.6445,
+      "step": 2068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8763005492917928e-05,
+      "loss": 0.2427,
+      "step": 2069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.876182619499387e-05,
+      "loss": 0.341,
+      "step": 2070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8760646359390834e-05,
+      "loss": 0.3954,
+      "step": 2071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8759465986154916e-05,
+      "loss": 1.3455,
+      "step": 2072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8758285075332235e-05,
+      "loss": 0.6743,
+      "step": 2073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8757103626968923e-05,
+      "loss": 1.4276,
+      "step": 2074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8755921641111148e-05,
+      "loss": 0.9957,
+      "step": 2075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.875473911780509e-05,
+      "loss": 0.2609,
+      "step": 2076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.875355605709694e-05,
+      "loss": 0.2871,
+      "step": 2077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8752372459032934e-05,
+      "loss": 1.1269,
+      "step": 2078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8751188323659306e-05,
+      "loss": 1.8778,
+      "step": 2079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.875000365102233e-05,
+      "loss": 0.2897,
+      "step": 2080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.874881844116828e-05,
+      "loss": 1.3052,
+      "step": 2081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.874763269414347e-05,
+      "loss": 1.4083,
+      "step": 2082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8746446409994225e-05,
+      "loss": 0.2816,
+      "step": 2083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8745259588766895e-05,
+      "loss": 0.7409,
+      "step": 2084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.874407223050785e-05,
+      "loss": 1.355,
+      "step": 2085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8742884335263476e-05,
+      "loss": 1.211,
+      "step": 2086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8741695903080195e-05,
+      "loss": 0.6792,
+      "step": 2087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8740506934004427e-05,
+      "loss": 0.545,
+      "step": 2088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8739317428082633e-05,
+      "loss": 1.5157,
+      "step": 2089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.873812738536128e-05,
+      "loss": 1.0535,
+      "step": 2090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.873693680588687e-05,
+      "loss": 0.6584,
+      "step": 2091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.873574568970592e-05,
+      "loss": 0.5132,
+      "step": 2092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8734554036864966e-05,
+      "loss": 0.8132,
+      "step": 2093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8733361847410563e-05,
+      "loss": 0.4817,
+      "step": 2094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.873216912138929e-05,
+      "loss": 1.4248,
+      "step": 2095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.873097585884775e-05,
+      "loss": 0.7393,
+      "step": 2096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.872978205983257e-05,
+      "loss": 0.8648,
+      "step": 2097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.872858772439038e-05,
+      "loss": 0.4702,
+      "step": 2098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.872739285256785e-05,
+      "loss": 0.8048,
+      "step": 2099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8726197444411666e-05,
+      "loss": 0.4019,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8725001499968525e-05,
+      "loss": 1.5081,
+      "step": 2101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.872380501928516e-05,
+      "loss": 0.5281,
+      "step": 2102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8722608002408315e-05,
+      "loss": 0.5214,
+      "step": 2103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.872141044938476e-05,
+      "loss": 0.6972,
+      "step": 2104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8720212360261286e-05,
+      "loss": 1.0127,
+      "step": 2105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.871901373508469e-05,
+      "loss": 1.1871,
+      "step": 2106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.871781457390182e-05,
+      "loss": 0.771,
+      "step": 2107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8716614876759518e-05,
+      "loss": 1.613,
+      "step": 2108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8715414643704655e-05,
+      "loss": 0.3573,
+      "step": 2109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8714213874784128e-05,
+      "loss": 0.2875,
+      "step": 2110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8713012570044853e-05,
+      "loss": 0.6021,
+      "step": 2111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8711810729533762e-05,
+      "loss": 1.5356,
+      "step": 2112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8710608353297813e-05,
+      "loss": 0.4691,
+      "step": 2113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8709405441383983e-05,
+      "loss": 0.8017,
+      "step": 2114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8708201993839265e-05,
+      "loss": 1.1179,
+      "step": 2115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.870699801071069e-05,
+      "loss": 1.2228,
+      "step": 2116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8705793492045288e-05,
+      "loss": 0.2065,
+      "step": 2117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.870458843789012e-05,
+      "loss": 0.6362,
+      "step": 2118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8703382848292276e-05,
+      "loss": 0.2249,
+      "step": 2119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.870217672329885e-05,
+      "loss": 0.5522,
+      "step": 2120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8700970062956974e-05,
+      "loss": 1.5922,
+      "step": 2121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8699762867313784e-05,
+      "loss": 1.4154,
+      "step": 2122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8698555136416445e-05,
+      "loss": 0.9187,
+      "step": 2123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8697346870312157e-05,
+      "loss": 0.6616,
+      "step": 2124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8696138069048112e-05,
+      "loss": 0.7355,
+      "step": 2125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8694928732671545e-05,
+      "loss": 1.31,
+      "step": 2126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8693718861229706e-05,
+      "loss": 0.6643,
+      "step": 2127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8692508454769867e-05,
+      "loss": 0.5123,
+      "step": 2128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.869129751333931e-05,
+      "loss": 0.3815,
+      "step": 2129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8690086036985355e-05,
+      "loss": 0.3577,
+      "step": 2130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.868887402575533e-05,
+      "loss": 1.4006,
+      "step": 2131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8687661479696594e-05,
+      "loss": 1.2517,
+      "step": 2132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8686448398856517e-05,
+      "loss": 1.1618,
+      "step": 2133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8685234783282494e-05,
+      "loss": 1.08,
+      "step": 2134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8684020633021948e-05,
+      "loss": 0.732,
+      "step": 2135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8682805948122307e-05,
+      "loss": 0.8782,
+      "step": 2136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.868159072863104e-05,
+      "loss": 0.7517,
+      "step": 2137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8680374974595614e-05,
+      "loss": 1.1441,
+      "step": 2138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8679158686063534e-05,
+      "loss": 1.7504,
+      "step": 2139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.867794186308232e-05,
+      "loss": 1.2453,
+      "step": 2140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8676724505699522e-05,
+      "loss": 1.2913,
+      "step": 2141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.867550661396269e-05,
+      "loss": 0.2214,
+      "step": 2142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8674288187919414e-05,
+      "loss": 0.6554,
+      "step": 2143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.86730692276173e-05,
+      "loss": 1.1088,
+      "step": 2144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8671849733103966e-05,
+      "loss": 0.6773,
+      "step": 2145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.867062970442707e-05,
+      "loss": 0.5307,
+      "step": 2146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8669409141634268e-05,
+      "loss": 0.2343,
+      "step": 2147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.866818804477325e-05,
+      "loss": 0.3464,
+      "step": 2148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.866696641389172e-05,
+      "loss": 0.9147,
+      "step": 2149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.866574424903742e-05,
+      "loss": 0.4967,
+      "step": 2150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8664521550258095e-05,
+      "loss": 1.1242,
+      "step": 2151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.866329831760151e-05,
+      "loss": 0.4567,
+      "step": 2152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8662074551115465e-05,
+      "loss": 0.4025,
+      "step": 2153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.866085025084777e-05,
+      "loss": 0.6363,
+      "step": 2154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.865962541684626e-05,
+      "loss": 0.2645,
+      "step": 2155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8658400049158787e-05,
+      "loss": 1.5326,
+      "step": 2156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.865717414783323e-05,
+      "loss": 1.3762,
+      "step": 2157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8655947712917477e-05,
+      "loss": 0.3712,
+      "step": 2158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8654720744459456e-05,
+      "loss": 0.3402,
+      "step": 2159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8653493242507105e-05,
+      "loss": 1.614,
+      "step": 2160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8652265207108373e-05,
+      "loss": 0.4682,
+      "step": 2161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8651036638311248e-05,
+      "loss": 1.7344,
+      "step": 2162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8649807536163725e-05,
+      "loss": 0.8574,
+      "step": 2163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.864857790071383e-05,
+      "loss": 1.6177,
+      "step": 2164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8647347732009605e-05,
+      "loss": 0.7825,
+      "step": 2165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.864611703009911e-05,
+      "loss": 0.2654,
+      "step": 2166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.864488579503043e-05,
+      "loss": 1.2046,
+      "step": 2167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8643654026851665e-05,
+      "loss": 0.6392,
+      "step": 2168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8642421725610953e-05,
+      "loss": 0.6202,
+      "step": 2169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8641188891356433e-05,
+      "loss": 0.4313,
+      "step": 2170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8639955524136273e-05,
+      "loss": 0.5896,
+      "step": 2171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8638721623998654e-05,
+      "loss": 0.3353,
+      "step": 2172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8637487190991797e-05,
+      "loss": 0.741,
+      "step": 2173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8636252225163927e-05,
+      "loss": 1.6102,
+      "step": 2174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.863501672656329e-05,
+      "loss": 0.4724,
+      "step": 2175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.863378069523816e-05,
+      "loss": 0.3239,
+      "step": 2176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8632544131236833e-05,
+      "loss": 1.5522,
+      "step": 2177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.863130703460762e-05,
+      "loss": 0.7459,
+      "step": 2178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8630069405398853e-05,
+      "loss": 0.8885,
+      "step": 2179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8628831243658884e-05,
+      "loss": 0.8305,
+      "step": 2180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.86275925494361e-05,
+      "loss": 1.8273,
+      "step": 2181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.862635332277888e-05,
+      "loss": 0.5418,
+      "step": 2182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8625113563735653e-05,
+      "loss": 0.3591,
+      "step": 2183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8623873272354854e-05,
+      "loss": 1.0142,
+      "step": 2184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8622632448684943e-05,
+      "loss": 0.859,
+      "step": 2185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8621391092774396e-05,
+      "loss": 0.3063,
+      "step": 2186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8620149204671713e-05,
+      "loss": 0.8682,
+      "step": 2187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.861890678442542e-05,
+      "loss": 0.7554,
+      "step": 2188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8617663832084057e-05,
+      "loss": 0.3365,
+      "step": 2189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8616420347696177e-05,
+      "loss": 0.6594,
+      "step": 2190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.861517633131038e-05,
+      "loss": 1.2238,
+      "step": 2191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.861393178297526e-05,
+      "loss": 0.5763,
+      "step": 2192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8612686702739438e-05,
+      "loss": 0.8063,
+      "step": 2193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.861144109065157e-05,
+      "loss": 0.6775,
+      "step": 2194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8610194946760316e-05,
+      "loss": 0.7446,
+      "step": 2195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8608948271114366e-05,
+      "loss": 0.1582,
+      "step": 2196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8607701063762424e-05,
+      "loss": 0.5162,
+      "step": 2197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8606453324753224e-05,
+      "loss": 0.3953,
+      "step": 2198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.860520505413551e-05,
+      "loss": 0.437,
+      "step": 2199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8603956251958067e-05,
+      "loss": 0.4171,
+      "step": 2200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8602706918269664e-05,
+      "loss": 1.3163,
+      "step": 2201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8601457053119126e-05,
+      "loss": 0.3083,
+      "step": 2202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8600206656555282e-05,
+      "loss": 1.6492,
+      "step": 2203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8598955728626986e-05,
+      "loss": 0.975,
+      "step": 2204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8597704269383118e-05,
+      "loss": 0.6802,
+      "step": 2205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8596452278872562e-05,
+      "loss": 1.0387,
+      "step": 2206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8595199757144243e-05,
+      "loss": 0.9584,
+      "step": 2207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8593946704247092e-05,
+      "loss": 0.2651,
+      "step": 2208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8592693120230068e-05,
+      "loss": 0.4256,
+      "step": 2209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.859143900514215e-05,
+      "loss": 0.5455,
+      "step": 2210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8590184359032332e-05,
+      "loss": 0.2315,
+      "step": 2211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.858892918194964e-05,
+      "loss": 0.8256,
+      "step": 2212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.858767347394311e-05,
+      "loss": 0.3737,
+      "step": 2213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8586417235061805e-05,
+      "loss": 0.5723,
+      "step": 2214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8585160465354808e-05,
+      "loss": 0.1025,
+      "step": 2215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8583903164871213e-05,
+      "loss": 1.4288,
+      "step": 2216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8582645333660152e-05,
+      "loss": 1.1072,
+      "step": 2217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8581386971770772e-05,
+      "loss": 0.773,
+      "step": 2218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8580128079252225e-05,
+      "loss": 0.408,
+      "step": 2219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8578868656153704e-05,
+      "loss": 1.8768,
+      "step": 2220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8577608702524414e-05,
+      "loss": 1.0886,
+      "step": 2221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8576348218413582e-05,
+      "loss": 0.4769,
+      "step": 2222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8575087203870457e-05,
+      "loss": 0.9422,
+      "step": 2223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8573825658944307e-05,
+      "loss": 0.9174,
+      "step": 2224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8572563583684414e-05,
+      "loss": 0.2905,
+      "step": 2225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8571300978140097e-05,
+      "loss": 0.85,
+      "step": 2226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8570037842360683e-05,
+      "loss": 0.6437,
+      "step": 2227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.856877417639552e-05,
+      "loss": 1.5173,
+      "step": 2228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8567509980293986e-05,
+      "loss": 1.2332,
+      "step": 2229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.856624525410547e-05,
+      "loss": 1.6446,
+      "step": 2230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8564979997879386e-05,
+      "loss": 0.8696,
+      "step": 2231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.856371421166517e-05,
+      "loss": 0.6708,
+      "step": 2232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8562447895512266e-05,
+      "loss": 0.3959,
+      "step": 2233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.856118104947016e-05,
+      "loss": 1.5553,
+      "step": 2234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.855991367358835e-05,
+      "loss": 0.7535,
+      "step": 2235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8558645767916347e-05,
+      "loss": 1.1361,
+      "step": 2236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.855737733250369e-05,
+      "loss": 0.7062,
+      "step": 2237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8556108367399932e-05,
+      "loss": 0.9392,
+      "step": 2238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.855483887265466e-05,
+      "loss": 0.8748,
+      "step": 2239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8553568848317472e-05,
+      "loss": 0.9156,
+      "step": 2240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8552298294437986e-05,
+      "loss": 1.0527,
+      "step": 2241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8551027211065844e-05,
+      "loss": 0.5927,
+      "step": 2242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.854975559825071e-05,
+      "loss": 0.7502,
+      "step": 2243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.854848345604226e-05,
+      "loss": 1.1279,
+      "step": 2244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.85472107844902e-05,
+      "loss": 0.5083,
+      "step": 2245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8545937583644256e-05,
+      "loss": 0.7914,
+      "step": 2246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.854466385355417e-05,
+      "loss": 0.514,
+      "step": 2247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8543389594269705e-05,
+      "loss": 0.2934,
+      "step": 2248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8542114805840655e-05,
+      "loss": 0.6945,
+      "step": 2249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.854083948831682e-05,
+      "loss": 0.5369,
+      "step": 2250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8539563641748027e-05,
+      "loss": 1.442,
+      "step": 2251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.853828726618412e-05,
+      "loss": 1.7108,
+      "step": 2252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.853701036167498e-05,
+      "loss": 0.7414,
+      "step": 2253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8535732928270484e-05,
+      "loss": 1.5435,
+      "step": 2254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8534454966020548e-05,
+      "loss": 1.0161,
+      "step": 2255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8533176474975095e-05,
+      "loss": 0.461,
+      "step": 2256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8531897455184087e-05,
+      "loss": 0.6555,
+      "step": 2257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8530617906697488e-05,
+      "loss": 1.545,
+      "step": 2258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.852933782956529e-05,
+      "loss": 0.5011,
+      "step": 2259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.852805722383751e-05,
+      "loss": 0.277,
+      "step": 2260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8526776089564183e-05,
+      "loss": 1.0479,
+      "step": 2261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8525494426795362e-05,
+      "loss": 0.289,
+      "step": 2262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8524212235581113e-05,
+      "loss": 1.4451,
+      "step": 2263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8522929515971547e-05,
+      "loss": 0.3868,
+      "step": 2264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8521646268016767e-05,
+      "loss": 0.3696,
+      "step": 2265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8520362491766917e-05,
+      "loss": 0.7179,
+      "step": 2266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8519078187272153e-05,
+      "loss": 0.6991,
+      "step": 2267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8517793354582653e-05,
+      "loss": 0.8296,
+      "step": 2268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8516507993748615e-05,
+      "loss": 0.5615,
+      "step": 2269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.851522210482026e-05,
+      "loss": 0.2266,
+      "step": 2270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8513935687847823e-05,
+      "loss": 0.4334,
+      "step": 2271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8512648742881573e-05,
+      "loss": 0.3621,
+      "step": 2272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.851136126997179e-05,
+      "loss": 0.5672,
+      "step": 2273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.851007326916877e-05,
+      "loss": 0.6527,
+      "step": 2274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.850878474052284e-05,
+      "loss": 0.5483,
+      "step": 2275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8507495684084344e-05,
+      "loss": 0.8868,
+      "step": 2276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.850620609990364e-05,
+      "loss": 0.5603,
+      "step": 2277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.850491598803112e-05,
+      "loss": 0.3865,
+      "step": 2278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8503625348517182e-05,
+      "loss": 0.8078,
+      "step": 2279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8502334181412258e-05,
+      "loss": 1.4848,
+      "step": 2280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.850104248676679e-05,
+      "loss": 0.9032,
+      "step": 2281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8499750264631247e-05,
+      "loss": 0.4345,
+      "step": 2282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.849845751505612e-05,
+      "loss": 0.3407,
+      "step": 2283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8497164238091912e-05,
+      "loss": 0.8656,
+      "step": 2284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8495870433789154e-05,
+      "loss": 0.1427,
+      "step": 2285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8494576102198385e-05,
+      "loss": 0.7255,
+      "step": 2286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8493281243370193e-05,
+      "loss": 0.902,
+      "step": 2287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.849198585735516e-05,
+      "loss": 0.1859,
+      "step": 2288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8490689944203898e-05,
+      "loss": 1.3428,
+      "step": 2289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8489393503967032e-05,
+      "loss": 1.2865,
+      "step": 2290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8488096536695225e-05,
+      "loss": 0.673,
+      "step": 2291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8486799042439143e-05,
+      "loss": 0.9434,
+      "step": 2292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8485501021249478e-05,
+      "loss": 1.3295,
+      "step": 2293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.848420247317695e-05,
+      "loss": 0.4196,
+      "step": 2294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8482903398272296e-05,
+      "loss": 0.2449,
+      "step": 2295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8481603796586264e-05,
+      "loss": 0.6025,
+      "step": 2296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8480303668169632e-05,
+      "loss": 0.5782,
+      "step": 2297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8479003013073196e-05,
+      "loss": 0.332,
+      "step": 2298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8477701831347774e-05,
+      "loss": 0.1901,
+      "step": 2299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8476400123044204e-05,
+      "loss": 0.7209,
+      "step": 2300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8475097888213344e-05,
+      "loss": 1.6628,
+      "step": 2301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8473795126906072e-05,
+      "loss": 1.4217,
+      "step": 2302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8472491839173292e-05,
+      "loss": 1.479,
+      "step": 2303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8471188025065914e-05,
+      "loss": 1.8512,
+      "step": 2304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8469883684634885e-05,
+      "loss": 0.4489,
+      "step": 2305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8468578817931166e-05,
+      "loss": 0.5134,
+      "step": 2306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8467273425005738e-05,
+      "loss": 1.2628,
+      "step": 2307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.84659675059096e-05,
+      "loss": 0.417,
+      "step": 2308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8464661060693782e-05,
+      "loss": 0.4304,
+      "step": 2309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.846335408940932e-05,
+      "loss": 0.4859,
+      "step": 2310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8462046592107274e-05,
+      "loss": 0.1871,
+      "step": 2311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8460738568838746e-05,
+      "loss": 0.4369,
+      "step": 2312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8459430019654823e-05,
+      "loss": 0.5547,
+      "step": 2313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.845812094460664e-05,
+      "loss": 0.6887,
+      "step": 2314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.845681134374534e-05,
+      "loss": 0.1954,
+      "step": 2315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8455501217122088e-05,
+      "loss": 0.4573,
+      "step": 2316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8454190564788073e-05,
+      "loss": 0.5414,
+      "step": 2317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.84528793867945e-05,
+      "loss": 1.7716,
+      "step": 2318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8451567683192605e-05,
+      "loss": 0.3306,
+      "step": 2319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.845025545403363e-05,
+      "loss": 1.7633,
+      "step": 2320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8448942699368843e-05,
+      "loss": 1.0782,
+      "step": 2321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8447629419249533e-05,
+      "loss": 0.6922,
+      "step": 2322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.844631561372702e-05,
+      "loss": 0.3645,
+      "step": 2323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.844500128285262e-05,
+      "loss": 0.4017,
+      "step": 2324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.84436864266777e-05,
+      "loss": 0.5701,
+      "step": 2325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.844237104525362e-05,
+      "loss": 0.8476,
+      "step": 2326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.844105513863178e-05,
+      "loss": 1.7606,
+      "step": 2327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8439738706863593e-05,
+      "loss": 0.7135,
+      "step": 2328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8438421750000484e-05,
+      "loss": 1.6956,
+      "step": 2329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8437104268093907e-05,
+      "loss": 0.5893,
+      "step": 2330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8435786261195347e-05,
+      "loss": 1.4687,
+      "step": 2331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8434467729356293e-05,
+      "loss": 1.4355,
+      "step": 2332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8433148672628262e-05,
+      "loss": 1.7407,
+      "step": 2333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8431829091062788e-05,
+      "loss": 0.9194,
+      "step": 2334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8430508984711424e-05,
+      "loss": 1.4749,
+      "step": 2335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.842918835362576e-05,
+      "loss": 0.8393,
+      "step": 2336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8427867197857378e-05,
+      "loss": 0.8625,
+      "step": 2337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8426545517457902e-05,
+      "loss": 0.7762,
+      "step": 2338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8425223312478975e-05,
+      "loss": 0.1598,
+      "step": 2339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.842390058297225e-05,
+      "loss": 0.9616,
+      "step": 2340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.842257732898941e-05,
+      "loss": 0.6108,
+      "step": 2341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8421253550582153e-05,
+      "loss": 0.517,
+      "step": 2342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.84199292478022e-05,
+      "loss": 0.7725,
+      "step": 2343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8418604420701292e-05,
+      "loss": 0.5179,
+      "step": 2344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8417279069331187e-05,
+      "loss": 0.785,
+      "step": 2345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8415953193743677e-05,
+      "loss": 0.2993,
+      "step": 2346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8414626793990553e-05,
+      "loss": 0.4976,
+      "step": 2347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8413299870123643e-05,
+      "loss": 0.5491,
+      "step": 2348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.841197242219479e-05,
+      "loss": 0.617,
+      "step": 2349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.841064445025586e-05,
+      "loss": 0.481,
+      "step": 2350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8409315954358734e-05,
+      "loss": 1.086,
+      "step": 2351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8407986934555318e-05,
+      "loss": 1.3927,
+      "step": 2352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8406657390897536e-05,
+      "loss": 0.6717,
+      "step": 2353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.840532732343733e-05,
+      "loss": 0.3717,
+      "step": 2354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8403996732226678e-05,
+      "loss": 0.4516,
+      "step": 2355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8402665617317557e-05,
+      "loss": 0.5142,
+      "step": 2356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8401333978761975e-05,
+      "loss": 0.4016,
+      "step": 2357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8400001816611964e-05,
+      "loss": 0.0523,
+      "step": 2358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.839866913091957e-05,
+      "loss": 0.2515,
+      "step": 2359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8397335921736858e-05,
+      "loss": 0.8816,
+      "step": 2360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.839600218911592e-05,
+      "loss": 0.8239,
+      "step": 2361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8394667933108863e-05,
+      "loss": 1.3041,
+      "step": 2362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8393333153767822e-05,
+      "loss": 1.3931,
+      "step": 2363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.839199785114494e-05,
+      "loss": 0.6433,
+      "step": 2364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8390662025292395e-05,
+      "loss": 0.7839,
+      "step": 2365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.838932567626237e-05,
+      "loss": 0.6231,
+      "step": 2366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8387988804107087e-05,
+      "loss": 0.6103,
+      "step": 2367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8386651408878768e-05,
+      "loss": 0.2832,
+      "step": 2368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8385313490629675e-05,
+      "loss": 0.6527,
+      "step": 2369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8383975049412073e-05,
+      "loss": 0.7316,
+      "step": 2370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8382636085278257e-05,
+      "loss": 0.3794,
+      "step": 2371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.838129659828054e-05,
+      "loss": 1.1205,
+      "step": 2372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8379956588471264e-05,
+      "loss": 1.6199,
+      "step": 2373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.837861605590278e-05,
+      "loss": 0.8711,
+      "step": 2374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.837727500062745e-05,
+      "loss": 0.6236,
+      "step": 2375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8375933422697685e-05,
+      "loss": 0.6297,
+      "step": 2376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8374591322165895e-05,
+      "loss": 1.739,
+      "step": 2377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.837324869908452e-05,
+      "loss": 0.8146,
+      "step": 2378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8371905553506014e-05,
+      "loss": 1.2015,
+      "step": 2379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.837056188548285e-05,
+      "loss": 0.5816,
+      "step": 2380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8369217695067534e-05,
+      "loss": 0.7858,
+      "step": 2381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.836787298231258e-05,
+      "loss": 1.205,
+      "step": 2382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8366527747270522e-05,
+      "loss": 0.5187,
+      "step": 2383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8365181989993923e-05,
+      "loss": 0.9454,
+      "step": 2384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8363835710535367e-05,
+      "loss": 0.6533,
+      "step": 2385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8362488908947443e-05,
+      "loss": 0.7456,
+      "step": 2386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.836114158528278e-05,
+      "loss": 1.1223,
+      "step": 2387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8359793739594014e-05,
+      "loss": 0.9089,
+      "step": 2388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8358445371933806e-05,
+      "loss": 1.5883,
+      "step": 2389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8357096482354837e-05,
+      "loss": 1.0179,
+      "step": 2390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.835574707090981e-05,
+      "loss": 1.4273,
+      "step": 2391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8354397137651445e-05,
+      "loss": 1.1146,
+      "step": 2392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.835304668263249e-05,
+      "loss": 0.572,
+      "step": 2393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.83516957059057e-05,
+      "loss": 0.618,
+      "step": 2394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8350344207523864e-05,
+      "loss": 0.6031,
+      "step": 2395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8348992187539777e-05,
+      "loss": 0.7153,
+      "step": 2396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8347639646006275e-05,
+      "loss": 0.7097,
+      "step": 2397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8346286582976194e-05,
+      "loss": 0.8157,
+      "step": 2398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.83449329985024e-05,
+      "loss": 0.4566,
+      "step": 2399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8343578892637774e-05,
+      "loss": 0.9057,
+      "step": 2400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8342224265435233e-05,
+      "loss": 1.8414,
+      "step": 2401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8340869116947695e-05,
+      "loss": 1.2314,
+      "step": 2402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8339513447228105e-05,
+      "loss": 1.2238,
+      "step": 2403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.833815725632943e-05,
+      "loss": 0.4516,
+      "step": 2404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.833680054430466e-05,
+      "loss": 0.2382,
+      "step": 2405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8335443311206797e-05,
+      "loss": 1.4578,
+      "step": 2406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.833408555708888e-05,
+      "loss": 0.6199,
+      "step": 2407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8332727282003937e-05,
+      "loss": 0.3358,
+      "step": 2408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8331368486005052e-05,
+      "loss": 1.0793,
+      "step": 2409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8330009169145313e-05,
+      "loss": 0.1524,
+      "step": 2410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8328649331477825e-05,
+      "loss": 0.1169,
+      "step": 2411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8327288973055715e-05,
+      "loss": 0.3371,
+      "step": 2412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8325928093932135e-05,
+      "loss": 0.4052,
+      "step": 2413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8324566694160258e-05,
+      "loss": 0.518,
+      "step": 2414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.832320477379327e-05,
+      "loss": 0.3789,
+      "step": 2415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.832184233288438e-05,
+      "loss": 0.3348,
+      "step": 2416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8320479371486828e-05,
+      "loss": 1.4692,
+      "step": 2417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.831911588965386e-05,
+      "loss": 0.4627,
+      "step": 2418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8317751887438744e-05,
+      "loss": 0.7154,
+      "step": 2419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8316387364894773e-05,
+      "loss": 1.6406,
+      "step": 2420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8315022322075265e-05,
+      "loss": 0.2176,
+      "step": 2421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8313656759033552e-05,
+      "loss": 0.7014,
+      "step": 2422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.831229067582298e-05,
+      "loss": 1.0666,
+      "step": 2423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8310924072496926e-05,
+      "loss": 1.5503,
+      "step": 2424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8309556949108787e-05,
+      "loss": 0.5107,
+      "step": 2425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8308189305711973e-05,
+      "loss": 0.4967,
+      "step": 2426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8306821142359918e-05,
+      "loss": 0.2605,
+      "step": 2427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8305452459106078e-05,
+      "loss": 0.5465,
+      "step": 2428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8304083256003934e-05,
+      "loss": 1.388,
+      "step": 2429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8302713533106967e-05,
+      "loss": 0.4863,
+      "step": 2430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8301343290468706e-05,
+      "loss": 0.9005,
+      "step": 2431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8299972528142678e-05,
+      "loss": 0.6361,
+      "step": 2432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8298601246182445e-05,
+      "loss": 0.9905,
+      "step": 2433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.829722944464158e-05,
+      "loss": 0.4787,
+      "step": 2434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8295857123573684e-05,
+      "loss": 0.4497,
+      "step": 2435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8294484283032367e-05,
+      "loss": 1.1969,
+      "step": 2436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.829311092307127e-05,
+      "loss": 1.2402,
+      "step": 2437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.829173704374405e-05,
+      "loss": 0.9006,
+      "step": 2438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.829036264510439e-05,
+      "loss": 0.6199,
+      "step": 2439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8288987727205982e-05,
+      "loss": 1.3409,
+      "step": 2440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.828761229010255e-05,
+      "loss": 0.9979,
+      "step": 2441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.828623633384782e-05,
+      "loss": 0.9949,
+      "step": 2442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.828485985849557e-05,
+      "loss": 0.6276,
+      "step": 2443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8283482864099562e-05,
+      "loss": 0.8854,
+      "step": 2444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8282105350713604e-05,
+      "loss": 0.4137,
+      "step": 2445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.828072731839152e-05,
+      "loss": 0.8608,
+      "step": 2446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.827934876718714e-05,
+      "loss": 0.1629,
+      "step": 2447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.827796969715433e-05,
+      "loss": 0.5754,
+      "step": 2448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.827659010834697e-05,
+      "loss": 1.0083,
+      "step": 2449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8275210000818966e-05,
+      "loss": 0.6974,
+      "step": 2450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.827382937462423e-05,
+      "loss": 1.4156,
+      "step": 2451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8272448229816707e-05,
+      "loss": 1.0312,
+      "step": 2452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8271066566450364e-05,
+      "loss": 1.2317,
+      "step": 2453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8269684384579176e-05,
+      "loss": 1.7931,
+      "step": 2454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8268301684257146e-05,
+      "loss": 0.3011,
+      "step": 2455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8266918465538305e-05,
+      "loss": 0.1528,
+      "step": 2456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8265534728476685e-05,
+      "loss": 1.5696,
+      "step": 2457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.826415047312636e-05,
+      "loss": 1.7693,
+      "step": 2458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8262765699541398e-05,
+      "loss": 0.5569,
+      "step": 2459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8261380407775917e-05,
+      "loss": 1.0565,
+      "step": 2460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.825999459788403e-05,
+      "loss": 0.2397,
+      "step": 2461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8258608269919892e-05,
+      "loss": 0.7677,
+      "step": 2462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8257221423937658e-05,
+      "loss": 1.8223,
+      "step": 2463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.825583405999152e-05,
+      "loss": 0.3719,
+      "step": 2464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.825444617813567e-05,
+      "loss": 1.4185,
+      "step": 2465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.825305777842435e-05,
+      "loss": 0.7226,
+      "step": 2466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8251668860911794e-05,
+      "loss": 1.7135,
+      "step": 2467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.825027942565227e-05,
+      "loss": 0.1352,
+      "step": 2468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8248889472700068e-05,
+      "loss": 2.1564,
+      "step": 2469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8247499002109482e-05,
+      "loss": 0.4928,
+      "step": 2470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8246108013934853e-05,
+      "loss": 0.7996,
+      "step": 2471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8244716508230517e-05,
+      "loss": 1.6098,
+      "step": 2472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.824332448505085e-05,
+      "loss": 0.4708,
+      "step": 2473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8241931944450226e-05,
+      "loss": 1.4939,
+      "step": 2474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8240538886483062e-05,
+      "loss": 1.0713,
+      "step": 2475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8239145311203783e-05,
+      "loss": 1.4969,
+      "step": 2476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8237751218666833e-05,
+      "loss": 1.2921,
+      "step": 2477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8236356608926683e-05,
+      "loss": 0.6725,
+      "step": 2478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.823496148203782e-05,
+      "loss": 0.2307,
+      "step": 2479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8233565838054755e-05,
+      "loss": 1.1564,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8232169677032007e-05,
+      "loss": 0.928,
+      "step": 2481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8230772999024137e-05,
+      "loss": 0.5948,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8229375804085706e-05,
+      "loss": 1.1634,
+      "step": 2483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.82279780922713e-05,
+      "loss": 0.7166,
+      "step": 2484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8226579863635536e-05,
+      "loss": 0.8679,
+      "step": 2485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.822518111823304e-05,
+      "loss": 1.0678,
+      "step": 2486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.822378185611846e-05,
+      "loss": 0.6409,
+      "step": 2487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8222382077346468e-05,
+      "loss": 0.9251,
+      "step": 2488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.822098178197175e-05,
+      "loss": 0.2083,
+      "step": 2489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.821958097004902e-05,
+      "loss": 1.2307,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.821817964163301e-05,
+      "loss": 1.1221,
+      "step": 2491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.821677779677846e-05,
+      "loss": 1.0706,
+      "step": 2492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8215375435540153e-05,
+      "loss": 0.7131,
+      "step": 2493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8213972557972872e-05,
+      "loss": 2.0153,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8212569164131434e-05,
+      "loss": 0.718,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8211165254070663e-05,
+      "loss": 0.9998,
+      "step": 2496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.820976082784541e-05,
+      "loss": 0.9591,
+      "step": 2497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8208355885510553e-05,
+      "loss": 0.1465,
+      "step": 2498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8206950427120984e-05,
+      "loss": 0.5116,
+      "step": 2499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8205544452731606e-05,
+      "loss": 0.4495,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.820413796239736e-05,
+      "loss": 1.4976,
+      "step": 2501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.820273095617319e-05,
+      "loss": 1.0922,
+      "step": 2502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.820132343411407e-05,
+      "loss": 0.6776,
+      "step": 2503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8199915396275e-05,
+      "loss": 0.3167,
+      "step": 2504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8198506842710986e-05,
+      "loss": 0.8686,
+      "step": 2505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.819709777347706e-05,
+      "loss": 1.7006,
+      "step": 2506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8195688188628277e-05,
+      "loss": 0.5059,
+      "step": 2507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.819427808821971e-05,
+      "loss": 1.7158,
+      "step": 2508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8192867472306448e-05,
+      "loss": 0.9211,
+      "step": 2509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.819145634094361e-05,
+      "loss": 0.4383,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8190044694186326e-05,
+      "loss": 0.9074,
+      "step": 2511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8188632532089754e-05,
+      "loss": 1.5228,
+      "step": 2512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.818721985470906e-05,
+      "loss": 0.1644,
+      "step": 2513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8185806662099444e-05,
+      "loss": 0.3984,
+      "step": 2514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8184392954316116e-05,
+      "loss": 0.3401,
+      "step": 2515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8182978731414316e-05,
+      "loss": 0.8103,
+      "step": 2516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.818156399344929e-05,
+      "loss": 0.72,
+      "step": 2517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8180148740476317e-05,
+      "loss": 0.8483,
+      "step": 2518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.817873297255069e-05,
+      "loss": 1.4009,
+      "step": 2519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8177316689727724e-05,
+      "loss": 1.5385,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8175899892062757e-05,
+      "loss": 1.2578,
+      "step": 2521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8174482579611134e-05,
+      "loss": 1.6751,
+      "step": 2522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8173064752428242e-05,
+      "loss": 0.4361,
+      "step": 2523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8171646410569466e-05,
+      "loss": 1.0738,
+      "step": 2524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.817022755409023e-05,
+      "loss": 1.5396,
+      "step": 2525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8168808183045964e-05,
+      "loss": 1.1586,
+      "step": 2526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8167388297492118e-05,
+      "loss": 0.3833,
+      "step": 2527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.816596789748418e-05,
+      "loss": 1.1406,
+      "step": 2528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8164546983077633e-05,
+      "loss": 0.3555,
+      "step": 2529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8163125554328003e-05,
+      "loss": 1.426,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8161703611290817e-05,
+      "loss": 0.7532,
+      "step": 2531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.816028115402164e-05,
+      "loss": 1.6836,
+      "step": 2532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8158858182576035e-05,
+      "loss": 0.6934,
+      "step": 2533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.815743469700961e-05,
+      "loss": 0.4877,
+      "step": 2534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8156010697377977e-05,
+      "loss": 1.1628,
+      "step": 2535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8154586183736772e-05,
+      "loss": 0.8057,
+      "step": 2536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.815316115614165e-05,
+      "loss": 0.1805,
+      "step": 2537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8151735614648285e-05,
+      "loss": 0.2808,
+      "step": 2538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.815030955931238e-05,
+      "loss": 0.8578,
+      "step": 2539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8148882990189652e-05,
+      "loss": 0.6748,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8147455907335826e-05,
+      "loss": 0.5305,
+      "step": 2541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.814602831080667e-05,
+      "loss": 0.7414,
+      "step": 2542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8144600200657953e-05,
+      "loss": 0.5897,
+      "step": 2543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.814317157694548e-05,
+      "loss": 0.5943,
+      "step": 2544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8141742439725062e-05,
+      "loss": 1.1445,
+      "step": 2545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8140312789052534e-05,
+      "loss": 0.7426,
+      "step": 2546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8138882624983758e-05,
+      "loss": 0.6826,
+      "step": 2547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8137451947574612e-05,
+      "loss": 0.1161,
+      "step": 2548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8136020756880985e-05,
+      "loss": 0.7232,
+      "step": 2549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8134589052958804e-05,
+      "loss": 0.5236,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8133156835863998e-05,
+      "loss": 0.6205,
+      "step": 2551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8131724105652526e-05,
+      "loss": 1.1006,
+      "step": 2552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.813029086238037e-05,
+      "loss": 0.4567,
+      "step": 2553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8128857106103525e-05,
+      "loss": 1.4275,
+      "step": 2554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8127422836878006e-05,
+      "loss": 0.6222,
+      "step": 2555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8125988054759852e-05,
+      "loss": 0.1673,
+      "step": 2556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8124552759805115e-05,
+      "loss": 1.5985,
+      "step": 2557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8123116952069883e-05,
+      "loss": 0.6259,
+      "step": 2558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8121680631610243e-05,
+      "loss": 0.5104,
+      "step": 2559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.812024379848232e-05,
+      "loss": 1.633,
+      "step": 2560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8118806452742247e-05,
+      "loss": 0.8539,
+      "step": 2561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8117368594446186e-05,
+      "loss": 0.1565,
+      "step": 2562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8115930223650313e-05,
+      "loss": 0.0567,
+      "step": 2563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8114491340410824e-05,
+      "loss": 0.2133,
+      "step": 2564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8113051944783937e-05,
+      "loss": 1.6222,
+      "step": 2565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.811161203682589e-05,
+      "loss": 0.3915,
+      "step": 2566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.811017161659294e-05,
+      "loss": 0.8772,
+      "step": 2567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8108730684141362e-05,
+      "loss": 1.6632,
+      "step": 2568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.810728923952746e-05,
+      "loss": 0.6288,
+      "step": 2569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.810584728280755e-05,
+      "loss": 1.2203,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8104404814037967e-05,
+      "loss": 0.4926,
+      "step": 2571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8102961833275074e-05,
+      "loss": 0.6127,
+      "step": 2572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.810151834057524e-05,
+      "loss": 0.7567,
+      "step": 2573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.810007433599487e-05,
+      "loss": 2.0027,
+      "step": 2574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.809862981959038e-05,
+      "loss": 1.7955,
+      "step": 2575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8097184791418206e-05,
+      "loss": 1.5395,
+      "step": 2576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8095739251534805e-05,
+      "loss": 0.7743,
+      "step": 2577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.809429319999666e-05,
+      "loss": 0.6427,
+      "step": 2578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.809284663686026e-05,
+      "loss": 0.4518,
+      "step": 2579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.809139956218213e-05,
+      "loss": 0.7609,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8089951976018806e-05,
+      "loss": 0.317,
+      "step": 2581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8088503878426847e-05,
+      "loss": 1.2122,
+      "step": 2582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8087055269462825e-05,
+      "loss": 1.016,
+      "step": 2583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8085606149183346e-05,
+      "loss": 0.9981,
+      "step": 2584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.808415651764502e-05,
+      "loss": 1.1051,
+      "step": 2585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8082706374904488e-05,
+      "loss": 0.8578,
+      "step": 2586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8081255721018407e-05,
+      "loss": 0.7178,
+      "step": 2587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8079804556043455e-05,
+      "loss": 0.687,
+      "step": 2588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.807835288003633e-05,
+      "loss": 0.7374,
+      "step": 2589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8076900693053747e-05,
+      "loss": 0.8209,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8075447995152448e-05,
+      "loss": 0.3677,
+      "step": 2591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8073994786389182e-05,
+      "loss": 0.5775,
+      "step": 2592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8072541066820737e-05,
+      "loss": 0.2502,
+      "step": 2593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8071086836503902e-05,
+      "loss": 0.649,
+      "step": 2594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8069632095495498e-05,
+      "loss": 0.3428,
+      "step": 2595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8068176843852363e-05,
+      "loss": 1.2674,
+      "step": 2596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.806672108163135e-05,
+      "loss": 0.6866,
+      "step": 2597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.806526480888934e-05,
+      "loss": 0.5387,
+      "step": 2598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.806380802568323e-05,
+      "loss": 0.7416,
+      "step": 2599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8062350732069933e-05,
+      "loss": 0.4193,
+      "step": 2600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.806089292810639e-05,
+      "loss": 0.4264,
+      "step": 2601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8059434613849558e-05,
+      "loss": 0.3385,
+      "step": 2602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.805797578935641e-05,
+      "loss": 1.2593,
+      "step": 2603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8056516454683947e-05,
+      "loss": 0.475,
+      "step": 2604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8055056609889182e-05,
+      "loss": 0.4459,
+      "step": 2605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8053596255029153e-05,
+      "loss": 1.13,
+      "step": 2606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8052135390160917e-05,
+      "loss": 0.6242,
+      "step": 2607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8050674015341557e-05,
+      "loss": 0.3546,
+      "step": 2608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8049212130628152e-05,
+      "loss": 0.4753,
+      "step": 2609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8047749736077838e-05,
+      "loss": 0.1414,
+      "step": 2610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8046286831747737e-05,
+      "loss": 0.7163,
+      "step": 2611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8044823417695012e-05,
+      "loss": 0.2986,
+      "step": 2612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8043359493976836e-05,
+      "loss": 0.9592,
+      "step": 2613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8041895060650408e-05,
+      "loss": 1.0325,
+      "step": 2614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.804043011777294e-05,
+      "loss": 1.9734,
+      "step": 2615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8038964665401672e-05,
+      "loss": 0.7335,
+      "step": 2616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8037498703593862e-05,
+      "loss": 0.6052,
+      "step": 2617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8036032232406774e-05,
+      "loss": 0.1972,
+      "step": 2618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8034565251897717e-05,
+      "loss": 1.3754,
+      "step": 2619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8033097762123996e-05,
+      "loss": 0.9516,
+      "step": 2620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.803162976314295e-05,
+      "loss": 0.0335,
+      "step": 2621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8030161255011942e-05,
+      "loss": 0.6552,
+      "step": 2622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.802869223778833e-05,
+      "loss": 0.6474,
+      "step": 2623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8027222711529528e-05,
+      "loss": 0.6042,
+      "step": 2624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.802575267629294e-05,
+      "loss": 1.6717,
+      "step": 2625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8024282132135994e-05,
+      "loss": 0.636,
+      "step": 2626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8022811079116163e-05,
+      "loss": 1.0585,
+      "step": 2627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8021339517290907e-05,
+      "loss": 0.7685,
+      "step": 2628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8019867446717725e-05,
+      "loss": 0.289,
+      "step": 2629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8018394867454132e-05,
+      "loss": 2.3215,
+      "step": 2630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8016921779557662e-05,
+      "loss": 0.3542,
+      "step": 2631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8015448183085867e-05,
+      "loss": 1.0683,
+      "step": 2632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.801397407809632e-05,
+      "loss": 0.4103,
+      "step": 2633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8012499464646622e-05,
+      "loss": 0.3338,
+      "step": 2634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8011024342794382e-05,
+      "loss": 0.8608,
+      "step": 2635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8009548712597233e-05,
+      "loss": 0.7587,
+      "step": 2636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8008072574112827e-05,
+      "loss": 0.4536,
+      "step": 2637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8006595927398834e-05,
+      "loss": 0.8216,
+      "step": 2638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.800511877251296e-05,
+      "loss": 1.3937,
+      "step": 2639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8003641109512906e-05,
+      "loss": 0.7494,
+      "step": 2640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.800216293845641e-05,
+      "loss": 0.9651,
+      "step": 2641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8000684259401222e-05,
+      "loss": 0.9029,
+      "step": 2642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7999205072405113e-05,
+      "loss": 0.3503,
+      "step": 2643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.799772537752588e-05,
+      "loss": 0.2615,
+      "step": 2644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7996245174821333e-05,
+      "loss": 1.1785,
+      "step": 2645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7994764464349306e-05,
+      "loss": 1.4029,
+      "step": 2646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7993283246167644e-05,
+      "loss": 0.7375,
+      "step": 2647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.799180152033423e-05,
+      "loss": 0.3507,
+      "step": 2648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7990319286906946e-05,
+      "loss": 0.3958,
+      "step": 2649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7988836545943706e-05,
+      "loss": 0.6615,
+      "step": 2650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7987353297502444e-05,
+      "loss": 0.3681,
+      "step": 2651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7985869541641108e-05,
+      "loss": 0.4446,
+      "step": 2652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7984385278417665e-05,
+      "loss": 1.3166,
+      "step": 2653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7982900507890113e-05,
+      "loss": 0.2556,
+      "step": 2654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.798141523011646e-05,
+      "loss": 0.6795,
+      "step": 2655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7979929445154736e-05,
+      "loss": 1.0179,
+      "step": 2656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7978443153062988e-05,
+      "loss": 0.4764,
+      "step": 2657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.797695635389929e-05,
+      "loss": 1.1036,
+      "step": 2658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7975469047721733e-05,
+      "loss": 1.372,
+      "step": 2659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7973981234588422e-05,
+      "loss": 1.8305,
+      "step": 2660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7972492914557487e-05,
+      "loss": 1.4713,
+      "step": 2661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.797100408768708e-05,
+      "loss": 1.6202,
+      "step": 2662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7969514754035373e-05,
+      "loss": 0.5843,
+      "step": 2663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7968024913660545e-05,
+      "loss": 0.412,
+      "step": 2664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7966534566620807e-05,
+      "loss": 0.3473,
+      "step": 2665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7965043712974396e-05,
+      "loss": 0.156,
+      "step": 2666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7963552352779553e-05,
+      "loss": 0.9213,
+      "step": 2667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7962060486094544e-05,
+      "loss": 0.3469,
+      "step": 2668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7960568112977665e-05,
+      "loss": 1.4076,
+      "step": 2669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7959075233487213e-05,
+      "loss": 0.9498,
+      "step": 2670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.795758184768152e-05,
+      "loss": 0.5555,
+      "step": 2671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7956087955618942e-05,
+      "loss": 0.7116,
+      "step": 2672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.795459355735783e-05,
+      "loss": 0.3512,
+      "step": 2673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.795309865295658e-05,
+      "loss": 0.9039,
+      "step": 2674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7951603242473596e-05,
+      "loss": 1.3201,
+      "step": 2675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7950107325967305e-05,
+      "loss": 0.6176,
+      "step": 2676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7948610903496156e-05,
+      "loss": 1.021,
+      "step": 2677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7947113975118608e-05,
+      "loss": 1.2823,
+      "step": 2678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.794561654089315e-05,
+      "loss": 0.3705,
+      "step": 2679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7944118600878286e-05,
+      "loss": 0.3209,
+      "step": 2680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7942620155132547e-05,
+      "loss": 0.5982,
+      "step": 2681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7941121203714466e-05,
+      "loss": 0.4717,
+      "step": 2682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.793962174668262e-05,
+      "loss": 0.7654,
+      "step": 2683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7938121784095584e-05,
+      "loss": 0.267,
+      "step": 2684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7936621316011967e-05,
+      "loss": 1.4805,
+      "step": 2685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7935120342490394e-05,
+      "loss": 1.1848,
+      "step": 2686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7933618863589503e-05,
+      "loss": 0.4685,
+      "step": 2687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.793211687936796e-05,
+      "loss": 0.959,
+      "step": 2688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.793061438988445e-05,
+      "loss": 0.6146,
+      "step": 2689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7929111395197672e-05,
+      "loss": 0.8609,
+      "step": 2690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.792760789536635e-05,
+      "loss": 0.8339,
+      "step": 2691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7926103890449225e-05,
+      "loss": 0.8016,
+      "step": 2692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7924599380505066e-05,
+      "loss": 0.4501,
+      "step": 2693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7923094365592648e-05,
+      "loss": 0.4935,
+      "step": 2694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7921588845770774e-05,
+      "loss": 0.6648,
+      "step": 2695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7920082821098263e-05,
+      "loss": 0.8984,
+      "step": 2696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7918576291633957e-05,
+      "loss": 1.6825,
+      "step": 2697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7917069257436717e-05,
+      "loss": 0.673,
+      "step": 2698
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7915561718565426e-05,
+      "loss": 1.18,
+      "step": 2699
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.791405367507898e-05,
+      "loss": 0.6263,
+      "step": 2700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.79125451270363e-05,
+      "loss": 0.3666,
+      "step": 2701
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7911036074496327e-05,
+      "loss": 0.3331,
+      "step": 2702
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7909526517518024e-05,
+      "loss": 1.1277,
+      "step": 2703
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7908016456160356e-05,
+      "loss": 0.2279,
+      "step": 2704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7906505890482337e-05,
+      "loss": 0.167,
+      "step": 2705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7904994820542976e-05,
+      "loss": 1.3657,
+      "step": 2706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7903483246401313e-05,
+      "loss": 0.3175,
+      "step": 2707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7901971168116405e-05,
+      "loss": 0.1972,
+      "step": 2708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7900458585747335e-05,
+      "loss": 0.9716,
+      "step": 2709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7898945499353195e-05,
+      "loss": 0.4037,
+      "step": 2710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.78974319089931e-05,
+      "loss": 0.8989,
+      "step": 2711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7895917814726194e-05,
+      "loss": 0.7758,
+      "step": 2712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7894403216611627e-05,
+      "loss": 0.7161,
+      "step": 2713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.789288811470857e-05,
+      "loss": 1.0061,
+      "step": 2714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7891372509076233e-05,
+      "loss": 0.5378,
+      "step": 2715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7889856399773817e-05,
+      "loss": 0.8718,
+      "step": 2716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7888339786860562e-05,
+      "loss": 1.3486,
+      "step": 2717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7886822670395728e-05,
+      "loss": 1.4621,
+      "step": 2718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7885305050438582e-05,
+      "loss": 0.3989,
+      "step": 2719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7883786927048417e-05,
+      "loss": 0.4811,
+      "step": 2720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7882268300284555e-05,
+      "loss": 1.7945,
+      "step": 2721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7880749170206327e-05,
+      "loss": 0.9516,
+      "step": 2722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7879229536873075e-05,
+      "loss": 0.2561,
+      "step": 2723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.787770940034419e-05,
+      "loss": 1.7368,
+      "step": 2724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7876188760679046e-05,
+      "loss": 0.6457,
+      "step": 2725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7874667617937066e-05,
+      "loss": 0.3136,
+      "step": 2726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7873145972177673e-05,
+      "loss": 0.6807,
+      "step": 2727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.787162382346033e-05,
+      "loss": 1.6656,
+      "step": 2728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7870101171844498e-05,
+      "loss": 1.3017,
+      "step": 2729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7868578017389674e-05,
+      "loss": 1.7709,
+      "step": 2730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.786705436015536e-05,
+      "loss": 0.2209,
+      "step": 2731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7865530200201097e-05,
+      "loss": 1.2868,
+      "step": 2732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.786400553758642e-05,
+      "loss": 1.2386,
+      "step": 2733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7862480372370915e-05,
+      "loss": 0.6157,
+      "step": 2734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7860954704614156e-05,
+      "loss": 1.1359,
+      "step": 2735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7859428534375762e-05,
+      "loss": 1.0287,
+      "step": 2736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.785790186171535e-05,
+      "loss": 0.7281,
+      "step": 2737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.785637468669258e-05,
+      "loss": 0.0747,
+      "step": 2738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.785484700936711e-05,
+      "loss": 0.75,
+      "step": 2739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.785331882979863e-05,
+      "loss": 0.8961,
+      "step": 2740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7851790148046847e-05,
+      "loss": 0.4915,
+      "step": 2741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7850260964171482e-05,
+      "loss": 0.998,
+      "step": 2742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7848731278232293e-05,
+      "loss": 1.0158,
+      "step": 2743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7847201090289033e-05,
+      "loss": 0.6337,
+      "step": 2744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.784567040040149e-05,
+      "loss": 0.2313,
+      "step": 2745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7844139208629477e-05,
+      "loss": 0.4795,
+      "step": 2746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7842607515032806e-05,
+      "loss": 0.6406,
+      "step": 2747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7841075319671324e-05,
+      "loss": 0.8807,
+      "step": 2748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7839542622604898e-05,
+      "loss": 0.6184,
+      "step": 2749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.783800942389341e-05,
+      "loss": 0.3352,
+      "step": 2750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7836475723596762e-05,
+      "loss": 1.4159,
+      "step": 2751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7834941521774873e-05,
+      "loss": 1.079,
+      "step": 2752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.783340681848769e-05,
+      "loss": 0.3696,
+      "step": 2753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7831871613795172e-05,
+      "loss": 0.5292,
+      "step": 2754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7830335907757304e-05,
+      "loss": 1.0336,
+      "step": 2755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7828799700434073e-05,
+      "loss": 1.2384,
+      "step": 2756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7827262991885515e-05,
+      "loss": 0.4537,
+      "step": 2757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7825725782171662e-05,
+      "loss": 0.4631,
+      "step": 2758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.782418807135258e-05,
+      "loss": 0.2873,
+      "step": 2759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7822649859488337e-05,
+      "loss": 0.7362,
+      "step": 2760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7821111146639032e-05,
+      "loss": 1.5247,
+      "step": 2761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7819571932864796e-05,
+      "loss": 1.3586,
+      "step": 2762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7818032218225755e-05,
+      "loss": 1.8664,
+      "step": 2763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.781649200278207e-05,
+      "loss": 1.6245,
+      "step": 2764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.781495128659392e-05,
+      "loss": 0.7672,
+      "step": 2765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7813410069721495e-05,
+      "loss": 1.2534,
+      "step": 2766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7811868352225017e-05,
+      "loss": 1.8889,
+      "step": 2767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7810326134164718e-05,
+      "loss": 0.3715,
+      "step": 2768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7808783415600853e-05,
+      "loss": 1.2978,
+      "step": 2769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7807240196593697e-05,
+      "loss": 1.1745,
+      "step": 2770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7805696477203546e-05,
+      "loss": 0.8171,
+      "step": 2771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.780415225749071e-05,
+      "loss": 1.4511,
+      "step": 2772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7802607537515524e-05,
+      "loss": 1.7506,
+      "step": 2773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7801062317338346e-05,
+      "loss": 1.0863,
+      "step": 2774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.779951659701954e-05,
+      "loss": 1.7965,
+      "step": 2775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7797970376619497e-05,
+      "loss": 0.2648,
+      "step": 2776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.779642365619864e-05,
+      "loss": 1.5216,
+      "step": 2777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7794876435817386e-05,
+      "loss": 0.9592,
+      "step": 2778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7793328715536195e-05,
+      "loss": 0.8121,
+      "step": 2779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.779178049541553e-05,
+      "loss": 0.3982,
+      "step": 2780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7790231775515884e-05,
+      "loss": 1.0997,
+      "step": 2781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7788682555897767e-05,
+      "loss": 1.316,
+      "step": 2782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7787132836621707e-05,
+      "loss": 1.4046,
+      "step": 2783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.778558261774825e-05,
+      "loss": 1.1286,
+      "step": 2784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.778403189933797e-05,
+      "loss": 1.1666,
+      "step": 2785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7782480681451446e-05,
+      "loss": 0.6712,
+      "step": 2786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7780928964149283e-05,
+      "loss": 0.5602,
+      "step": 2787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7779376747492117e-05,
+      "loss": 0.7135,
+      "step": 2788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.777782403154059e-05,
+      "loss": 0.5074,
+      "step": 2789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.777627081635536e-05,
+      "loss": 0.7741,
+      "step": 2790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.777471710199712e-05,
+      "loss": 0.2465,
+      "step": 2791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.777316288852657e-05,
+      "loss": 0.6535,
+      "step": 2792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7771608176004433e-05,
+      "loss": 0.6955,
+      "step": 2793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.777005296449146e-05,
+      "loss": 0.858,
+      "step": 2794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7768497254048402e-05,
+      "loss": 0.8393,
+      "step": 2795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.776694104473605e-05,
+      "loss": 0.6836,
+      "step": 2796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7765384336615196e-05,
+      "loss": 0.5613,
+      "step": 2797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7763827129746674e-05,
+      "loss": 0.4322,
+      "step": 2798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7762269424191317e-05,
+      "loss": 0.5921,
+      "step": 2799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7760711220009984e-05,
+      "loss": 0.3924,
+      "step": 2800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7759152517263555e-05,
+      "loss": 0.4441,
+      "step": 2801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7757593316012933e-05,
+      "loss": 1.4903,
+      "step": 2802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7756033616319035e-05,
+      "loss": 0.5704,
+      "step": 2803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7754473418242795e-05,
+      "loss": 0.3132,
+      "step": 2804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7752912721845177e-05,
+      "loss": 0.4356,
+      "step": 2805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7751351527187153e-05,
+      "loss": 1.0993,
+      "step": 2806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.774978983432972e-05,
+      "loss": 1.1898,
+      "step": 2807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7748227643333894e-05,
+      "loss": 1.1257,
+      "step": 2808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.774666495426071e-05,
+      "loss": 1.2183,
+      "step": 2809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7745101767171228e-05,
+      "loss": 1.1445,
+      "step": 2810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7743538082126512e-05,
+      "loss": 0.8997,
+      "step": 2811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7741973899187665e-05,
+      "loss": 0.4514,
+      "step": 2812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7740409218415798e-05,
+      "loss": 0.9855,
+      "step": 2813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7738844039872036e-05,
+      "loss": 0.0885,
+      "step": 2814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7737278363617542e-05,
+      "loss": 1.7519,
+      "step": 2815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7735712189713485e-05,
+      "loss": 0.4069,
+      "step": 2816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.773414551822105e-05,
+      "loss": 0.8787,
+      "step": 2817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.773257834920145e-05,
+      "loss": 0.4858,
+      "step": 2818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7731010682715918e-05,
+      "loss": 0.5619,
+      "step": 2819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.77294425188257e-05,
+      "loss": 0.2182,
+      "step": 2820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.772787385759207e-05,
+      "loss": 1.3261,
+      "step": 2821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.772630469907631e-05,
+      "loss": 1.3203,
+      "step": 2822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7724735043339723e-05,
+      "loss": 1.2474,
+      "step": 2823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7723164890443647e-05,
+      "loss": 0.3753,
+      "step": 2824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7721594240449427e-05,
+      "loss": 0.9802,
+      "step": 2825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7720023093418415e-05,
+      "loss": 0.4674,
+      "step": 2826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7718451449412017e-05,
+      "loss": 1.3513,
+      "step": 2827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7716879308491623e-05,
+      "loss": 1.1117,
+      "step": 2828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7715306670718664e-05,
+      "loss": 1.0235,
+      "step": 2829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.771373353615458e-05,
+      "loss": 1.0067,
+      "step": 2830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7712159904860832e-05,
+      "loss": 0.7533,
+      "step": 2831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7710585776898908e-05,
+      "loss": 0.6667,
+      "step": 2832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7709011152330306e-05,
+      "loss": 0.624,
+      "step": 2833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7707436031216546e-05,
+      "loss": 1.0206,
+      "step": 2834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7705860413619174e-05,
+      "loss": 1.4426,
+      "step": 2835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7704284299599742e-05,
+      "loss": 0.4994,
+      "step": 2836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7702707689219834e-05,
+      "loss": 0.7222,
+      "step": 2837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7701130582541048e-05,
+      "loss": 1.0379,
+      "step": 2838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7699552979625005e-05,
+      "loss": 0.8786,
+      "step": 2839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7697974880533334e-05,
+      "loss": 1.1926,
+      "step": 2840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7696396285327704e-05,
+      "loss": 0.726,
+      "step": 2841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.769481719406978e-05,
+      "loss": 0.6561,
+      "step": 2842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7693237606821267e-05,
+      "loss": 0.6188,
+      "step": 2843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7691657523643873e-05,
+      "loss": 1.6841,
+      "step": 2844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7690076944599334e-05,
+      "loss": 0.6863,
+      "step": 2845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7688495869749402e-05,
+      "loss": 0.4993,
+      "step": 2846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7686914299155858e-05,
+      "loss": 0.632,
+      "step": 2847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7685332232880486e-05,
+      "loss": 0.7783,
+      "step": 2848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.76837496709851e-05,
+      "loss": 0.4864,
+      "step": 2849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7682166613531535e-05,
+      "loss": 0.5278,
+      "step": 2850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7680583060581633e-05,
+      "loss": 1.2914,
+      "step": 2851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7678999012197272e-05,
+      "loss": 0.9198,
+      "step": 2852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7677414468440342e-05,
+      "loss": 0.7022,
+      "step": 2853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7675829429372744e-05,
+      "loss": 0.2343,
+      "step": 2854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7674243895056416e-05,
+      "loss": 1.7862,
+      "step": 2855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7672657865553295e-05,
+      "loss": 1.4999,
+      "step": 2856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.767107134092535e-05,
+      "loss": 1.2007,
+      "step": 2857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7669484321234575e-05,
+      "loss": 0.3697,
+      "step": 2858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7667896806542965e-05,
+      "loss": 0.5423,
+      "step": 2859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7666308796912552e-05,
+      "loss": 1.6996,
+      "step": 2860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.766472029240538e-05,
+      "loss": 0.3726,
+      "step": 2861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7663131293083504e-05,
+      "loss": 0.3201,
+      "step": 2862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7661541799009017e-05,
+      "loss": 1.961,
+      "step": 2863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7659951810244015e-05,
+      "loss": 0.6307,
+      "step": 2864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7658361326850627e-05,
+      "loss": 0.4331,
+      "step": 2865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.765677034889098e-05,
+      "loss": 0.1572,
+      "step": 2866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7655178876427243e-05,
+      "loss": 0.4516,
+      "step": 2867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7653586909521594e-05,
+      "loss": 0.2809,
+      "step": 2868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7651994448236235e-05,
+      "loss": 0.2152,
+      "step": 2869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7650401492633378e-05,
+      "loss": 0.6143,
+      "step": 2870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7648808042775263e-05,
+      "loss": 1.092,
+      "step": 2871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.764721409872415e-05,
+      "loss": 1.1085,
+      "step": 2872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7645619660542307e-05,
+      "loss": 1.8961,
+      "step": 2873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7644024728292034e-05,
+      "loss": 0.9602,
+      "step": 2874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7642429302035646e-05,
+      "loss": 1.2722,
+      "step": 2875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.764083338183548e-05,
+      "loss": 0.6625,
+      "step": 2876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7639236967753885e-05,
+      "loss": 1.2781,
+      "step": 2877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.763764005985323e-05,
+      "loss": 0.8998,
+      "step": 2878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.763604265819591e-05,
+      "loss": 0.2915,
+      "step": 2879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7634444762844338e-05,
+      "loss": 1.577,
+      "step": 2880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7632846373860946e-05,
+      "loss": 0.5055,
+      "step": 2881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7631247491308172e-05,
+      "loss": 0.742,
+      "step": 2882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7629648115248497e-05,
+      "loss": 0.4027,
+      "step": 2883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.762804824574441e-05,
+      "loss": 1.2207,
+      "step": 2884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7626447882858408e-05,
+      "loss": 0.695,
+      "step": 2885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7624847026653022e-05,
+      "loss": 0.5347,
+      "step": 2886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7623245677190803e-05,
+      "loss": 0.4142,
+      "step": 2887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.762164383453431e-05,
+      "loss": 0.6903,
+      "step": 2888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7620041498746134e-05,
+      "loss": 0.7202,
+      "step": 2889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.761843866988887e-05,
+      "loss": 1.2524,
+      "step": 2890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.761683534802515e-05,
+      "loss": 1.2697,
+      "step": 2891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.761523153321761e-05,
+      "loss": 0.8404,
+      "step": 2892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7613627225528912e-05,
+      "loss": 1.2214,
+      "step": 2893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.761202242502174e-05,
+      "loss": 1.4127,
+      "step": 2894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7610417131758793e-05,
+      "loss": 1.2025,
+      "step": 2895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7608811345802785e-05,
+      "loss": 0.5084,
+      "step": 2896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7607205067216468e-05,
+      "loss": 0.7656,
+      "step": 2897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7605598296062586e-05,
+      "loss": 0.3393,
+      "step": 2898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7603991032403927e-05,
+      "loss": 0.9296,
+      "step": 2899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7602383276303274e-05,
+      "loss": 0.374,
+      "step": 2900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7600775027823453e-05,
+      "loss": 1.5398,
+      "step": 2901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7599166287027298e-05,
+      "loss": 1.3224,
+      "step": 2902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7597557053977656e-05,
+      "loss": 1.2963,
+      "step": 2903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.759594732873741e-05,
+      "loss": 0.4387,
+      "step": 2904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7594337111369446e-05,
+      "loss": 0.2009,
+      "step": 2905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7592726401936682e-05,
+      "loss": 0.2606,
+      "step": 2906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7591115200502038e-05,
+      "loss": 1.6085,
+      "step": 2907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.758950350712847e-05,
+      "loss": 0.4777,
+      "step": 2908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.758789132187895e-05,
+      "loss": 0.2877,
+      "step": 2909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7586278644816466e-05,
+      "loss": 0.6364,
+      "step": 2910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7584665476004027e-05,
+      "loss": 0.5121,
+      "step": 2911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7583051815504653e-05,
+      "loss": 2.0424,
+      "step": 2912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7581437663381396e-05,
+      "loss": 0.3715,
+      "step": 2913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.757982301969732e-05,
+      "loss": 0.769,
+      "step": 2914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.757820788451551e-05,
+      "loss": 0.1684,
+      "step": 2915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7576592257899067e-05,
+      "loss": 0.414,
+      "step": 2916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7574976139911117e-05,
+      "loss": 0.4369,
+      "step": 2917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.757335953061481e-05,
+      "loss": 0.8109,
+      "step": 2918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.757174243007329e-05,
+      "loss": 0.2119,
+      "step": 2919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7570124838349746e-05,
+      "loss": 1.4707,
+      "step": 2920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7568506755507387e-05,
+      "loss": 0.7194,
+      "step": 2921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7566888181609416e-05,
+      "loss": 1.0815,
+      "step": 2922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.756526911671908e-05,
+      "loss": 1.0795,
+      "step": 2923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7563649560899636e-05,
+      "loss": 1.1142,
+      "step": 2924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7562029514214364e-05,
+      "loss": 1.7316,
+      "step": 2925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7560408976726556e-05,
+      "loss": 0.6831,
+      "step": 2926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.755878794849952e-05,
+      "loss": 0.9896,
+      "step": 2927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.75571664295966e-05,
+      "loss": 0.3433,
+      "step": 2928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7555544420081146e-05,
+      "loss": 1.5052,
+      "step": 2929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7553921920016533e-05,
+      "loss": 0.7203,
+      "step": 2930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7552298929466145e-05,
+      "loss": 0.5722,
+      "step": 2931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.75506754484934e-05,
+      "loss": 1.7112,
+      "step": 2932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7549051477161726e-05,
+      "loss": 0.5484,
+      "step": 2933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7547427015534575e-05,
+      "loss": 0.9826,
+      "step": 2934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7545802063675406e-05,
+      "loss": 1.265,
+      "step": 2935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7544176621647716e-05,
+      "loss": 1.1046,
+      "step": 2936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7542550689515007e-05,
+      "loss": 0.6638,
+      "step": 2937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7540924267340808e-05,
+      "loss": 0.7868,
+      "step": 2938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7539297355188658e-05,
+      "loss": 0.8243,
+      "step": 2939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7537669953122124e-05,
+      "loss": 1.5498,
+      "step": 2940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.753604206120479e-05,
+      "loss": 1.0855,
+      "step": 2941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.753441367950026e-05,
+      "loss": 0.5518,
+      "step": 2942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7532784808072157e-05,
+      "loss": 1.4112,
+      "step": 2943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.753115544698411e-05,
+      "loss": 0.7539,
+      "step": 2944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.752952559629979e-05,
+      "loss": 0.4714,
+      "step": 2945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.752789525608287e-05,
+      "loss": 0.5484,
+      "step": 2946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7526264426397052e-05,
+      "loss": 1.4872,
+      "step": 2947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.752463310730605e-05,
+      "loss": 0.4213,
+      "step": 2948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7523001298873604e-05,
+      "loss": 0.6586,
+      "step": 2949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7521369001163466e-05,
+      "loss": 1.1825,
+      "step": 2950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7519736214239405e-05,
+      "loss": 1.6558,
+      "step": 2951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7518102938165228e-05,
+      "loss": 0.7478,
+      "step": 2952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7516469173004736e-05,
+      "loss": 0.2785,
+      "step": 2953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7514834918821767e-05,
+      "loss": 1.155,
+      "step": 2954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.751320017568017e-05,
+      "loss": 1.0092,
+      "step": 2955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7511564943643815e-05,
+      "loss": 1.2541,
+      "step": 2956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7509929222776585e-05,
+      "loss": 0.2431,
+      "step": 2957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.75082930131424e-05,
+      "loss": 0.648,
+      "step": 2958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7506656314805177e-05,
+      "loss": 0.257,
+      "step": 2959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7505019127828873e-05,
+      "loss": 1.0709,
+      "step": 2960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.750338145227744e-05,
+      "loss": 1.5561,
+      "step": 2961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7501743288214872e-05,
+      "loss": 0.5644,
+      "step": 2962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.750010463570517e-05,
+      "loss": 1.016,
+      "step": 2963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.749846549481236e-05,
+      "loss": 0.1999,
+      "step": 2964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7496825865600478e-05,
+      "loss": 1.0338,
+      "step": 2965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7495185748133583e-05,
+      "loss": 0.2192,
+      "step": 2966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.749354514247576e-05,
+      "loss": 1.5829,
+      "step": 2967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.749190404869111e-05,
+      "loss": 0.1421,
+      "step": 2968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.749026246684375e-05,
+      "loss": 0.975,
+      "step": 2969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.748862039699781e-05,
+      "loss": 0.8282,
+      "step": 2970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7486977839217456e-05,
+      "loss": 0.5829,
+      "step": 2971
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7485334793566854e-05,
+      "loss": 1.5218,
+      "step": 2972
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.748369126011021e-05,
+      "loss": 0.7885,
+      "step": 2973
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.748204723891173e-05,
+      "loss": 1.7925,
+      "step": 2974
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7480402730035638e-05,
+      "loss": 0.9541,
+      "step": 2975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.74787577335462e-05,
+      "loss": 0.6416,
+      "step": 2976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.747711224950768e-05,
+      "loss": 0.8174,
+      "step": 2977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.747546627798437e-05,
+      "loss": 0.7706,
+      "step": 2978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7473819819040575e-05,
+      "loss": 1.17,
+      "step": 2979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.747217287274063e-05,
+      "loss": 1.0068,
+      "step": 2980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.747052543914887e-05,
+      "loss": 1.1618,
+      "step": 2981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7468877518329673e-05,
+      "loss": 1.317,
+      "step": 2982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7467229110347413e-05,
+      "loss": 1.1085,
+      "step": 2983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7465580215266504e-05,
+      "loss": 0.4751,
+      "step": 2984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.746393083315136e-05,
+      "loss": 0.8876,
+      "step": 2985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7462280964066427e-05,
+      "loss": 0.294,
+      "step": 2986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.746063060807617e-05,
+      "loss": 0.7488,
+      "step": 2987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.745897976524506e-05,
+      "loss": 0.4121,
+      "step": 2988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7457328435637602e-05,
+      "loss": 0.5834,
+      "step": 2989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7455676619318307e-05,
+      "loss": 0.3407,
+      "step": 2990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7454024316351726e-05,
+      "loss": 0.5505,
+      "step": 2991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7452371526802403e-05,
+      "loss": 1.8711,
+      "step": 2992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7450718250734916e-05,
+      "loss": 1.1418,
+      "step": 2993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7449064488213863e-05,
+      "loss": 0.9069,
+      "step": 2994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7447410239303848e-05,
+      "loss": 1.7943,
+      "step": 2995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7445755504069515e-05,
+      "loss": 1.1616,
+      "step": 2996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7444100282575508e-05,
+      "loss": 1.4639,
+      "step": 2997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.744244457488649e-05,
+      "loss": 0.6342,
+      "step": 2998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7440788381067167e-05,
+      "loss": 1.201,
+      "step": 2999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.743913170118223e-05,
+      "loss": 0.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7437474535296418e-05,
+      "loss": 1.5628,
+      "step": 3001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.743581688347447e-05,
+      "loss": 0.5382,
+      "step": 3002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7434158745781157e-05,
+      "loss": 0.618,
+      "step": 3003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.743250012228126e-05,
+      "loss": 1.6211,
+      "step": 3004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.743084101303958e-05,
+      "loss": 0.9179,
+      "step": 3005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.742918141812094e-05,
+      "loss": 1.1277,
+      "step": 3006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7427521337590184e-05,
+      "loss": 0.2039,
+      "step": 3007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7425860771512166e-05,
+      "loss": 1.8096,
+      "step": 3008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7424199719951772e-05,
+      "loss": 0.6055,
+      "step": 3009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.742253818297389e-05,
+      "loss": 0.2978,
+      "step": 3010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.742087616064345e-05,
+      "loss": 1.7886,
+      "step": 3011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7419213653025376e-05,
+      "loss": 0.6696,
+      "step": 3012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7417550660184628e-05,
+      "loss": 0.2368,
+      "step": 3013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7415887182186183e-05,
+      "loss": 0.6871,
+      "step": 3014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7414223219095028e-05,
+      "loss": 0.151,
+      "step": 3015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7412558770976176e-05,
+      "loss": 1.6867,
+      "step": 3016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7410893837894654e-05,
+      "loss": 1.0485,
+      "step": 3017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7409228419915518e-05,
+      "loss": 0.2898,
+      "step": 3018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7407562517103834e-05,
+      "loss": 0.4159,
+      "step": 3019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.740589612952469e-05,
+      "loss": 0.4021,
+      "step": 3020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7404229257243188e-05,
+      "loss": 0.4808,
+      "step": 3021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7402561900324456e-05,
+      "loss": 1.6865,
+      "step": 3022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7400894058833642e-05,
+      "loss": 1.4427,
+      "step": 3023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7399225732835903e-05,
+      "loss": 0.8112,
+      "step": 3024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7397556922396426e-05,
+      "loss": 0.2505,
+      "step": 3025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7395887627580413e-05,
+      "loss": 1.6216,
+      "step": 3026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7394217848453075e-05,
+      "loss": 0.7755,
+      "step": 3027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7392547585079658e-05,
+      "loss": 0.8277,
+      "step": 3028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.739087683752542e-05,
+      "loss": 0.5344,
+      "step": 3029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7389205605855632e-05,
+      "loss": 0.2546,
+      "step": 3030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7387533890135596e-05,
+      "loss": 0.3016,
+      "step": 3031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7385861690430624e-05,
+      "loss": 0.7221,
+      "step": 3032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.738418900680605e-05,
+      "loss": 1.4086,
+      "step": 3033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7382515839327223e-05,
+      "loss": 0.8433,
+      "step": 3034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.738084218805952e-05,
+      "loss": 0.6542,
+      "step": 3035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7379168053068325e-05,
+      "loss": 0.7079,
+      "step": 3036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7377493434419052e-05,
+      "loss": 0.8509,
+      "step": 3037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7375818332177125e-05,
+      "loss": 0.6217,
+      "step": 3038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7374142746407992e-05,
+      "loss": 1.2098,
+      "step": 3039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.737246667717712e-05,
+      "loss": 0.2104,
+      "step": 3040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7370790124549993e-05,
+      "loss": 0.6094,
+      "step": 3041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7369113088592115e-05,
+      "loss": 0.8253,
+      "step": 3042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.736743556936901e-05,
+      "loss": 0.8933,
+      "step": 3043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7365757566946213e-05,
+      "loss": 1.4814,
+      "step": 3044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7364079081389285e-05,
+      "loss": 0.1841,
+      "step": 3045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7362400112763813e-05,
+      "loss": 0.6179,
+      "step": 3046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7360720661135387e-05,
+      "loss": 1.021,
+      "step": 3047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.735904072656963e-05,
+      "loss": 0.9715,
+      "step": 3048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.735736030913217e-05,
+      "loss": 0.7116,
+      "step": 3049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7355679408888664e-05,
+      "loss": 0.713,
+      "step": 3050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.735399802590479e-05,
+      "loss": 0.9335,
+      "step": 3051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7352316160246236e-05,
+      "loss": 0.5234,
+      "step": 3052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7350633811978715e-05,
+      "loss": 0.9565,
+      "step": 3053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.734895098116795e-05,
+      "loss": 1.8595,
+      "step": 3054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.73472676678797e-05,
+      "loss": 0.2293,
+      "step": 3055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.734558387217973e-05,
+      "loss": 0.7279,
+      "step": 3056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7343899594133827e-05,
+      "loss": 0.8867,
+      "step": 3057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7342214833807784e-05,
+      "loss": 0.4919,
+      "step": 3058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7340529591267442e-05,
+      "loss": 0.4257,
+      "step": 3059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7338843866578637e-05,
+      "loss": 0.2582,
+      "step": 3060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.733715765980723e-05,
+      "loss": 0.9854,
+      "step": 3061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.73354709710191e-05,
+      "loss": 0.3675,
+      "step": 3062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.733378380028015e-05,
+      "loss": 0.5338,
+      "step": 3063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.73320961476563e-05,
+      "loss": 1.5949,
+      "step": 3064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.733040801321348e-05,
+      "loss": 0.5971,
+      "step": 3065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.732871939701765e-05,
+      "loss": 1.5067,
+      "step": 3066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.732703029913479e-05,
+      "loss": 1.2069,
+      "step": 3067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7325340719630884e-05,
+      "loss": 0.5485,
+      "step": 3068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7323650658571952e-05,
+      "loss": 0.3828,
+      "step": 3069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7321960116024023e-05,
+      "loss": 1.2638,
+      "step": 3070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7320269092053147e-05,
+      "loss": 1.1916,
+      "step": 3071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.731857758672539e-05,
+      "loss": 0.9853,
+      "step": 3072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.731688560010684e-05,
+      "loss": 0.7042,
+      "step": 3073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7315193132263605e-05,
+      "loss": 0.476,
+      "step": 3074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7313500183261814e-05,
+      "loss": 0.4145,
+      "step": 3075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7311806753167607e-05,
+      "loss": 0.7612,
+      "step": 3076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7310112842047146e-05,
+      "loss": 0.6239,
+      "step": 3077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.730841844996662e-05,
+      "loss": 1.0713,
+      "step": 3078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7306723576992213e-05,
+      "loss": 0.4158,
+      "step": 3079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7305028223190163e-05,
+      "loss": 1.1163,
+      "step": 3080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7303332388626693e-05,
+      "loss": 0.8609,
+      "step": 3081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7301636073368074e-05,
+      "loss": 1.5637,
+      "step": 3082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7299939277480573e-05,
+      "loss": 0.3165,
+      "step": 3083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.729824200103048e-05,
+      "loss": 1.389,
+      "step": 3084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.729654424408412e-05,
+      "loss": 1.4951,
+      "step": 3085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7294846006707814e-05,
+      "loss": 0.5491,
+      "step": 3086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7293147288967922e-05,
+      "loss": 0.8908,
+      "step": 3087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7291448090930805e-05,
+      "loss": 0.6932,
+      "step": 3088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7289748412662856e-05,
+      "loss": 0.4572,
+      "step": 3089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.728804825423048e-05,
+      "loss": 0.7259,
+      "step": 3090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7286347615700106e-05,
+      "loss": 1.1273,
+      "step": 3091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7284646497138174e-05,
+      "loss": 1.6712,
+      "step": 3092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7282944898611148e-05,
+      "loss": 1.0015,
+      "step": 3093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7281242820185517e-05,
+      "loss": 1.2669,
+      "step": 3094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7279540261927773e-05,
+      "loss": 0.9748,
+      "step": 3095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7277837223904438e-05,
+      "loss": 0.9098,
+      "step": 3096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.727613370618205e-05,
+      "loss": 0.8382,
+      "step": 3097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.727442970882717e-05,
+      "loss": 0.8174,
+      "step": 3098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.727272523190637e-05,
+      "loss": 1.0972,
+      "step": 3099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7271020275486243e-05,
+      "loss": 0.2051,
+      "step": 3100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7269314839633405e-05,
+      "loss": 1.5859,
+      "step": 3101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7267608924414487e-05,
+      "loss": 0.2664,
+      "step": 3102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.726590252989614e-05,
+      "loss": 1.136,
+      "step": 3103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7264195656145035e-05,
+      "loss": 0.394,
+      "step": 3104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7262488303227853e-05,
+      "loss": 0.3757,
+      "step": 3105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7260780471211313e-05,
+      "loss": 0.3716,
+      "step": 3106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.725907216016213e-05,
+      "loss": 1.6062,
+      "step": 3107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7257363370147055e-05,
+      "loss": 0.7179,
+      "step": 3108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7255654101232846e-05,
+      "loss": 0.5287,
+      "step": 3109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7253944353486287e-05,
+      "loss": 0.766,
+      "step": 3110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.725223412697418e-05,
+      "loss": 0.2874,
+      "step": 3111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7250523421763335e-05,
+      "loss": 0.4149,
+      "step": 3112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7248812237920606e-05,
+      "loss": 1.1665,
+      "step": 3113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7247100575512835e-05,
+      "loss": 0.6783,
+      "step": 3114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7245388434606905e-05,
+      "loss": 1.3292,
+      "step": 3115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7243675815269703e-05,
+      "loss": 0.247,
+      "step": 3116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7241962717568153e-05,
+      "loss": 0.348,
+      "step": 3117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7240249141569175e-05,
+      "loss": 1.5131,
+      "step": 3118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7238535087339725e-05,
+      "loss": 1.5195,
+      "step": 3119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.723682055494677e-05,
+      "loss": 0.5009,
+      "step": 3120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7235105544457298e-05,
+      "loss": 0.9832,
+      "step": 3121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7233390055938315e-05,
+      "loss": 0.3387,
+      "step": 3122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7231674089456838e-05,
+      "loss": 1.3203,
+      "step": 3123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7229957645079927e-05,
+      "loss": 0.3077,
+      "step": 3124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.722824072287463e-05,
+      "loss": 0.3016,
+      "step": 3125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7226523322908035e-05,
+      "loss": 0.955,
+      "step": 3126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.722480544524723e-05,
+      "loss": 1.1719,
+      "step": 3127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7223087089959352e-05,
+      "loss": 1.3176,
+      "step": 3128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7221368257111524e-05,
+      "loss": 1.4156,
+      "step": 3129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.72196489467709e-05,
+      "loss": 0.9623,
+      "step": 3130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7217929159004664e-05,
+      "loss": 0.8993,
+      "step": 3131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7216208893880002e-05,
+      "loss": 0.8662,
+      "step": 3132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7214488151464127e-05,
+      "loss": 0.6731,
+      "step": 3133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7212766931824267e-05,
+      "loss": 0.6951,
+      "step": 3134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7211045235027672e-05,
+      "loss": 1.4765,
+      "step": 3135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7209323061141613e-05,
+      "loss": 0.8327,
+      "step": 3136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7207600410233368e-05,
+      "loss": 1.8243,
+      "step": 3137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.720587728237025e-05,
+      "loss": 0.295,
+      "step": 3138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7204153677619572e-05,
+      "loss": 0.8578,
+      "step": 3139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.720242959604869e-05,
+      "loss": 0.4001,
+      "step": 3140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7200705037724952e-05,
+      "loss": 1.0972,
+      "step": 3141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.719898000271574e-05,
+      "loss": 0.9546,
+      "step": 3142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.719725449108845e-05,
+      "loss": 1.6181,
+      "step": 3143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.719552850291051e-05,
+      "loss": 1.3502,
+      "step": 3144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7193802038249337e-05,
+      "loss": 0.7173,
+      "step": 3145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.71920750971724e-05,
+      "loss": 0.8336,
+      "step": 3146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7190347679747164e-05,
+      "loss": 0.2379,
+      "step": 3147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.718861978604112e-05,
+      "loss": 0.5695,
+      "step": 3148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7186891416121775e-05,
+      "loss": 0.7985,
+      "step": 3149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.718516257005666e-05,
+      "loss": 0.5511,
+      "step": 3150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7183433247913325e-05,
+      "loss": 0.7571,
+      "step": 3151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7181703449759332e-05,
+      "loss": 0.5866,
+      "step": 3152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.717997317566226e-05,
+      "loss": 0.4957,
+      "step": 3153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.717824242568972e-05,
+      "loss": 1.61,
+      "step": 3154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.717651119990933e-05,
+      "loss": 1.3667,
+      "step": 3155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7174779498388723e-05,
+      "loss": 1.4905,
+      "step": 3156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7173047321195565e-05,
+      "loss": 0.2164,
+      "step": 3157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.717131466839753e-05,
+      "loss": 1.3548,
+      "step": 3158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.716958154006231e-05,
+      "loss": 0.2681,
+      "step": 3159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.716784793625763e-05,
+      "loss": 0.2954,
+      "step": 3160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.716611385705121e-05,
+      "loss": 1.2369,
+      "step": 3161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7164379302510807e-05,
+      "loss": 0.5836,
+      "step": 3162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7162644272704185e-05,
+      "loss": 0.6459,
+      "step": 3163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7160908767699145e-05,
+      "loss": 0.3252,
+      "step": 3164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.715917278756348e-05,
+      "loss": 1.0549,
+      "step": 3165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.715743633236502e-05,
+      "loss": 0.5717,
+      "step": 3166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.715569940217161e-05,
+      "loss": 0.3107,
+      "step": 3167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7153961997051116e-05,
+      "loss": 1.0223,
+      "step": 3168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.715222411707141e-05,
+      "loss": 1.6723,
+      "step": 3169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.71504857623004e-05,
+      "loss": 0.845,
+      "step": 3170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7148746932806e-05,
+      "loss": 1.8594,
+      "step": 3171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7147007628656146e-05,
+      "loss": 0.7846,
+      "step": 3172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7145267849918794e-05,
+      "loss": 0.3978,
+      "step": 3173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7143527596661916e-05,
+      "loss": 1.1475,
+      "step": 3174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.714178686895351e-05,
+      "loss": 1.6269,
+      "step": 3175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7140045666861583e-05,
+      "loss": 0.2188,
+      "step": 3176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7138303990454164e-05,
+      "loss": 0.6578,
+      "step": 3177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.71365618397993e-05,
+      "loss": 0.3318,
+      "step": 3178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7134819214965058e-05,
+      "loss": 0.4964,
+      "step": 3179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7133076116019523e-05,
+      "loss": 0.7377,
+      "step": 3180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7131332543030803e-05,
+      "loss": 0.7717,
+      "step": 3181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7129588496067016e-05,
+      "loss": 0.5136,
+      "step": 3182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.71278439751963e-05,
+      "loss": 0.8282,
+      "step": 3183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7126098980486816e-05,
+      "loss": 0.652,
+      "step": 3184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.712435351200674e-05,
+      "loss": 1.0774,
+      "step": 3185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7122607569824274e-05,
+      "loss": 1.0537,
+      "step": 3186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.712086115400763e-05,
+      "loss": 0.4811,
+      "step": 3187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7119114264625033e-05,
+      "loss": 0.2778,
+      "step": 3188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7117366901744746e-05,
+      "loss": 0.1812,
+      "step": 3189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.711561906543503e-05,
+      "loss": 0.6416,
+      "step": 3190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7113870755764177e-05,
+      "loss": 0.2535,
+      "step": 3191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7112121972800503e-05,
+      "loss": 1.2551,
+      "step": 3192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.711037271661232e-05,
+      "loss": 1.2019,
+      "step": 3193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7108622987267973e-05,
+      "loss": 0.7858,
+      "step": 3194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7106872784835833e-05,
+      "loss": 0.3229,
+      "step": 3195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7105122109384276e-05,
+      "loss": 1.0838,
+      "step": 3196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.71033709609817e-05,
+      "loss": 1.0445,
+      "step": 3197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.710161933969653e-05,
+      "loss": 0.6177,
+      "step": 3198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7099867245597197e-05,
+      "loss": 0.6358,
+      "step": 3199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7098114678752155e-05,
+      "loss": 0.5261,
+      "step": 3200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7096361639229877e-05,
+      "loss": 1.9297,
+      "step": 3201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.709460812709885e-05,
+      "loss": 0.4192,
+      "step": 3202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.70928541424276e-05,
+      "loss": 1.7751,
+      "step": 3203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.709109968528465e-05,
+      "loss": 1.3746,
+      "step": 3204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7089344755738532e-05,
+      "loss": 0.4124,
+      "step": 3205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.708758935385783e-05,
+      "loss": 0.2133,
+      "step": 3206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.708583347971112e-05,
+      "loss": 0.424,
+      "step": 3207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.708407713336701e-05,
+      "loss": 0.5454,
+      "step": 3208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.708232031489411e-05,
+      "loss": 1.3841,
+      "step": 3209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.708056302436107e-05,
+      "loss": 1.2345,
+      "step": 3210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7078805261836545e-05,
+      "loss": 1.0756,
+      "step": 3211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7077047027389214e-05,
+      "loss": 0.3538,
+      "step": 3212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.707528832108776e-05,
+      "loss": 0.2894,
+      "step": 3213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7073529143000914e-05,
+      "loss": 0.3361,
+      "step": 3214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.707176949319739e-05,
+      "loss": 0.8566,
+      "step": 3215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.707000937174595e-05,
+      "loss": 1.0822,
+      "step": 3216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.706824877871536e-05,
+      "loss": 0.6905,
+      "step": 3217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7066487714174403e-05,
+      "loss": 0.663,
+      "step": 3218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.706472617819189e-05,
+      "loss": 1.3,
+      "step": 3219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7062964170836642e-05,
+      "loss": 0.9025,
+      "step": 3220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.70612016921775e-05,
+      "loss": 1.0929,
+      "step": 3221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7059438742283328e-05,
+      "loss": 0.324,
+      "step": 3222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7057675321223e-05,
+      "loss": 1.4222,
+      "step": 3223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.705591142906542e-05,
+      "loss": 1.04,
+      "step": 3224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.70541470658795e-05,
+      "loss": 0.0979,
+      "step": 3225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.705238223173417e-05,
+      "loss": 0.4283,
+      "step": 3226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.705061692669839e-05,
+      "loss": 1.1447,
+      "step": 3227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.704885115084113e-05,
+      "loss": 1.5919,
+      "step": 3228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.704708490423137e-05,
+      "loss": 0.7834,
+      "step": 3229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.704531818693813e-05,
+      "loss": 1.0979,
+      "step": 3230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7043550999030434e-05,
+      "loss": 0.3949,
+      "step": 3231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7041783340577324e-05,
+      "loss": 0.303,
+      "step": 3232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.704001521164786e-05,
+      "loss": 0.2983,
+      "step": 3233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7038246612311126e-05,
+      "loss": 0.2876,
+      "step": 3234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7036477542636226e-05,
+      "loss": 0.5978,
+      "step": 3235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7034708002692273e-05,
+      "loss": 0.3432,
+      "step": 3236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.70329379925484e-05,
+      "loss": 0.9942,
+      "step": 3237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7031167512273772e-05,
+      "loss": 0.7478,
+      "step": 3238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7029396561937556e-05,
+      "loss": 1.4912,
+      "step": 3239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7027625141608947e-05,
+      "loss": 0.1006,
+      "step": 3240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7025853251357152e-05,
+      "loss": 1.4909,
+      "step": 3241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7024080891251394e-05,
+      "loss": 0.125,
+      "step": 3242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.702230806136093e-05,
+      "loss": 1.3989,
+      "step": 3243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.702053476175502e-05,
+      "loss": 1.341,
+      "step": 3244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7018760992502948e-05,
+      "loss": 0.7533,
+      "step": 3245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7016986753674013e-05,
+      "loss": 0.2954,
+      "step": 3246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.701521204533754e-05,
+      "loss": 0.8085,
+      "step": 3247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7013436867562867e-05,
+      "loss": 0.6689,
+      "step": 3248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.701166122041935e-05,
+      "loss": 0.63,
+      "step": 3249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.700988510397636e-05,
+      "loss": 0.6681,
+      "step": 3250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7008108518303294e-05,
+      "loss": 1.6562,
+      "step": 3251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.700633146346956e-05,
+      "loss": 1.3138,
+      "step": 3252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.70045539395446e-05,
+      "loss": 1.401,
+      "step": 3253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.700277594659784e-05,
+      "loss": 1.7606,
+      "step": 3254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.700099748469877e-05,
+      "loss": 1.0625,
+      "step": 3255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.699921855391687e-05,
+      "loss": 1.1262,
+      "step": 3256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6997439154321634e-05,
+      "loss": 0.1527,
+      "step": 3257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6995659285982586e-05,
+      "loss": 0.7329,
+      "step": 3258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.699387894896927e-05,
+      "loss": 0.6057,
+      "step": 3259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6992098143351246e-05,
+      "loss": 1.5865,
+      "step": 3260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6990316869198087e-05,
+      "loss": 0.4473,
+      "step": 3261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6988535126579387e-05,
+      "loss": 0.8227,
+      "step": 3262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.698675291556476e-05,
+      "loss": 0.6901,
+      "step": 3263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6984970236223843e-05,
+      "loss": 0.7904,
+      "step": 3264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6983187088626278e-05,
+      "loss": 0.9882,
+      "step": 3265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6981403472841737e-05,
+      "loss": 0.7653,
+      "step": 3266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6979619388939905e-05,
+      "loss": 0.5881,
+      "step": 3267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6977834836990488e-05,
+      "loss": 0.277,
+      "step": 3268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.697604981706321e-05,
+      "loss": 0.9082,
+      "step": 3269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.697426432922781e-05,
+      "loss": 1.2231,
+      "step": 3270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6972478373554052e-05,
+      "loss": 0.3076,
+      "step": 3271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6970691950111705e-05,
+      "loss": 0.6461,
+      "step": 3272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6968905058970575e-05,
+      "loss": 1.168,
+      "step": 3273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6967117700200468e-05,
+      "loss": 0.9109,
+      "step": 3274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6965329873871227e-05,
+      "loss": 0.8072,
+      "step": 3275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6963541580052695e-05,
+      "loss": 0.9346,
+      "step": 3276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6961752818814734e-05,
+      "loss": 0.7394,
+      "step": 3277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.695996359022725e-05,
+      "loss": 0.7949,
+      "step": 3278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.695817389436014e-05,
+      "loss": 0.4185,
+      "step": 3279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.695638373128332e-05,
+      "loss": 0.9229,
+      "step": 3280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6954593101066743e-05,
+      "loss": 0.527,
+      "step": 3281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6952802003780365e-05,
+      "loss": 0.788,
+      "step": 3282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6951010439494165e-05,
+      "loss": 1.0528,
+      "step": 3283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6949218408278143e-05,
+      "loss": 0.4659,
+      "step": 3284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.694742591020231e-05,
+      "loss": 1.2662,
+      "step": 3285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.69456329453367e-05,
+      "loss": 0.8702,
+      "step": 3286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.694383951375137e-05,
+      "loss": 1.3987,
+      "step": 3287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6942045615516383e-05,
+      "loss": 0.1725,
+      "step": 3288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6940251250701833e-05,
+      "loss": 0.3259,
+      "step": 3289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6938456419377816e-05,
+      "loss": 0.2468,
+      "step": 3290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6936661121614472e-05,
+      "loss": 0.8702,
+      "step": 3291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6934865357481937e-05,
+      "loss": 1.1876,
+      "step": 3292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6933069127050365e-05,
+      "loss": 0.5073,
+      "step": 3293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6931272430389947e-05,
+      "loss": 0.7683,
+      "step": 3294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6929475267570873e-05,
+      "loss": 0.9037,
+      "step": 3295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6927677638663357e-05,
+      "loss": 0.8892,
+      "step": 3296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.692587954373764e-05,
+      "loss": 0.5317,
+      "step": 3297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.692408098286397e-05,
+      "loss": 0.7258,
+      "step": 3298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6922281956112616e-05,
+      "loss": 0.2522,
+      "step": 3299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6920482463553876e-05,
+      "loss": 0.6127,
+      "step": 3300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6918682505258044e-05,
+      "loss": 1.1106,
+      "step": 3301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6916882081295455e-05,
+      "loss": 0.5413,
+      "step": 3302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.691508119173644e-05,
+      "loss": 0.2409,
+      "step": 3303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.691327983665137e-05,
+      "loss": 1.5185,
+      "step": 3304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6911478016110622e-05,
+      "loss": 0.8939,
+      "step": 3305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.69096757301846e-05,
+      "loss": 0.8913,
+      "step": 3306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.690787297894371e-05,
+      "loss": 1.449,
+      "step": 3307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6906069762458386e-05,
+      "loss": 1.7752,
+      "step": 3308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.690426608079909e-05,
+      "loss": 0.7564,
+      "step": 3309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6902461934036284e-05,
+      "loss": 1.0467,
+      "step": 3310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.690065732224046e-05,
+      "loss": 1.6937,
+      "step": 3311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.689885224548212e-05,
+      "loss": 0.5055,
+      "step": 3312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.68970467038318e-05,
+      "loss": 0.5526,
+      "step": 3313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6895240697360036e-05,
+      "loss": 0.3568,
+      "step": 3314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6893434226137386e-05,
+      "loss": 1.7348,
+      "step": 3315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6891627290234432e-05,
+      "loss": 0.4143,
+      "step": 3316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6889819889721775e-05,
+      "loss": 1.1735,
+      "step": 3317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.688801202467003e-05,
+      "loss": 0.5565,
+      "step": 3318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6886203695149825e-05,
+      "loss": 0.255,
+      "step": 3319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6884394901231817e-05,
+      "loss": 0.4273,
+      "step": 3320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.688258564298667e-05,
+      "loss": 1.6846,
+      "step": 3321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6880775920485087e-05,
+      "loss": 0.4923,
+      "step": 3322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.687896573379776e-05,
+      "loss": 0.3572,
+      "step": 3323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6877155082995416e-05,
+      "loss": 0.6939,
+      "step": 3324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6875343968148804e-05,
+      "loss": 2.2136,
+      "step": 3325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6873532389328683e-05,
+      "loss": 0.2524,
+      "step": 3326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6871720346605825e-05,
+      "loss": 0.785,
+      "step": 3327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6869907840051035e-05,
+      "loss": 0.6506,
+      "step": 3328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.686809486973512e-05,
+      "loss": 0.4326,
+      "step": 3329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6866281435728922e-05,
+      "loss": 1.5486,
+      "step": 3330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6864467538103294e-05,
+      "loss": 0.9818,
+      "step": 3331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.686265317692909e-05,
+      "loss": 1.1909,
+      "step": 3332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.686083835227722e-05,
+      "loss": 0.5699,
+      "step": 3333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.685902306421857e-05,
+      "loss": 1.2874,
+      "step": 3334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6857207312824074e-05,
+      "loss": 0.8347,
+      "step": 3335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6855391098164673e-05,
+      "loss": 1.0064,
+      "step": 3336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6853574420311323e-05,
+      "loss": 1.455,
+      "step": 3337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.685175727933501e-05,
+      "loss": 1.0073,
+      "step": 3338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.684993967530672e-05,
+      "loss": 1.1332,
+      "step": 3339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.684812160829748e-05,
+      "loss": 1.2803,
+      "step": 3340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.684630307837831e-05,
+      "loss": 0.9162,
+      "step": 3341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.684448408562027e-05,
+      "loss": 0.6021,
+      "step": 3342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6842664630094417e-05,
+      "loss": 0.4186,
+      "step": 3343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.684084471187185e-05,
+      "loss": 0.6693,
+      "step": 3344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.683902433102367e-05,
+      "loss": 0.7975,
+      "step": 3345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6837203487621e-05,
+      "loss": 0.326,
+      "step": 3346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6835382181734976e-05,
+      "loss": 0.6286,
+      "step": 3347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6833560413436762e-05,
+      "loss": 0.3288,
+      "step": 3348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.683173818279754e-05,
+      "loss": 0.9283,
+      "step": 3349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6829915489888488e-05,
+      "loss": 0.6968,
+      "step": 3350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6828092334780838e-05,
+      "loss": 0.3589,
+      "step": 3351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6826268717545808e-05,
+      "loss": 1.2399,
+      "step": 3352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.682444463825466e-05,
+      "loss": 1.4079,
+      "step": 3353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.682262009697865e-05,
+      "loss": 0.3243,
+      "step": 3354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6820795093789067e-05,
+      "loss": 0.3601,
+      "step": 3355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6818969628757212e-05,
+      "loss": 0.4429,
+      "step": 3356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6817143701954413e-05,
+      "loss": 0.6966,
+      "step": 3357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.681531731345201e-05,
+      "loss": 0.2718,
+      "step": 3358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6813490463321357e-05,
+      "loss": 0.1664,
+      "step": 3359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.681166315163382e-05,
+      "loss": 0.2744,
+      "step": 3360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.680983537846081e-05,
+      "loss": 0.8368,
+      "step": 3361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6808007143873724e-05,
+      "loss": 0.9882,
+      "step": 3362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6806178447944007e-05,
+      "loss": 0.5906,
+      "step": 3363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6804349290743094e-05,
+      "loss": 0.6033,
+      "step": 3364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6802519672342453e-05,
+      "loss": 0.2049,
+      "step": 3365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.680068959281357e-05,
+      "loss": 0.3035,
+      "step": 3366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6798859052227946e-05,
+      "loss": 0.4499,
+      "step": 3367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6797028050657105e-05,
+      "loss": 0.3798,
+      "step": 3368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.679519658817258e-05,
+      "loss": 1.0905,
+      "step": 3369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6793364664845926e-05,
+      "loss": 1.0125,
+      "step": 3370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6791532280748718e-05,
+      "loss": 0.4191,
+      "step": 3371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.678969943595255e-05,
+      "loss": 0.8512,
+      "step": 3372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6787866130529033e-05,
+      "loss": 1.4681,
+      "step": 3373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.678603236454979e-05,
+      "loss": 0.6632,
+      "step": 3374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6784198138086467e-05,
+      "loss": 1.0893,
+      "step": 3375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6782363451210733e-05,
+      "loss": 1.1126,
+      "step": 3376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.678052830399427e-05,
+      "loss": 1.2893,
+      "step": 3377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6778692696508765e-05,
+      "loss": 0.9957,
+      "step": 3378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.677685662882595e-05,
+      "loss": 1.258,
+      "step": 3379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6775020101017558e-05,
+      "loss": 1.2565,
+      "step": 3380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6773183113155338e-05,
+      "loss": 0.7913,
+      "step": 3381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6771345665311063e-05,
+      "loss": 0.3546,
+      "step": 3382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.676950775755653e-05,
+      "loss": 1.0603,
+      "step": 3383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6767669389963537e-05,
+      "loss": 1.1606,
+      "step": 3384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6765830562603914e-05,
+      "loss": 2.033,
+      "step": 3385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6763991275549503e-05,
+      "loss": 0.7891,
+      "step": 3386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.676215152887217e-05,
+      "loss": 0.7441,
+      "step": 3387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6760311322643786e-05,
+      "loss": 0.3825,
+      "step": 3388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6758470656936256e-05,
+      "loss": 0.5199,
+      "step": 3389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.675662953182149e-05,
+      "loss": 0.8058,
+      "step": 3390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6754787947371428e-05,
+      "loss": 0.6197,
+      "step": 3391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.675294590365802e-05,
+      "loss": 0.6285,
+      "step": 3392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6751103400753226e-05,
+      "loss": 1.0081,
+      "step": 3393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6749260438729042e-05,
+      "loss": 0.5226,
+      "step": 3394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6747417017657473e-05,
+      "loss": 0.3114,
+      "step": 3395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6745573137610538e-05,
+      "loss": 0.5636,
+      "step": 3396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.674372879866028e-05,
+      "loss": 0.5811,
+      "step": 3397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6741884000878765e-05,
+      "loss": 0.6907,
+      "step": 3398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6740038744338055e-05,
+      "loss": 0.4881,
+      "step": 3399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6738193029110258e-05,
+      "loss": 0.7309,
+      "step": 3400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6736346855267476e-05,
+      "loss": 0.122,
+      "step": 3401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6734500222881842e-05,
+      "loss": 0.2578,
+      "step": 3402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6732653132025518e-05,
+      "loss": 0.2501,
+      "step": 3403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6730805582770654e-05,
+      "loss": 0.4362,
+      "step": 3404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6728957575189436e-05,
+      "loss": 0.5854,
+      "step": 3405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6727109109354076e-05,
+      "loss": 0.4155,
+      "step": 3406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.672526018533679e-05,
+      "loss": 0.4432,
+      "step": 3407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.672341080320981e-05,
+      "loss": 0.2691,
+      "step": 3408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6721560963045398e-05,
+      "loss": 0.3383,
+      "step": 3409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.671971066491583e-05,
+      "loss": 0.465,
+      "step": 3410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.671785990889339e-05,
+      "loss": 1.7861,
+      "step": 3411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6716008695050396e-05,
+      "loss": 0.151,
+      "step": 3412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.671415702345917e-05,
+      "loss": 0.1854,
+      "step": 3413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6712304894192058e-05,
+      "loss": 0.9073,
+      "step": 3414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.671045230732142e-05,
+      "loss": 0.8076,
+      "step": 3415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6708599262919644e-05,
+      "loss": 0.7682,
+      "step": 3416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.670674576105913e-05,
+      "loss": 0.6494,
+      "step": 3417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.670489180181229e-05,
+      "loss": 0.9185,
+      "step": 3418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6703037385251556e-05,
+      "loss": 1.2848,
+      "step": 3419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6701182511449385e-05,
+      "loss": 1.1788,
+      "step": 3420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6699327180478253e-05,
+      "loss": 0.8183,
+      "step": 3421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6697471392410636e-05,
+      "loss": 0.1767,
+      "step": 3422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6695615147319047e-05,
+      "loss": 0.8418,
+      "step": 3423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6693758445276015e-05,
+      "loss": 0.9959,
+      "step": 3424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6691901286354074e-05,
+      "loss": 0.8536,
+      "step": 3425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6690043670625788e-05,
+      "loss": 1.5788,
+      "step": 3426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.668818559816373e-05,
+      "loss": 0.8945,
+      "step": 3427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.66863270690405e-05,
+      "loss": 0.8844,
+      "step": 3428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.668446808332871e-05,
+      "loss": 0.8653,
+      "step": 3429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6682608641100992e-05,
+      "loss": 1.6874,
+      "step": 3430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6680748742429997e-05,
+      "loss": 1.226,
+      "step": 3431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6678888387388387e-05,
+      "loss": 0.9831,
+      "step": 3432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6677027576048845e-05,
+      "loss": 0.3724,
+      "step": 3433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6675166308484082e-05,
+      "loss": 1.1123,
+      "step": 3434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6673304584766812e-05,
+      "loss": 0.6087,
+      "step": 3435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6671442404969774e-05,
+      "loss": 0.8382,
+      "step": 3436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6669579769165726e-05,
+      "loss": 0.3273,
+      "step": 3437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.666771667742744e-05,
+      "loss": 0.3642,
+      "step": 3438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6665853129827704e-05,
+      "loss": 0.9434,
+      "step": 3439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6663989126439336e-05,
+      "loss": 0.5348,
+      "step": 3440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6662124667335157e-05,
+      "loss": 0.9952,
+      "step": 3441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6660259752588015e-05,
+      "loss": 0.539,
+      "step": 3442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6658394382270772e-05,
+      "loss": 1.0596,
+      "step": 3443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6656528556456307e-05,
+      "loss": 0.8064,
+      "step": 3444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.665466227521752e-05,
+      "loss": 0.3852,
+      "step": 3445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6652795538627322e-05,
+      "loss": 0.5085,
+      "step": 3446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6650928346758662e-05,
+      "loss": 0.4764,
+      "step": 3447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6649060699684474e-05,
+      "loss": 0.3043,
+      "step": 3448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6647192597477735e-05,
+      "loss": 0.2945,
+      "step": 3449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6645324040211435e-05,
+      "loss": 0.5423,
+      "step": 3450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6643455027958576e-05,
+      "loss": 0.4076,
+      "step": 3451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.664158556079218e-05,
+      "loss": 0.3669,
+      "step": 3452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6639715638785292e-05,
+      "loss": 0.7506,
+      "step": 3453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6637845262010964e-05,
+      "loss": 0.6057,
+      "step": 3454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.663597443054228e-05,
+      "loss": 0.3919,
+      "step": 3455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.663410314445233e-05,
+      "loss": 0.5001,
+      "step": 3456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6632231403814223e-05,
+      "loss": 0.4514,
+      "step": 3457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6630359208701088e-05,
+      "loss": 0.2867,
+      "step": 3458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.662848655918608e-05,
+      "loss": 0.4211,
+      "step": 3459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.662661345534236e-05,
+      "loss": 0.1634,
+      "step": 3460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6624739897243108e-05,
+      "loss": 0.2981,
+      "step": 3461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6622865884961524e-05,
+      "loss": 0.7002,
+      "step": 3462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6620991418570835e-05,
+      "loss": 0.1864,
+      "step": 3463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6619116498144272e-05,
+      "loss": 0.676,
+      "step": 3464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6617241123755083e-05,
+      "loss": 0.7045,
+      "step": 3465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6615365295476544e-05,
+      "loss": 1.1498,
+      "step": 3466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6613489013381943e-05,
+      "loss": 0.38,
+      "step": 3467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6611612277544593e-05,
+      "loss": 0.5103,
+      "step": 3468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6609735088037812e-05,
+      "loss": 1.6887,
+      "step": 3469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.660785744493494e-05,
+      "loss": 0.1434,
+      "step": 3470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.660597934830935e-05,
+      "loss": 1.2348,
+      "step": 3471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.660410079823441e-05,
+      "loss": 1.3196,
+      "step": 3472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6602221794783508e-05,
+      "loss": 1.0607,
+      "step": 3473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.660034233803007e-05,
+      "loss": 1.3507,
+      "step": 3474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.659846242804753e-05,
+      "loss": 0.241,
+      "step": 3475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6596582064909324e-05,
+      "loss": 0.5883,
+      "step": 3476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6594701248688925e-05,
+      "loss": 1.9398,
+      "step": 3477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.659281997945982e-05,
+      "loss": 0.5243,
+      "step": 3478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.65909382572955e-05,
+      "loss": 0.702,
+      "step": 3479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6589056082269497e-05,
+      "loss": 0.6214,
+      "step": 3480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6587173454455343e-05,
+      "loss": 0.8672,
+      "step": 3481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.658529037392659e-05,
+      "loss": 1.3497,
+      "step": 3482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6583406840756813e-05,
+      "loss": 0.6123,
+      "step": 3483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6581522855019604e-05,
+      "loss": 0.5084,
+      "step": 3484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.657963841678857e-05,
+      "loss": 0.3158,
+      "step": 3485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6577753526137337e-05,
+      "loss": 0.8871,
+      "step": 3486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6575868183139545e-05,
+      "loss": 1.4688,
+      "step": 3487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.657398238786886e-05,
+      "loss": 1.0489,
+      "step": 3488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6572096140398955e-05,
+      "loss": 1.1244,
+      "step": 3489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6570209440803528e-05,
+      "loss": 1.0261,
+      "step": 3490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6568322289156294e-05,
+      "loss": 0.5691,
+      "step": 3491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6566434685530988e-05,
+      "loss": 1.3852,
+      "step": 3492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6564546630001356e-05,
+      "loss": 0.4414,
+      "step": 3493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6562658122641163e-05,
+      "loss": 0.8667,
+      "step": 3494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.65607691635242e-05,
+      "loss": 0.6333,
+      "step": 3495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.655887975272426e-05,
+      "loss": 0.4941,
+      "step": 3496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6556989890315172e-05,
+      "loss": 0.2884,
+      "step": 3497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6555099576370762e-05,
+      "loss": 0.9489,
+      "step": 3498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.65532088109649e-05,
+      "loss": 0.278,
+      "step": 3499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6551317594171448e-05,
+      "loss": 0.5222,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6549425926064298e-05,
+      "loss": 1.4394,
+      "step": 3501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6547533806717364e-05,
+      "loss": 0.4953,
+      "step": 3502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6545641236204564e-05,
+      "loss": 1.3811,
+      "step": 3503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6543748214599845e-05,
+      "loss": 0.5685,
+      "step": 3504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.654185474197717e-05,
+      "loss": 1.083,
+      "step": 3505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6539960818410515e-05,
+      "loss": 0.5548,
+      "step": 3506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6538066443973874e-05,
+      "loss": 0.4015,
+      "step": 3507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6536171618741267e-05,
+      "loss": 1.5511,
+      "step": 3508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.653427634278672e-05,
+      "loss": 0.6993,
+      "step": 3509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6532380616184284e-05,
+      "loss": 1.6051,
+      "step": 3510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6530484439008024e-05,
+      "loss": 0.2235,
+      "step": 3511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6528587811332026e-05,
+      "loss": 0.5832,
+      "step": 3512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6526690733230395e-05,
+      "loss": 0.4262,
+      "step": 3513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6524793204777243e-05,
+      "loss": 0.1131,
+      "step": 3514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6522895226046713e-05,
+      "loss": 1.3181,
+      "step": 3515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6520996797112957e-05,
+      "loss": 0.5816,
+      "step": 3516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.651909791805015e-05,
+      "loss": 0.4681,
+      "step": 3517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.651719858893248e-05,
+      "loss": 0.8116,
+      "step": 3518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6515298809834155e-05,
+      "loss": 0.6993,
+      "step": 3519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6513398580829395e-05,
+      "loss": 1.8988,
+      "step": 3520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6511497901992447e-05,
+      "loss": 1.0565,
+      "step": 3521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6509596773397575e-05,
+      "loss": 0.6425,
+      "step": 3522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.650769519511905e-05,
+      "loss": 0.9662,
+      "step": 3523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.650579316723117e-05,
+      "loss": 0.3245,
+      "step": 3524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6503890689808248e-05,
+      "loss": 0.5069,
+      "step": 3525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6501987762924614e-05,
+      "loss": 0.5492,
+      "step": 3526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6500084386654616e-05,
+      "loss": 1.9582,
+      "step": 3527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.649818056107262e-05,
+      "loss": 1.2882,
+      "step": 3528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6496276286253006e-05,
+      "loss": 0.4121,
+      "step": 3529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.649437156227018e-05,
+      "loss": 1.0353,
+      "step": 3530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6492466389198556e-05,
+      "loss": 0.6667,
+      "step": 3531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6490560767112574e-05,
+      "loss": 0.651,
+      "step": 3532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6488654696086686e-05,
+      "loss": 0.5184,
+      "step": 3533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.648674817619535e-05,
+      "loss": 0.5268,
+      "step": 3534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.648484120751308e-05,
+      "loss": 0.8756,
+      "step": 3535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.648293379011436e-05,
+      "loss": 1.1446,
+      "step": 3536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6481025924073722e-05,
+      "loss": 0.6566,
+      "step": 3537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6479117609465706e-05,
+      "loss": 0.9283,
+      "step": 3538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6477208846364872e-05,
+      "loss": 0.945,
+      "step": 3539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.647529963484579e-05,
+      "loss": 0.6153,
+      "step": 3540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.647338997498306e-05,
+      "loss": 0.558,
+      "step": 3541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.647147986685129e-05,
+      "loss": 1.1769,
+      "step": 3542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.646956931052511e-05,
+      "loss": 1.0052,
+      "step": 3543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6467658306079173e-05,
+      "loss": 0.4781,
+      "step": 3544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6465746853588124e-05,
+      "loss": 0.1924,
+      "step": 3545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6463834953126662e-05,
+      "loss": 0.9663,
+      "step": 3546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6461922604769474e-05,
+      "loss": 0.5175,
+      "step": 3547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6460009808591284e-05,
+      "loss": 0.8176,
+      "step": 3548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.645809656466682e-05,
+      "loss": 1.0148,
+      "step": 3549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6456182873070838e-05,
+      "loss": 1.1148,
+      "step": 3550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.64542687338781e-05,
+      "loss": 1.3173,
+      "step": 3551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.64523541471634e-05,
+      "loss": 1.2007,
+      "step": 3552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6450439113001534e-05,
+      "loss": 0.249,
+      "step": 3553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.644852363146733e-05,
+      "loss": 0.5289,
+      "step": 3554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.644660770263562e-05,
+      "loss": 1.9786,
+      "step": 3555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6444691326581264e-05,
+      "loss": 1.5345,
+      "step": 3556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6442774503379133e-05,
+      "loss": 1.2944,
+      "step": 3557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.644085723310412e-05,
+      "loss": 0.4714,
+      "step": 3558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6438939515831134e-05,
+      "loss": 0.6874,
+      "step": 3559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6437021351635098e-05,
+      "loss": 1.3525,
+      "step": 3560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6435102740590954e-05,
+      "loss": 0.1444,
+      "step": 3561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6433183682773672e-05,
+      "loss": 2.2013,
+      "step": 3562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6431264178258216e-05,
+      "loss": 1.6523,
+      "step": 3563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6429344227119598e-05,
+      "loss": 0.2422,
+      "step": 3564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6427423829432816e-05,
+      "loss": 0.3166,
+      "step": 3565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6425502985272907e-05,
+      "loss": 0.2652,
+      "step": 3566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6423581694714923e-05,
+      "loss": 0.211,
+      "step": 3567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6421659957833923e-05,
+      "loss": 0.41,
+      "step": 3568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6419737774704992e-05,
+      "loss": 1.8987,
+      "step": 3569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6417815145403228e-05,
+      "loss": 0.5181,
+      "step": 3570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6415892070003757e-05,
+      "loss": 0.6335,
+      "step": 3571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6413968548581705e-05,
+      "loss": 0.7232,
+      "step": 3572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6412044581212224e-05,
+      "loss": 0.4283,
+      "step": 3573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6410120167970497e-05,
+      "loss": 0.5925,
+      "step": 3574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.64081953089317e-05,
+      "loss": 1.0288,
+      "step": 3575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6406270004171037e-05,
+      "loss": 0.3311,
+      "step": 3576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6404344253763733e-05,
+      "loss": 0.801,
+      "step": 3577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6402418057785034e-05,
+      "loss": 0.9683,
+      "step": 3578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.640049141631019e-05,
+      "loss": 0.2968,
+      "step": 3579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6398564329414474e-05,
+      "loss": 0.522,
+      "step": 3580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6396636797173182e-05,
+      "loss": 1.0415,
+      "step": 3581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6394708819661625e-05,
+      "loss": 1.1002,
+      "step": 3582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6392780396955124e-05,
+      "loss": 1.6135,
+      "step": 3583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.639085152912903e-05,
+      "loss": 1.1093,
+      "step": 3584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.638892221625869e-05,
+      "loss": 1.8163,
+      "step": 3585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.63869924584195e-05,
+      "loss": 1.3341,
+      "step": 3586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.638506225568685e-05,
+      "loss": 0.4209,
+      "step": 3587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6383131608136154e-05,
+      "loss": 0.3399,
+      "step": 3588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6381200515842842e-05,
+      "loss": 0.9551,
+      "step": 3589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6379268978882363e-05,
+      "loss": 1.5302,
+      "step": 3590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6377336997330183e-05,
+      "loss": 1.425,
+      "step": 3591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6375404571261784e-05,
+      "loss": 0.7154,
+      "step": 3592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6373471700752666e-05,
+      "loss": 0.241,
+      "step": 3593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6371538385878348e-05,
+      "loss": 0.6663,
+      "step": 3594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6369604626714365e-05,
+      "loss": 0.8717,
+      "step": 3595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6367670423336275e-05,
+      "loss": 0.8358,
+      "step": 3596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6365735775819635e-05,
+      "loss": 1.158,
+      "step": 3597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6363800684240045e-05,
+      "loss": 0.5242,
+      "step": 3598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6361865148673103e-05,
+      "loss": 1.0398,
+      "step": 3599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6359929169194436e-05,
+      "loss": 0.3427,
+      "step": 3600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6357992745879675e-05,
+      "loss": 1.6589,
+      "step": 3601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6356055878804482e-05,
+      "loss": 1.0776,
+      "step": 3602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.635411856804453e-05,
+      "loss": 0.3214,
+      "step": 3603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6352180813675516e-05,
+      "loss": 1.6722,
+      "step": 3604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.635024261577314e-05,
+      "loss": 0.4059,
+      "step": 3605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6348303974413138e-05,
+      "loss": 1.5513,
+      "step": 3606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.634636488967124e-05,
+      "loss": 0.5137,
+      "step": 3607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6344425361623214e-05,
+      "loss": 0.5828,
+      "step": 3608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.634248539034484e-05,
+      "loss": 0.2622,
+      "step": 3609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6340544975911914e-05,
+      "loss": 1.07,
+      "step": 3610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6338604118400244e-05,
+      "loss": 0.2388,
+      "step": 3611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.633666281788566e-05,
+      "loss": 1.7012,
+      "step": 3612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6334721074444013e-05,
+      "loss": 0.7132,
+      "step": 3613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.633277888815116e-05,
+      "loss": 1.4487,
+      "step": 3614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6330836259082995e-05,
+      "loss": 1.1846,
+      "step": 3615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.632889318731541e-05,
+      "loss": 0.7058,
+      "step": 3616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6326949672924316e-05,
+      "loss": 0.2439,
+      "step": 3617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6325005715985658e-05,
+      "loss": 0.1263,
+      "step": 3618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6323061316575378e-05,
+      "loss": 1.3119,
+      "step": 3619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.632111647476945e-05,
+      "loss": 0.2399,
+      "step": 3620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6319171190643858e-05,
+      "loss": 0.1973,
+      "step": 3621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6317225464274602e-05,
+      "loss": 1.0982,
+      "step": 3622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.631527929573771e-05,
+      "loss": 1.6811,
+      "step": 3623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6313332685109208e-05,
+      "loss": 1.8014,
+      "step": 3624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.631138563246516e-05,
+      "loss": 0.9517,
+      "step": 3625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.630943813788163e-05,
+      "loss": 1.1289,
+      "step": 3626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.630749020143472e-05,
+      "loss": 0.9672,
+      "step": 3627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.630554182320052e-05,
+      "loss": 0.6329,
+      "step": 3628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.630359300325517e-05,
+      "loss": 1.0834,
+      "step": 3629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6301643741674795e-05,
+      "loss": 0.5404,
+      "step": 3630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6299694038535566e-05,
+      "loss": 1.1846,
+      "step": 3631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6297743893913654e-05,
+      "loss": 0.937,
+      "step": 3632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6295793307885252e-05,
+      "loss": 1.0168,
+      "step": 3633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.629384228052657e-05,
+      "loss": 1.5502,
+      "step": 3634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6291890811913834e-05,
+      "loss": 0.979,
+      "step": 3635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6289938902123287e-05,
+      "loss": 0.1747,
+      "step": 3636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6287986551231198e-05,
+      "loss": 0.6076,
+      "step": 3637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.628603375931384e-05,
+      "loss": 0.6952,
+      "step": 3638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6284080526447507e-05,
+      "loss": 0.2494,
+      "step": 3639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.628212685270852e-05,
+      "loss": 0.2704,
+      "step": 3640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6280172738173208e-05,
+      "loss": 0.1865,
+      "step": 3641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.627821818291791e-05,
+      "loss": 0.563,
+      "step": 3642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6276263187019003e-05,
+      "loss": 0.9749,
+      "step": 3643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6274307750552858e-05,
+      "loss": 1.1008,
+      "step": 3644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6272351873595886e-05,
+      "loss": 0.7138,
+      "step": 3645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6270395556224498e-05,
+      "loss": 0.6546,
+      "step": 3646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6268438798515125e-05,
+      "loss": 0.7637,
+      "step": 3647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6266481600544223e-05,
+      "loss": 0.2847,
+      "step": 3648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6264523962388256e-05,
+      "loss": 0.608,
+      "step": 3649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6262565884123716e-05,
+      "loss": 0.4417,
+      "step": 3650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6260607365827102e-05,
+      "loss": 0.4757,
+      "step": 3651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.625864840757493e-05,
+      "loss": 1.379,
+      "step": 3652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6256689009443742e-05,
+      "loss": 0.4424,
+      "step": 3653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6254729171510095e-05,
+      "loss": 2.0498,
+      "step": 3654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6252768893850555e-05,
+      "loss": 0.2796,
+      "step": 3655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.625080817654171e-05,
+      "loss": 0.4495,
+      "step": 3656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.624884701966017e-05,
+      "loss": 0.3484,
+      "step": 3657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.624688542328256e-05,
+      "loss": 0.9957,
+      "step": 3658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.624492338748551e-05,
+      "loss": 0.7118,
+      "step": 3659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.624296091234569e-05,
+      "loss": 0.5317,
+      "step": 3660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6240997997939765e-05,
+      "loss": 0.4064,
+      "step": 3661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.623903464434443e-05,
+      "loss": 0.2432,
+      "step": 3662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6237070851636396e-05,
+      "loss": 0.5197,
+      "step": 3663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6235106619892384e-05,
+      "loss": 0.3296,
+      "step": 3664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6233141949189146e-05,
+      "loss": 1.5427,
+      "step": 3665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.623117683960343e-05,
+      "loss": 0.4992,
+      "step": 3666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6229211291212024e-05,
+      "loss": 0.6284,
+      "step": 3667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6227245304091714e-05,
+      "loss": 0.5629,
+      "step": 3668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6225278878319326e-05,
+      "loss": 0.446,
+      "step": 3669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.622331201397167e-05,
+      "loss": 0.0624,
+      "step": 3670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.622134471112561e-05,
+      "loss": 0.2973,
+      "step": 3671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.621937696985799e-05,
+      "loss": 1.3189,
+      "step": 3672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.621740879024571e-05,
+      "loss": 1.542,
+      "step": 3673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.621544017236565e-05,
+      "loss": 1.4903,
+      "step": 3674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6213471116294737e-05,
+      "loss": 1.4125,
+      "step": 3675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6211501622109898e-05,
+      "loss": 0.737,
+      "step": 3676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6209531689888083e-05,
+      "loss": 1.189,
+      "step": 3677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6207561319706253e-05,
+      "loss": 0.6691,
+      "step": 3678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6205590511641397e-05,
+      "loss": 1.2921,
+      "step": 3679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6203619265770515e-05,
+      "loss": 0.6214,
+      "step": 3680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6201647582170622e-05,
+      "loss": 1.3403,
+      "step": 3681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6199675460918752e-05,
+      "loss": 0.8607,
+      "step": 3682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.619770290209196e-05,
+      "loss": 0.6417,
+      "step": 3683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6195729905767306e-05,
+      "loss": 0.7946,
+      "step": 3684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6193756472021883e-05,
+      "loss": 1.0183,
+      "step": 3685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6191782600932793e-05,
+      "loss": 0.6694,
+      "step": 3686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6189808292577153e-05,
+      "loss": 0.7511,
+      "step": 3687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6187833547032106e-05,
+      "loss": 1.5179,
+      "step": 3688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6185858364374795e-05,
+      "loss": 0.8747,
+      "step": 3689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6183882744682398e-05,
+      "loss": 0.3404,
+      "step": 3690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6181906688032107e-05,
+      "loss": 0.6175,
+      "step": 3691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.617993019450112e-05,
+      "loss": 1.1449,
+      "step": 3692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6177953264166666e-05,
+      "loss": 1.1472,
+      "step": 3693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6175975897105974e-05,
+      "loss": 0.6455,
+      "step": 3694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6173998093396307e-05,
+      "loss": 0.5371,
+      "step": 3695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6172019853114943e-05,
+      "loss": 0.8926,
+      "step": 3696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6170041176339165e-05,
+      "loss": 0.6372,
+      "step": 3697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6168062063146283e-05,
+      "loss": 0.3502,
+      "step": 3698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.616608251361362e-05,
+      "loss": 0.4644,
+      "step": 3699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6164102527818526e-05,
+      "loss": 0.2642,
+      "step": 3700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6162122105838344e-05,
+      "loss": 0.3031,
+      "step": 3701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6160141247750462e-05,
+      "loss": 2.1201,
+      "step": 3702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.615815995363227e-05,
+      "loss": 0.1576,
+      "step": 3703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.615617822356118e-05,
+      "loss": 1.1502,
+      "step": 3704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6154196057614615e-05,
+      "loss": 0.2784,
+      "step": 3705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6152213455870014e-05,
+      "loss": 1.5583,
+      "step": 3706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6150230418404848e-05,
+      "loss": 0.6582,
+      "step": 3707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.614824694529659e-05,
+      "loss": 0.5726,
+      "step": 3708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.614626303662274e-05,
+      "loss": 0.4181,
+      "step": 3709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.61442786924608e-05,
+      "loss": 0.4424,
+      "step": 3710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6142293912888305e-05,
+      "loss": 1.4756,
+      "step": 3711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6140308697982805e-05,
+      "loss": 0.2609,
+      "step": 3712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6138323047821857e-05,
+      "loss": 0.0372,
+      "step": 3713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.613633696248304e-05,
+      "loss": 1.3051,
+      "step": 3714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6134350442043956e-05,
+      "loss": 0.9296,
+      "step": 3715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6132363486582213e-05,
+      "loss": 1.2015,
+      "step": 3716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.613037609617545e-05,
+      "loss": 1.0719,
+      "step": 3717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.612838827090131e-05,
+      "loss": 0.2429,
+      "step": 3718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6126400010837456e-05,
+      "loss": 0.5302,
+      "step": 3719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6124411316061573e-05,
+      "loss": 1.772,
+      "step": 3720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.612242218665136e-05,
+      "loss": 0.7285,
+      "step": 3721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.612043262268453e-05,
+      "loss": 1.1349,
+      "step": 3722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6118442624238822e-05,
+      "loss": 0.7485,
+      "step": 3723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.611645219139198e-05,
+      "loss": 0.7447,
+      "step": 3724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.611446132422177e-05,
+      "loss": 0.6277,
+      "step": 3725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6112470022805985e-05,
+      "loss": 1.018,
+      "step": 3726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6110478287222418e-05,
+      "loss": 0.9364,
+      "step": 3727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.610848611754889e-05,
+      "loss": 1.5514,
+      "step": 3728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6106493513863226e-05,
+      "loss": 0.8296,
+      "step": 3729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6104500476243293e-05,
+      "loss": 0.7578,
+      "step": 3730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.610250700476695e-05,
+      "loss": 0.9695,
+      "step": 3731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.610051309951208e-05,
+      "loss": 0.297,
+      "step": 3732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6098518760556603e-05,
+      "loss": 1.0655,
+      "step": 3733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6096523987978416e-05,
+      "loss": 0.1886,
+      "step": 3734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.609452878185547e-05,
+      "loss": 0.8375,
+      "step": 3735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6092533142265707e-05,
+      "loss": 0.6839,
+      "step": 3736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6090537069287107e-05,
+      "loss": 0.4662,
+      "step": 3737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6088540562997655e-05,
+      "loss": 0.4236,
+      "step": 3738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6086543623475358e-05,
+      "loss": 1.2864,
+      "step": 3739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6084546250798226e-05,
+      "loss": 0.6246,
+      "step": 3740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6082548445044312e-05,
+      "loss": 0.417,
+      "step": 3741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6080550206291658e-05,
+      "loss": 1.0593,
+      "step": 3742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6078551534618344e-05,
+      "loss": 0.6395,
+      "step": 3743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6076552430102452e-05,
+      "loss": 0.6663,
+      "step": 3744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6074552892822093e-05,
+      "loss": 0.5687,
+      "step": 3745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.607255292285539e-05,
+      "loss": 0.4737,
+      "step": 3746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6070552520280477e-05,
+      "loss": 0.1289,
+      "step": 3747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6068551685175518e-05,
+      "loss": 1.2632,
+      "step": 3748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6066550417618677e-05,
+      "loss": 0.7764,
+      "step": 3749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6064548717688155e-05,
+      "loss": 0.92,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.606254658546215e-05,
+      "loss": 0.3143,
+      "step": 3751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.606054402101889e-05,
+      "loss": 1.7319,
+      "step": 3752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6058541024436616e-05,
+      "loss": 0.3722,
+      "step": 3753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6056537595793585e-05,
+      "loss": 1.6337,
+      "step": 3754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.605453373516807e-05,
+      "loss": 0.3674,
+      "step": 3755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6052529442638365e-05,
+      "loss": 1.9033,
+      "step": 3756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6050524718282777e-05,
+      "loss": 0.1003,
+      "step": 3757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6048519562179628e-05,
+      "loss": 0.4142,
+      "step": 3758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6046513974407273e-05,
+      "loss": 0.7525,
+      "step": 3759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6044507955044055e-05,
+      "loss": 0.8943,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6042501504168355e-05,
+      "loss": 1.1106,
+      "step": 3761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.604049462185857e-05,
+      "loss": 1.7609,
+      "step": 3762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6038487308193103e-05,
+      "loss": 0.8371,
+      "step": 3763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6036479563250387e-05,
+      "loss": 0.9088,
+      "step": 3764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6034471387108863e-05,
+      "loss": 0.6391,
+      "step": 3765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.603246277984699e-05,
+      "loss": 1.3606,
+      "step": 3766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6030453741543246e-05,
+      "loss": 0.6574,
+      "step": 3767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6028444272276122e-05,
+      "loss": 0.5472,
+      "step": 3768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.602643437212413e-05,
+      "loss": 1.7032,
+      "step": 3769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.60244240411658e-05,
+      "loss": 0.731,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.602241327947968e-05,
+      "loss": 0.0942,
+      "step": 3771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6020402087144316e-05,
+      "loss": 0.9381,
+      "step": 3772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6018390464238298e-05,
+      "loss": 0.9465,
+      "step": 3773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.601637841084022e-05,
+      "loss": 1.4339,
+      "step": 3774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.601436592702869e-05,
+      "loss": 1.4506,
+      "step": 3775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.601235301288234e-05,
+      "loss": 1.2619,
+      "step": 3776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.601033966847981e-05,
+      "loss": 0.8818,
+      "step": 3777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.600832589389977e-05,
+      "loss": 0.3788,
+      "step": 3778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6006311689220893e-05,
+      "loss": 1.8177,
+      "step": 3779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6004297054521874e-05,
+      "loss": 0.8396,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6002281989881427e-05,
+      "loss": 0.8208,
+      "step": 3781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6000266495378284e-05,
+      "loss": 1.235,
+      "step": 3782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5998250571091187e-05,
+      "loss": 0.8773,
+      "step": 3783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.59962342170989e-05,
+      "loss": 0.9258,
+      "step": 3784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5994217433480203e-05,
+      "loss": 1.123,
+      "step": 3785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.599220022031389e-05,
+      "loss": 0.4463,
+      "step": 3786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5990182577678786e-05,
+      "loss": 0.6511,
+      "step": 3787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5988164505653704e-05,
+      "loss": 1.5633,
+      "step": 3788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5986146004317497e-05,
+      "loss": 0.4581,
+      "step": 3789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5984127073749026e-05,
+      "loss": 0.6701,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5982107714027184e-05,
+      "loss": 1.0759,
+      "step": 3791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5980087925230853e-05,
+      "loss": 0.5118,
+      "step": 3792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5978067707438957e-05,
+      "loss": 0.8167,
+      "step": 3793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5976047060730418e-05,
+      "loss": 0.3064,
+      "step": 3794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5974025985184188e-05,
+      "loss": 0.8569,
+      "step": 3795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5972004480879234e-05,
+      "loss": 0.3734,
+      "step": 3796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5969982547894532e-05,
+      "loss": 0.441,
+      "step": 3797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.596796018630908e-05,
+      "loss": 0.7697,
+      "step": 3798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5965937396201897e-05,
+      "loss": 0.2338,
+      "step": 3799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5963914177652003e-05,
+      "loss": 0.3517,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.596189053073846e-05,
+      "loss": 1.2254,
+      "step": 3801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5959866455540325e-05,
+      "loss": 1.2597,
+      "step": 3802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5957841952136678e-05,
+      "loss": 0.4293,
+      "step": 3803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.595581702060662e-05,
+      "loss": 1.2904,
+      "step": 3804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5953791661029263e-05,
+      "loss": 1.3695,
+      "step": 3805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5951765873483745e-05,
+      "loss": 0.4787,
+      "step": 3806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5949739658049204e-05,
+      "loss": 0.179,
+      "step": 3807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5947713014804812e-05,
+      "loss": 0.9478,
+      "step": 3808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5945685943829752e-05,
+      "loss": 0.4268,
+      "step": 3809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5943658445203216e-05,
+      "loss": 1.1682,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5941630519004423e-05,
+      "loss": 0.5109,
+      "step": 3811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.59396021653126e-05,
+      "loss": 1.0713,
+      "step": 3812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5937573384207003e-05,
+      "loss": 0.2276,
+      "step": 3813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5935544175766895e-05,
+      "loss": 0.9752,
+      "step": 3814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5933514540071555e-05,
+      "loss": 0.5129,
+      "step": 3815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5931484477200284e-05,
+      "loss": 0.1802,
+      "step": 3816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.592945398723239e-05,
+      "loss": 0.5685,
+      "step": 3817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.592742307024722e-05,
+      "loss": 0.4453,
+      "step": 3818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5925391726324108e-05,
+      "loss": 0.1922,
+      "step": 3819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.592335995554243e-05,
+      "loss": 0.1365,
+      "step": 3820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5921327757981557e-05,
+      "loss": 0.2729,
+      "step": 3821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5919295133720896e-05,
+      "loss": 0.9366,
+      "step": 3822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.591726208283986e-05,
+      "loss": 2.1523,
+      "step": 3823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.591522860541788e-05,
+      "loss": 0.2762,
+      "step": 3824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.59131947015344e-05,
+      "loss": 1.7008,
+      "step": 3825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5911160371268897e-05,
+      "loss": 0.4931,
+      "step": 3826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5909125614700847e-05,
+      "loss": 0.367,
+      "step": 3827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5907090431909744e-05,
+      "loss": 1.5396,
+      "step": 3828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5905054822975112e-05,
+      "loss": 0.6447,
+      "step": 3829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5903018787976474e-05,
+      "loss": 0.8966,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5900982326993382e-05,
+      "loss": 0.8085,
+      "step": 3831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5898945440105404e-05,
+      "loss": 0.7569,
+      "step": 3832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.589690812739212e-05,
+      "loss": 1.0709,
+      "step": 3833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5894870388933128e-05,
+      "loss": 0.4037,
+      "step": 3834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.589283222480804e-05,
+      "loss": 0.4034,
+      "step": 3835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5890793635096495e-05,
+      "loss": 1.143,
+      "step": 3836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5888754619878138e-05,
+      "loss": 0.7528,
+      "step": 3837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.588671517923263e-05,
+      "loss": 1.4577,
+      "step": 3838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.588467531323966e-05,
+      "loss": 0.7565,
+      "step": 3839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.588263502197892e-05,
+      "loss": 0.9974,
+      "step": 3840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5880594305530128e-05,
+      "loss": 1.1343,
+      "step": 3841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5878553163973013e-05,
+      "loss": 0.7496,
+      "step": 3842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5876511597387322e-05,
+      "loss": 0.7201,
+      "step": 3843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5874469605852827e-05,
+      "loss": 1.0481,
+      "step": 3844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5872427189449304e-05,
+      "loss": 1.0672,
+      "step": 3845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.587038434825655e-05,
+      "loss": 0.7349,
+      "step": 3846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5868341082354377e-05,
+      "loss": 0.6533,
+      "step": 3847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5866297391822624e-05,
+      "loss": 0.3693,
+      "step": 3848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5864253276741138e-05,
+      "loss": 0.248,
+      "step": 3849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5862208737189772e-05,
+      "loss": 0.5361,
+      "step": 3850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.586016377324842e-05,
+      "loss": 0.2118,
+      "step": 3851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.585811838499697e-05,
+      "loss": 1.4342,
+      "step": 3852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5856072572515344e-05,
+      "loss": 2.077,
+      "step": 3853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5854026335883466e-05,
+      "loss": 1.5646,
+      "step": 3854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5851979675181287e-05,
+      "loss": 1.4033,
+      "step": 3855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5849932590488764e-05,
+      "loss": 0.8304,
+      "step": 3856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.584788508188589e-05,
+      "loss": 0.3135,
+      "step": 3857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5845837149452652e-05,
+      "loss": 1.5861,
+      "step": 3858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5843788793269068e-05,
+      "loss": 0.6285,
+      "step": 3859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5841740013415162e-05,
+      "loss": 1.8801,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5839690809970982e-05,
+      "loss": 0.1971,
+      "step": 3861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5837641183016597e-05,
+      "loss": 0.5475,
+      "step": 3862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5835591132632084e-05,
+      "loss": 0.3727,
+      "step": 3863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5833540658897533e-05,
+      "loss": 0.8746,
+      "step": 3864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5831489761893066e-05,
+      "loss": 1.2601,
+      "step": 3865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5829438441698806e-05,
+      "loss": 0.7152,
+      "step": 3866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.58273866983949e-05,
+      "loss": 0.5582,
+      "step": 3867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5825334532061514e-05,
+      "loss": 0.3281,
+      "step": 3868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.582328194277882e-05,
+      "loss": 1.5604,
+      "step": 3869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5821228930627018e-05,
+      "loss": 0.9226,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5819175495686317e-05,
+      "loss": 0.2768,
+      "step": 3871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.581712163803695e-05,
+      "loss": 1.0929,
+      "step": 3872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.581506735775916e-05,
+      "loss": 0.4965,
+      "step": 3873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5813012654933205e-05,
+      "loss": 1.5249,
+      "step": 3874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.581095752963936e-05,
+      "loss": 1.0685,
+      "step": 3875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5808901981957932e-05,
+      "loss": 0.4066,
+      "step": 3876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5806846011969224e-05,
+      "loss": 1.0939,
+      "step": 3877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5804789619753564e-05,
+      "loss": 0.5621,
+      "step": 3878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.580273280539129e-05,
+      "loss": 0.7161,
+      "step": 3879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5800675568962778e-05,
+      "loss": 0.8043,
+      "step": 3880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.579861791054839e-05,
+      "loss": 0.9646,
+      "step": 3881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5796559830228526e-05,
+      "loss": 0.75,
+      "step": 3882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.579450132808359e-05,
+      "loss": 1.5993,
+      "step": 3883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5792442404194018e-05,
+      "loss": 1.3109,
+      "step": 3884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5790383058640247e-05,
+      "loss": 0.5016,
+      "step": 3885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5788323291502734e-05,
+      "loss": 1.2225,
+      "step": 3886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5786263102861963e-05,
+      "loss": 0.86,
+      "step": 3887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5784202492798414e-05,
+      "loss": 0.4431,
+      "step": 3888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5782141461392606e-05,
+      "loss": 0.6577,
+      "step": 3889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5780080008725065e-05,
+      "loss": 0.6688,
+      "step": 3890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.577801813487632e-05,
+      "loss": 0.6747,
+      "step": 3891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5775955839926943e-05,
+      "loss": 0.5401,
+      "step": 3892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5773893123957503e-05,
+      "loss": 0.8026,
+      "step": 3893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.577182998704859e-05,
+      "loss": 0.6318,
+      "step": 3894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.576976642928081e-05,
+      "loss": 0.6065,
+      "step": 3895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5767702450734793e-05,
+      "loss": 0.5594,
+      "step": 3896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5765638051491174e-05,
+      "loss": 0.7692,
+      "step": 3897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.576357323163061e-05,
+      "loss": 0.7562,
+      "step": 3898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.576150799123378e-05,
+      "loss": 0.7313,
+      "step": 3899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5759442330381363e-05,
+      "loss": 0.3041,
+      "step": 3900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5757376249154073e-05,
+      "loss": 1.4608,
+      "step": 3901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5755309747632633e-05,
+      "loss": 1.4858,
+      "step": 3902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.575324282589778e-05,
+      "loss": 1.5015,
+      "step": 3903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5751175484030262e-05,
+      "loss": 1.4336,
+      "step": 3904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.574910772211086e-05,
+      "loss": 1.3613,
+      "step": 3905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5747039540220364e-05,
+      "loss": 1.2597,
+      "step": 3906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5744970938439572e-05,
+      "loss": 1.3286,
+      "step": 3907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.574290191684931e-05,
+      "loss": 0.4684,
+      "step": 3908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5740832475530406e-05,
+      "loss": 0.6278,
+      "step": 3909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5738762614563723e-05,
+      "loss": 0.4614,
+      "step": 3910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5736692334030128e-05,
+      "loss": 0.2836,
+      "step": 3911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.573462163401051e-05,
+      "loss": 1.07,
+      "step": 3912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5732550514585768e-05,
+      "loss": 1.6319,
+      "step": 3913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5730478975836823e-05,
+      "loss": 1.2232,
+      "step": 3914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5728407017844614e-05,
+      "loss": 1.1908,
+      "step": 3915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5726334640690084e-05,
+      "loss": 1.2951,
+      "step": 3916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.572426184445421e-05,
+      "loss": 0.4409,
+      "step": 3917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5722188629217974e-05,
+      "loss": 0.1699,
+      "step": 3918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.572011499506238e-05,
+      "loss": 0.5753,
+      "step": 3919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.571804094206844e-05,
+      "loss": 1.2444,
+      "step": 3920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5715966470317188e-05,
+      "loss": 0.2265,
+      "step": 3921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5713891579889682e-05,
+      "loss": 2.0278,
+      "step": 3922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5711816270866978e-05,
+      "loss": 0.575,
+      "step": 3923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.570974054333017e-05,
+      "loss": 0.3185,
+      "step": 3924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.570766439736035e-05,
+      "loss": 0.324,
+      "step": 3925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5705587833038636e-05,
+      "loss": 1.0421,
+      "step": 3926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5703510850446157e-05,
+      "loss": 0.531,
+      "step": 3927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.570143344966407e-05,
+      "loss": 0.6941,
+      "step": 3928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5699355630773527e-05,
+      "loss": 0.7179,
+      "step": 3929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.569727739385572e-05,
+      "loss": 0.692,
+      "step": 3930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5695198738991837e-05,
+      "loss": 0.8829,
+      "step": 3931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5693119666263108e-05,
+      "loss": 1.2705,
+      "step": 3932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5691040175750742e-05,
+      "loss": 0.7768,
+      "step": 3933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5688960267535998e-05,
+      "loss": 1.2774,
+      "step": 3934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5686879941700137e-05,
+      "loss": 0.3693,
+      "step": 3935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5684799198324436e-05,
+      "loss": 0.3958,
+      "step": 3936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.568271803749019e-05,
+      "loss": 0.5335,
+      "step": 3937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5680636459278714e-05,
+      "loss": 1.3761,
+      "step": 3938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5678554463771334e-05,
+      "loss": 0.9111,
+      "step": 3939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5676472051049392e-05,
+      "loss": 0.8285,
+      "step": 3940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.567438922119425e-05,
+      "loss": 0.9721,
+      "step": 3941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.567230597428729e-05,
+      "loss": 0.8568,
+      "step": 3942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5670222310409898e-05,
+      "loss": 0.7901,
+      "step": 3943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.566813822964349e-05,
+      "loss": 0.7242,
+      "step": 3944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5666053732069478e-05,
+      "loss": 1.0212,
+      "step": 3945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.566396881776932e-05,
+      "loss": 1.3353,
+      "step": 3946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.566188348682447e-05,
+      "loss": 0.7133,
+      "step": 3947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5659797739316397e-05,
+      "loss": 0.7299,
+      "step": 3948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5657711575326597e-05,
+      "loss": 0.2304,
+      "step": 3949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5655624994936576e-05,
+      "loss": 0.332,
+      "step": 3950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5653537998227856e-05,
+      "loss": 0.3898,
+      "step": 3951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.565145058528198e-05,
+      "loss": 1.227,
+      "step": 3952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5649362756180493e-05,
+      "loss": 0.4,
+      "step": 3953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5647274511004985e-05,
+      "loss": 0.405,
+      "step": 3954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.564518584983703e-05,
+      "loss": 1.4636,
+      "step": 3955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5643096772758242e-05,
+      "loss": 1.439,
+      "step": 3956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.564100727985023e-05,
+      "loss": 0.2308,
+      "step": 3957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5638917371194645e-05,
+      "loss": 0.2963,
+      "step": 3958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5636827046873135e-05,
+      "loss": 1.1281,
+      "step": 3959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.563473630696736e-05,
+      "loss": 0.6139,
+      "step": 3960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5632645151559025e-05,
+      "loss": 1.4387,
+      "step": 3961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5630553580729813e-05,
+      "loss": 1.5358,
+      "step": 3962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5628461594561456e-05,
+      "loss": 0.6405,
+      "step": 3963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5626369193135687e-05,
+      "loss": 0.6723,
+      "step": 3964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.562427637653425e-05,
+      "loss": 1.5704,
+      "step": 3965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.562218314483891e-05,
+      "loss": 0.7267,
+      "step": 3966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5620089498131462e-05,
+      "loss": 1.0879,
+      "step": 3967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5617995436493697e-05,
+      "loss": 0.8194,
+      "step": 3968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5615900960007433e-05,
+      "loss": 1.3542,
+      "step": 3969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5613806068754503e-05,
+      "loss": 1.1457,
+      "step": 3970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5611710762816752e-05,
+      "loss": 0.82,
+      "step": 3971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5609615042276045e-05,
+      "loss": 1.376,
+      "step": 3972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5607518907214266e-05,
+      "loss": 0.5065,
+      "step": 3973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5605422357713307e-05,
+      "loss": 0.789,
+      "step": 3974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5603325393855085e-05,
+      "loss": 1.3627,
+      "step": 3975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5601228015721526e-05,
+      "loss": 0.7102,
+      "step": 3976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.559913022339458e-05,
+      "loss": 1.1176,
+      "step": 3977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.55970320169562e-05,
+      "loss": 1.3264,
+      "step": 3978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.559493339648837e-05,
+      "loss": 0.7847,
+      "step": 3979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.559283436207308e-05,
+      "loss": 0.9862,
+      "step": 3980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5590734913792345e-05,
+      "loss": 0.5704,
+      "step": 3981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5588635051728187e-05,
+      "loss": 0.5767,
+      "step": 3982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5586534775962654e-05,
+      "loss": 0.7456,
+      "step": 3983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.558443408657779e-05,
+      "loss": 1.0687,
+      "step": 3984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5582332983655687e-05,
+      "loss": 0.117,
+      "step": 3985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5580231467278425e-05,
+      "loss": 0.203,
+      "step": 3986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.557812953752812e-05,
+      "loss": 0.4476,
+      "step": 3987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5576027194486884e-05,
+      "loss": 1.8096,
+      "step": 3988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.557392443823686e-05,
+      "loss": 0.1539,
+      "step": 3989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5571821268860206e-05,
+      "loss": 0.9434,
+      "step": 3990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5569717686439095e-05,
+      "loss": 0.9093,
+      "step": 3991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5567613691055714e-05,
+      "loss": 1.118,
+      "step": 3992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5565509282792262e-05,
+      "loss": 0.9055,
+      "step": 3993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5563404461730955e-05,
+      "loss": 0.6029,
+      "step": 3994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5561299227954045e-05,
+      "loss": 1.0229,
+      "step": 3995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.555919358154377e-05,
+      "loss": 0.6309,
+      "step": 3996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5557087522582412e-05,
+      "loss": 0.3861,
+      "step": 3997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5554981051152235e-05,
+      "loss": 0.5076,
+      "step": 3998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5552874167335556e-05,
+      "loss": 0.226,
+      "step": 3999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5550766871214687e-05,
+      "loss": 0.4597,
+      "step": 4000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5548659162871963e-05,
+      "loss": 0.7595,
+      "step": 4001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5546551042389723e-05,
+      "loss": 0.2489,
+      "step": 4002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5544442509850348e-05,
+      "loss": 1.2019,
+      "step": 4003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5542333565336204e-05,
+      "loss": 0.4425,
+      "step": 4004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5540224208929698e-05,
+      "loss": 0.4833,
+      "step": 4005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.553811444071324e-05,
+      "loss": 0.5388,
+      "step": 4006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5536004260769252e-05,
+      "loss": 1.0938,
+      "step": 4007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5533893669180193e-05,
+      "loss": 0.4147,
+      "step": 4008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.553178266602852e-05,
+      "loss": 1.2275,
+      "step": 4009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5529671251396703e-05,
+      "loss": 0.4183,
+      "step": 4010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5527559425367245e-05,
+      "loss": 0.3586,
+      "step": 4011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5525447188022646e-05,
+      "loss": 0.5126,
+      "step": 4012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5523334539445445e-05,
+      "loss": 1.2881,
+      "step": 4013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5521221479718173e-05,
+      "loss": 1.4558,
+      "step": 4014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.551910800892339e-05,
+      "loss": 0.7199,
+      "step": 4015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5516994127143674e-05,
+      "loss": 0.1921,
+      "step": 4016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5514879834461607e-05,
+      "loss": 0.8104,
+      "step": 4017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.551276513095981e-05,
+      "loss": 0.2394,
+      "step": 4018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.551065001672089e-05,
+      "loss": 1.1346,
+      "step": 4019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.550853449182749e-05,
+      "loss": 1.4769,
+      "step": 4020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5506418556362264e-05,
+      "loss": 1.4031,
+      "step": 4021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5504302210407888e-05,
+      "loss": 0.7485,
+      "step": 4022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.550218545404704e-05,
+      "loss": 0.9441,
+      "step": 4023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5500068287362432e-05,
+      "loss": 1.8073,
+      "step": 4024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5497950710436774e-05,
+      "loss": 0.5642,
+      "step": 4025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.54958327233528e-05,
+      "loss": 1.1114,
+      "step": 4026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5493714326193263e-05,
+      "loss": 0.8425,
+      "step": 4027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5491595519040934e-05,
+      "loss": 0.7624,
+      "step": 4028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.548947630197859e-05,
+      "loss": 0.2589,
+      "step": 4029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.548735667508903e-05,
+      "loss": 1.023,
+      "step": 4030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5485236638455065e-05,
+      "loss": 1.4173,
+      "step": 4031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5483116192159535e-05,
+      "loss": 0.5464,
+      "step": 4032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.548099533628528e-05,
+      "loss": 0.3611,
+      "step": 4033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5478874070915163e-05,
+      "loss": 0.7424,
+      "step": 4034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5476752396132063e-05,
+      "loss": 0.1802,
+      "step": 4035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5474630312018878e-05,
+      "loss": 1.1798,
+      "step": 4036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.547250781865851e-05,
+      "loss": 0.587,
+      "step": 4037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5470384916133894e-05,
+      "loss": 0.1941,
+      "step": 4038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.546826160452797e-05,
+      "loss": 1.3269,
+      "step": 4039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5466137883923697e-05,
+      "loss": 0.3574,
+      "step": 4040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5464013754404044e-05,
+      "loss": 0.2042,
+      "step": 4041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.546188921605201e-05,
+      "loss": 0.2584,
+      "step": 4042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5459764268950594e-05,
+      "loss": 0.3215,
+      "step": 4043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.545763891318282e-05,
+      "loss": 0.9372,
+      "step": 4044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5455513148831735e-05,
+      "loss": 0.1496,
+      "step": 4045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.545338697598038e-05,
+      "loss": 1.1497,
+      "step": 4046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5451260394711833e-05,
+      "loss": 0.5213,
+      "step": 4047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5449133405109174e-05,
+      "loss": 0.8021,
+      "step": 4048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5447006007255517e-05,
+      "loss": 0.7196,
+      "step": 4049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.544487820123397e-05,
+      "loss": 0.3007,
+      "step": 4050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5442749987127667e-05,
+      "loss": 0.8339,
+      "step": 4051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5440621365019768e-05,
+      "loss": 1.5511,
+      "step": 4052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5438492334993425e-05,
+      "loss": 0.2837,
+      "step": 4053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5436362897131828e-05,
+      "loss": 0.7517,
+      "step": 4054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5434233051518175e-05,
+      "loss": 0.7747,
+      "step": 4055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5432102798235683e-05,
+      "loss": 0.9567,
+      "step": 4056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.542997213736757e-05,
+      "loss": 1.6114,
+      "step": 4057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.542784106899709e-05,
+      "loss": 0.4672,
+      "step": 4058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5425709593207504e-05,
+      "loss": 0.2553,
+      "step": 4059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5423577710082087e-05,
+      "loss": 0.6611,
+      "step": 4060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5421445419704134e-05,
+      "loss": 1.1563,
+      "step": 4061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5419312722156955e-05,
+      "loss": 1.7398,
+      "step": 4062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5417179617523882e-05,
+      "loss": 0.2697,
+      "step": 4063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5415046105888238e-05,
+      "loss": 0.4018,
+      "step": 4064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5412912187333396e-05,
+      "loss": 0.2713,
+      "step": 4065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5410777861942717e-05,
+      "loss": 0.0861,
+      "step": 4066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5408643129799604e-05,
+      "loss": 1.3947,
+      "step": 4067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5406507990987447e-05,
+      "loss": 1.0968,
+      "step": 4068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5404372445589675e-05,
+      "loss": 2.0494,
+      "step": 4069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5402236493689725e-05,
+      "loss": 1.1778,
+      "step": 4070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5400100135371048e-05,
+      "loss": 0.9542,
+      "step": 4071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.539796337071711e-05,
+      "loss": 0.4934,
+      "step": 4072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5395826199811392e-05,
+      "loss": 1.0841,
+      "step": 4073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5393688622737407e-05,
+      "loss": 0.4909,
+      "step": 4074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5391550639578657e-05,
+      "loss": 0.8345,
+      "step": 4075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.538941225041868e-05,
+      "loss": 0.899,
+      "step": 4076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.538727345534102e-05,
+      "loss": 0.8239,
+      "step": 4077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5385134254429248e-05,
+      "loss": 0.2344,
+      "step": 4078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5382994647766933e-05,
+      "loss": 0.8111,
+      "step": 4079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.538085463543768e-05,
+      "loss": 0.1997,
+      "step": 4080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.537871421752509e-05,
+      "loss": 0.6883,
+      "step": 4081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.53765733941128e-05,
+      "loss": 0.605,
+      "step": 4082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5374432165284442e-05,
+      "loss": 0.911,
+      "step": 4083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5372290531123685e-05,
+      "loss": 0.57,
+      "step": 4084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.53701484917142e-05,
+      "loss": 0.3563,
+      "step": 4085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.536800604713967e-05,
+      "loss": 0.9486,
+      "step": 4086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.536586319748381e-05,
+      "loss": 1.4014,
+      "step": 4087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.536371994283034e-05,
+      "loss": 0.6974,
+      "step": 4088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5361576283262995e-05,
+      "loss": 0.3807,
+      "step": 4089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.535943221886553e-05,
+      "loss": 1.4021,
+      "step": 4090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.535728774972171e-05,
+      "loss": 0.1507,
+      "step": 4091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5355142875915326e-05,
+      "loss": 0.6577,
+      "step": 4092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5352997597530178e-05,
+      "loss": 1.1193,
+      "step": 4093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.535085191465008e-05,
+      "loss": 1.2198,
+      "step": 4094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5348705827358866e-05,
+      "loss": 0.1071,
+      "step": 4095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5346559335740387e-05,
+      "loss": 0.7124,
+      "step": 4096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5344412439878504e-05,
+      "loss": 0.3724,
+      "step": 4097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5342265139857097e-05,
+      "loss": 0.3551,
+      "step": 4098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5340117435760065e-05,
+      "loss": 0.1969,
+      "step": 4099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.533796932767131e-05,
+      "loss": 0.546,
+      "step": 4100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.533582081567477e-05,
+      "loss": 1.7194,
+      "step": 4101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5333671899854388e-05,
+      "loss": 0.2952,
+      "step": 4102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5331522580294113e-05,
+      "loss": 0.7254,
+      "step": 4103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.532937285707793e-05,
+      "loss": 0.2185,
+      "step": 4104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5327222730289822e-05,
+      "loss": 0.1202,
+      "step": 4105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5325072200013796e-05,
+      "loss": 0.8493,
+      "step": 4106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5322921266333883e-05,
+      "loss": 1.1451,
+      "step": 4107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5320769929334106e-05,
+      "loss": 2.4422,
+      "step": 4108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5318618189098528e-05,
+      "loss": 0.7251,
+      "step": 4109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5316466045711218e-05,
+      "loss": 0.4111,
+      "step": 4110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5314313499256257e-05,
+      "loss": 0.5003,
+      "step": 4111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.531216054981775e-05,
+      "loss": 1.1085,
+      "step": 4112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5310007197479808e-05,
+      "loss": 0.27,
+      "step": 4113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5307853442326575e-05,
+      "loss": 0.5665,
+      "step": 4114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5305699284442183e-05,
+      "loss": 1.13,
+      "step": 4115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5303544723910802e-05,
+      "loss": 1.3734,
+      "step": 4116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5301389760816612e-05,
+      "loss": 2.1567,
+      "step": 4117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5299234395243815e-05,
+      "loss": 0.1614,
+      "step": 4118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.529707862727661e-05,
+      "loss": 0.4862,
+      "step": 4119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5294922456999233e-05,
+      "loss": 1.0876,
+      "step": 4120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5292765884495917e-05,
+      "loss": 1.0001,
+      "step": 4121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5290608909850927e-05,
+      "loss": 0.6534,
+      "step": 4122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5288451533148537e-05,
+      "loss": 1.7297,
+      "step": 4123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.528629375447303e-05,
+      "loss": 1.1821,
+      "step": 4124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5284135573908718e-05,
+      "loss": 0.5885,
+      "step": 4125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5281976991539916e-05,
+      "loss": 0.9658,
+      "step": 4126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.527981800745097e-05,
+      "loss": 0.8538,
+      "step": 4127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5277658621726217e-05,
+      "loss": 0.3396,
+      "step": 4128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5275498834450042e-05,
+      "loss": 0.6859,
+      "step": 4129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5273338645706814e-05,
+      "loss": 0.7096,
+      "step": 4130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5271178055580946e-05,
+      "loss": 0.8998,
+      "step": 4131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5269017064156838e-05,
+      "loss": 0.4152,
+      "step": 4132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5266855671518932e-05,
+      "loss": 0.1802,
+      "step": 4133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.526469387775167e-05,
+      "loss": 0.5666,
+      "step": 4134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.526253168293951e-05,
+      "loss": 1.0489,
+      "step": 4135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.526036908716694e-05,
+      "loss": 0.1668,
+      "step": 4136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5258206090518448e-05,
+      "loss": 1.0314,
+      "step": 4137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.525604269307854e-05,
+      "loss": 0.9745,
+      "step": 4138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.525387889493174e-05,
+      "loss": 0.4242,
+      "step": 4139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5251714696162596e-05,
+      "loss": 0.1185,
+      "step": 4140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5249550096855656e-05,
+      "loss": 0.5256,
+      "step": 4141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5247385097095496e-05,
+      "loss": 0.9422,
+      "step": 4142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5245219696966704e-05,
+      "loss": 0.947,
+      "step": 4143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.524305389655388e-05,
+      "loss": 1.362,
+      "step": 4144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5240887695941638e-05,
+      "loss": 0.5554,
+      "step": 4145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5238721095214625e-05,
+      "loss": 0.836,
+      "step": 4146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5236554094457478e-05,
+      "loss": 0.5156,
+      "step": 4147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.523438669375487e-05,
+      "loss": 0.6068,
+      "step": 4148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5232218893191477e-05,
+      "loss": 0.2748,
+      "step": 4149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5230050692852003e-05,
+      "loss": 0.5288,
+      "step": 4150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5227882092821157e-05,
+      "loss": 1.4394,
+      "step": 4151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5225713093183664e-05,
+      "loss": 0.283,
+      "step": 4152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5223543694024263e-05,
+      "loss": 0.799,
+      "step": 4153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5221373895427726e-05,
+      "loss": 1.4818,
+      "step": 4154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.521920369747882e-05,
+      "loss": 1.5706,
+      "step": 4155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5217033100262337e-05,
+      "loss": 1.3055,
+      "step": 4156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.521486210386308e-05,
+      "loss": 0.3382,
+      "step": 4157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5212690708365877e-05,
+      "loss": 0.2754,
+      "step": 4158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5210518913855557e-05,
+      "loss": 0.2383,
+      "step": 4159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5208346720416978e-05,
+      "loss": 0.3728,
+      "step": 4160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5206174128135006e-05,
+      "loss": 0.4759,
+      "step": 4161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5204001137094526e-05,
+      "loss": 0.5422,
+      "step": 4162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.520182774738044e-05,
+      "loss": 0.3112,
+      "step": 4163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5199653959077656e-05,
+      "loss": 0.9766,
+      "step": 4164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5197479772271115e-05,
+      "loss": 1.4459,
+      "step": 4165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5195305187045752e-05,
+      "loss": 0.8456,
+      "step": 4166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5193130203486533e-05,
+      "loss": 1.5801,
+      "step": 4167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.519095482167844e-05,
+      "loss": 0.9173,
+      "step": 4168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5188779041706463e-05,
+      "loss": 0.3208,
+      "step": 4169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5186602863655602e-05,
+      "loss": 0.5833,
+      "step": 4170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5184426287610895e-05,
+      "loss": 1.1527,
+      "step": 4171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.518224931365737e-05,
+      "loss": 1.5185,
+      "step": 4172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5180071941880095e-05,
+      "loss": 1.5357,
+      "step": 4173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.517789417236413e-05,
+      "loss": 0.3735,
+      "step": 4174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.517571600519456e-05,
+      "loss": 1.8606,
+      "step": 4175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5173537440456492e-05,
+      "loss": 0.3359,
+      "step": 4176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5171358478235042e-05,
+      "loss": 0.8803,
+      "step": 4177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.516917911861534e-05,
+      "loss": 1.6772,
+      "step": 4178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5166999361682542e-05,
+      "loss": 1.689,
+      "step": 4179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5164819207521807e-05,
+      "loss": 1.0867,
+      "step": 4180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.516263865621831e-05,
+      "loss": 0.6361,
+      "step": 4181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5160457707857248e-05,
+      "loss": 2.0056,
+      "step": 4182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.515827636252384e-05,
+      "loss": 1.7912,
+      "step": 4183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.51560946203033e-05,
+      "loss": 0.4982,
+      "step": 4184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5153912481280878e-05,
+      "loss": 1.0964,
+      "step": 4185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.515172994554182e-05,
+      "loss": 0.1294,
+      "step": 4186
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5149547013171412e-05,
+      "loss": 0.9678,
+      "step": 4187
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5147363684254934e-05,
+      "loss": 0.5477,
+      "step": 4188
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.514517995887769e-05,
+      "loss": 0.3051,
+      "step": 4189
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5142995837125e-05,
+      "loss": 0.6219,
+      "step": 4190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.51408113190822e-05,
+      "loss": 0.509,
+      "step": 4191
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.513862640483463e-05,
+      "loss": 0.5634,
+      "step": 4192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5136441094467675e-05,
+      "loss": 0.9063,
+      "step": 4193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.51342553880667e-05,
+      "loss": 0.4502,
+      "step": 4194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.51320692857171e-05,
+      "loss": 0.8328,
+      "step": 4195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5129882787504294e-05,
+      "loss": 1.5063,
+      "step": 4196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.512769589351371e-05,
+      "loss": 0.5373,
+      "step": 4197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5125508603830785e-05,
+      "loss": 0.8524,
+      "step": 4198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5123320918540978e-05,
+      "loss": 0.6828,
+      "step": 4199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.512113283772977e-05,
+      "loss": 0.3656,
+      "step": 4200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.511894436148264e-05,
+      "loss": 0.5634,
+      "step": 4201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.51167554898851e-05,
+      "loss": 1.0816,
+      "step": 4202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.511456622302267e-05,
+      "loss": 0.3967,
+      "step": 4203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5112376560980876e-05,
+      "loss": 0.1792,
+      "step": 4204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5110186503845274e-05,
+      "loss": 0.4682,
+      "step": 4205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5107996051701434e-05,
+      "loss": 0.2595,
+      "step": 4206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5105805204634933e-05,
+      "loss": 0.7422,
+      "step": 4207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5103613962731377e-05,
+      "loss": 1.4294,
+      "step": 4208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5101422326076365e-05,
+      "loss": 0.1869,
+      "step": 4209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5099230294755533e-05,
+      "loss": 1.4245,
+      "step": 4210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.509703786885452e-05,
+      "loss": 0.4471,
+      "step": 4211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5094845048458994e-05,
+      "loss": 1.1786,
+      "step": 4212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.509265183365462e-05,
+      "loss": 1.6512,
+      "step": 4213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.509045822452709e-05,
+      "loss": 0.1885,
+      "step": 4214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5088264221162105e-05,
+      "loss": 1.007,
+      "step": 4215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5086069823645394e-05,
+      "loss": 0.7261,
+      "step": 4216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.508387503206269e-05,
+      "loss": 0.4941,
+      "step": 4217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5081679846499743e-05,
+      "loss": 0.4982,
+      "step": 4218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5079484267042318e-05,
+      "loss": 1.3774,
+      "step": 4219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.50772882937762e-05,
+      "loss": 0.5097,
+      "step": 4220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5075091926787183e-05,
+      "loss": 0.5727,
+      "step": 4221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.507289516616108e-05,
+      "loss": 0.434,
+      "step": 4222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5070698011983722e-05,
+      "loss": 0.2093,
+      "step": 4223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5068500464340953e-05,
+      "loss": 0.6066,
+      "step": 4224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5066302523318635e-05,
+      "loss": 0.6422,
+      "step": 4225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.506410418900263e-05,
+      "loss": 0.2608,
+      "step": 4226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5061905461478835e-05,
+      "loss": 0.5517,
+      "step": 4227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.505970634083316e-05,
+      "loss": 0.7929,
+      "step": 4228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5057506827151517e-05,
+      "loss": 0.8516,
+      "step": 4229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5055306920519845e-05,
+      "loss": 1.2539,
+      "step": 4230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5053106621024095e-05,
+      "loss": 1.3878,
+      "step": 4231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5050905928750235e-05,
+      "loss": 0.7735,
+      "step": 4232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5048704843784244e-05,
+      "loss": 1.3939,
+      "step": 4233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5046503366212123e-05,
+      "loss": 0.5013,
+      "step": 4234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5044301496119882e-05,
+      "loss": 1.0276,
+      "step": 4235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5042099233593546e-05,
+      "loss": 0.8194,
+      "step": 4236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.503989657871916e-05,
+      "loss": 0.9191,
+      "step": 4237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5037693531582787e-05,
+      "loss": 1.1808,
+      "step": 4238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5035490092270494e-05,
+      "loss": 1.0529,
+      "step": 4239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.503328626086838e-05,
+      "loss": 0.8463,
+      "step": 4240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.503108203746253e-05,
+      "loss": 0.3558,
+      "step": 4241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5028877422139087e-05,
+      "loss": 1.2356,
+      "step": 4242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5026672414984167e-05,
+      "loss": 0.7107,
+      "step": 4243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5024467016083936e-05,
+      "loss": 0.6746,
+      "step": 4244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5022261225524548e-05,
+      "loss": 0.8169,
+      "step": 4245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5020055043392188e-05,
+      "loss": 0.5464,
+      "step": 4246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5017848469773056e-05,
+      "loss": 0.9614,
+      "step": 4247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5015641504753357e-05,
+      "loss": 0.1115,
+      "step": 4248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5013434148419322e-05,
+      "loss": 1.2619,
+      "step": 4249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5011226400857194e-05,
+      "loss": 0.4445,
+      "step": 4250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.500901826215322e-05,
+      "loss": 1.4456,
+      "step": 4251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5006809732393686e-05,
+      "loss": 1.3795,
+      "step": 4252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5004600811664878e-05,
+      "loss": 0.5873,
+      "step": 4253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5002391500053095e-05,
+      "loss": 1.1942,
+      "step": 4254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5000181797644658e-05,
+      "loss": 0.0643,
+      "step": 4255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.4997971704525896e-05,
+      "loss": 0.3357,
+      "step": 4256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.4995761220783158e-05,
+      "loss": 0.3039,
+      "step": 4257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.499355034650282e-05,
+      "loss": 1.1055,
+      "step": 4258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.499133908177125e-05,
+      "loss": 0.409,
+      "step": 4259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4989127426674844e-05,
+      "loss": 0.1832,
+      "step": 4260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.498691538130002e-05,
+      "loss": 0.1177,
+      "step": 4261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.498470294573319e-05,
+      "loss": 2.0099,
+      "step": 4262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.498249012006081e-05,
+      "loss": 0.198,
+      "step": 4263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4980276904369322e-05,
+      "loss": 0.6936,
+      "step": 4264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4978063298745206e-05,
+      "loss": 0.9044,
+      "step": 4265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4975849303274942e-05,
+      "loss": 0.9675,
+      "step": 4266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.497363491804504e-05,
+      "loss": 1.3952,
+      "step": 4267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4971420143142013e-05,
+      "loss": 0.0694,
+      "step": 4268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4969204978652386e-05,
+      "loss": 0.7789,
+      "step": 4269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4966989424662712e-05,
+      "loss": 0.824,
+      "step": 4270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4964773481259558e-05,
+      "loss": 0.4014,
+      "step": 4271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.496255714852949e-05,
+      "loss": 0.7957,
+      "step": 4272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4960340426559112e-05,
+      "loss": 1.3232,
+      "step": 4273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.495812331543503e-05,
+      "loss": 0.7038,
+      "step": 4274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.495590581524386e-05,
+      "loss": 0.2883,
+      "step": 4275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.495368792607225e-05,
+      "loss": 0.2999,
+      "step": 4276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4951469648006844e-05,
+      "loss": 0.8074,
+      "step": 4277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.494925098113432e-05,
+      "loss": 0.6545,
+      "step": 4278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.494703192554136e-05,
+      "loss": 0.6171,
+      "step": 4279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4944812481314656e-05,
+      "loss": 0.8568,
+      "step": 4280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4942592648540926e-05,
+      "loss": 0.8791,
+      "step": 4281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4940372427306905e-05,
+      "loss": 0.6827,
+      "step": 4282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4938151817699336e-05,
+      "loss": 1.0466,
+      "step": 4283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4935930819804973e-05,
+      "loss": 0.6419,
+      "step": 4284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4933709433710596e-05,
+      "loss": 1.6247,
+      "step": 4285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.493148765950299e-05,
+      "loss": 1.0512,
+      "step": 4286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4929265497268974e-05,
+      "loss": 1.5437,
+      "step": 4287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4927042947095353e-05,
+      "loss": 0.2111,
+      "step": 4288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.492482000906897e-05,
+      "loss": 1.0503,
+      "step": 4289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.492259668327667e-05,
+      "loss": 1.1322,
+      "step": 4290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4920372969805334e-05,
+      "loss": 0.9211,
+      "step": 4291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4918148868741825e-05,
+      "loss": 0.0859,
+      "step": 4292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.491592438017305e-05,
+      "loss": 0.2428,
+      "step": 4293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4913699504185915e-05,
+      "loss": 0.814,
+      "step": 4294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4911474240867357e-05,
+      "loss": 0.6581,
+      "step": 4295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4909248590304305e-05,
+      "loss": 0.5437,
+      "step": 4296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4907022552583723e-05,
+      "loss": 0.137,
+      "step": 4297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.490479612779258e-05,
+      "loss": 0.5119,
+      "step": 4298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4902569316017865e-05,
+      "loss": 0.4043,
+      "step": 4299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4900342117346577e-05,
+      "loss": 1.0152,
+      "step": 4300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4898114531865736e-05,
+      "loss": 1.4389,
+      "step": 4301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4895886559662378e-05,
+      "loss": 0.7014,
+      "step": 4302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4893658200823547e-05,
+      "loss": 0.154,
+      "step": 4303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4891429455436303e-05,
+      "loss": 1.8962,
+      "step": 4304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.488920032358773e-05,
+      "loss": 0.2459,
+      "step": 4305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4886970805364915e-05,
+      "loss": 0.3588,
+      "step": 4306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4884740900854964e-05,
+      "loss": 1.2514,
+      "step": 4307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.488251061014501e-05,
+      "loss": 1.6338,
+      "step": 4308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4880279933322188e-05,
+      "loss": 1.5971,
+      "step": 4309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4878048870473645e-05,
+      "loss": 0.1693,
+      "step": 4310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4875817421686557e-05,
+      "loss": 0.0963,
+      "step": 4311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4873585587048098e-05,
+      "loss": 0.5144,
+      "step": 4312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4871353366645477e-05,
+      "loss": 0.2339,
+      "step": 4313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48691207605659e-05,
+      "loss": 0.4551,
+      "step": 4314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48668877688966e-05,
+      "loss": 1.0188,
+      "step": 4315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4864654391724813e-05,
+      "loss": 0.1531,
+      "step": 4316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4862420629137813e-05,
+      "loss": 0.7901,
+      "step": 4317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4860186481222865e-05,
+      "loss": 0.645,
+      "step": 4318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.485795194806725e-05,
+      "loss": 0.6291,
+      "step": 4319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4855717029758283e-05,
+      "loss": 1.0162,
+      "step": 4320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.485348172638328e-05,
+      "loss": 1.0832,
+      "step": 4321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4851246038029576e-05,
+      "loss": 0.4158,
+      "step": 4322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4849009964784513e-05,
+      "loss": 1.0472,
+      "step": 4323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4846773506735465e-05,
+      "loss": 0.9973,
+      "step": 4324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4844536663969806e-05,
+      "loss": 0.2676,
+      "step": 4325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.484229943657493e-05,
+      "loss": 0.2884,
+      "step": 4326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4840061824638253e-05,
+      "loss": 0.3982,
+      "step": 4327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4837823828247187e-05,
+      "loss": 0.238,
+      "step": 4328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.483558544748918e-05,
+      "loss": 1.2067,
+      "step": 4329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.483334668245168e-05,
+      "loss": 1.4291,
+      "step": 4330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4831107533222164e-05,
+      "loss": 0.9027,
+      "step": 4331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.482886799988811e-05,
+      "loss": 0.8419,
+      "step": 4332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4826628082537025e-05,
+      "loss": 1.4168,
+      "step": 4333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4824387781256414e-05,
+      "loss": 1.1505,
+      "step": 4334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.482214709613381e-05,
+      "loss": 0.9416,
+      "step": 4335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4819906027256757e-05,
+      "loss": 0.6907,
+      "step": 4336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.481766457471282e-05,
+      "loss": 1.158,
+      "step": 4337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4815422738589562e-05,
+      "loss": 1.0887,
+      "step": 4338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.481318051897458e-05,
+      "loss": 1.1427,
+      "step": 4339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.481093791595548e-05,
+      "loss": 0.7595,
+      "step": 4340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4808694929619875e-05,
+      "loss": 1.2211,
+      "step": 4341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4806451560055403e-05,
+      "loss": 0.6111,
+      "step": 4342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4804207807349712e-05,
+      "loss": 0.8131,
+      "step": 4343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4801963671590465e-05,
+      "loss": 1.0127,
+      "step": 4344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4799719152865345e-05,
+      "loss": 0.6427,
+      "step": 4345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.479747425126204e-05,
+      "loss": 0.2241,
+      "step": 4346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4795228966868268e-05,
+      "loss": 0.2144,
+      "step": 4347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.479298329977174e-05,
+      "loss": 0.1942,
+      "step": 4348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.479073725006021e-05,
+      "loss": 0.2511,
+      "step": 4349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.478849081782142e-05,
+      "loss": 0.5857,
+      "step": 4350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4786244003143146e-05,
+      "loss": 1.2789,
+      "step": 4351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4783996806113166e-05,
+      "loss": 1.2075,
+      "step": 4352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4781749226819287e-05,
+      "loss": 1.2822,
+      "step": 4353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.477950126534931e-05,
+      "loss": 0.4148,
+      "step": 4354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.477725292179108e-05,
+      "loss": 1.0641,
+      "step": 4355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4775004196232425e-05,
+      "loss": 1.2837,
+      "step": 4356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4772755088761207e-05,
+      "loss": 0.0903,
+      "step": 4357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4770505599465308e-05,
+      "loss": 0.9732,
+      "step": 4358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.476825572843261e-05,
+      "loss": 0.6024,
+      "step": 4359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4766005475751013e-05,
+      "loss": 1.4519,
+      "step": 4360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4763754841508445e-05,
+      "loss": 0.1239,
+      "step": 4361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.476150382579283e-05,
+      "loss": 0.5402,
+      "step": 4362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4759252428692113e-05,
+      "loss": 0.8511,
+      "step": 4363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4757000650294272e-05,
+      "loss": 0.3299,
+      "step": 4364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.475474849068727e-05,
+      "loss": 0.1267,
+      "step": 4365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4752495949959102e-05,
+      "loss": 1.2515,
+      "step": 4366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.475024302819778e-05,
+      "loss": 0.3721,
+      "step": 4367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.474798972549133e-05,
+      "loss": 0.2532,
+      "step": 4368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.474573604192778e-05,
+      "loss": 0.431,
+      "step": 4369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4743481977595184e-05,
+      "loss": 0.1951,
+      "step": 4370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.474122753258161e-05,
+      "loss": 1.4582,
+      "step": 4371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4738972706975146e-05,
+      "loss": 0.2062,
+      "step": 4372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4736717500863883e-05,
+      "loss": 1.5245,
+      "step": 4373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4734461914335934e-05,
+      "loss": 0.6344,
+      "step": 4374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4732205947479424e-05,
+      "loss": 0.557,
+      "step": 4375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4729949600382495e-05,
+      "loss": 1.0774,
+      "step": 4376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4727692873133304e-05,
+      "loss": 0.9643,
+      "step": 4377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4725435765820016e-05,
+      "loss": 0.307,
+      "step": 4378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4723178278530827e-05,
+      "loss": 0.8452,
+      "step": 4379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4720920411353935e-05,
+      "loss": 0.2504,
+      "step": 4380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4718662164377552e-05,
+      "loss": 1.0015,
+      "step": 4381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.471640353768991e-05,
+      "loss": 1.1165,
+      "step": 4382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4714144531379253e-05,
+      "loss": 0.894,
+      "step": 4383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4711885145533845e-05,
+      "loss": 0.3394,
+      "step": 4384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.470962538024196e-05,
+      "loss": 1.5701,
+      "step": 4385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.470736523559188e-05,
+      "loss": 0.7479,
+      "step": 4386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.470510471167192e-05,
+      "loss": 0.5381,
+      "step": 4387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4702843808570395e-05,
+      "loss": 1.8027,
+      "step": 4388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.470058252637564e-05,
+      "loss": 0.4892,
+      "step": 4389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4698320865176005e-05,
+      "loss": 0.9348,
+      "step": 4390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.469605882505985e-05,
+      "loss": 0.1395,
+      "step": 4391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4693796406115554e-05,
+      "loss": 1.2798,
+      "step": 4392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4691533608431514e-05,
+      "loss": 0.4925,
+      "step": 4393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.468927043209614e-05,
+      "loss": 0.8766,
+      "step": 4394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.468700687719785e-05,
+      "loss": 1.0624,
+      "step": 4395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4684742943825082e-05,
+      "loss": 0.5233,
+      "step": 4396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4682478632066288e-05,
+      "loss": 0.0775,
+      "step": 4397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4680213942009942e-05,
+      "loss": 1.2124,
+      "step": 4398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.467794887374452e-05,
+      "loss": 0.6094,
+      "step": 4399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4675683427358515e-05,
+      "loss": 0.6019,
+      "step": 4400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.467341760294045e-05,
+      "loss": 1.3161,
+      "step": 4401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4671151400578846e-05,
+      "loss": 1.3338,
+      "step": 4402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4668884820362244e-05,
+      "loss": 1.5853,
+      "step": 4403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4666617862379197e-05,
+      "loss": 1.027,
+      "step": 4404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.466435052671828e-05,
+      "loss": 1.0148,
+      "step": 4405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4662082813468077e-05,
+      "loss": 0.4088,
+      "step": 4406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4659814722717193e-05,
+      "loss": 0.5981,
+      "step": 4407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4657546254554236e-05,
+      "loss": 0.2684,
+      "step": 4408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4655277409067838e-05,
+      "loss": 1.6117,
+      "step": 4409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4653008186346643e-05,
+      "loss": 0.902,
+      "step": 4410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4650738586479312e-05,
+      "loss": 1.3974,
+      "step": 4411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4648468609554516e-05,
+      "loss": 0.3215,
+      "step": 4412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4646198255660953e-05,
+      "loss": 0.3862,
+      "step": 4413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4643927524887314e-05,
+      "loss": 0.0951,
+      "step": 4414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4641656417322323e-05,
+      "loss": 1.4546,
+      "step": 4415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4639384933054713e-05,
+      "loss": 0.0415,
+      "step": 4416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.463711307217323e-05,
+      "loss": 0.0972,
+      "step": 4417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.463484083476664e-05,
+      "loss": 0.2579,
+      "step": 4418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4632568220923717e-05,
+      "loss": 0.2701,
+      "step": 4419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4630295230733252e-05,
+      "loss": 0.6105,
+      "step": 4420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4628021864284053e-05,
+      "loss": 1.212,
+      "step": 4421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4625748121664938e-05,
+      "loss": 0.6181,
+      "step": 4422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4623474002964752e-05,
+      "loss": 0.3883,
+      "step": 4423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4621199508272337e-05,
+      "loss": 1.2319,
+      "step": 4424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.461892463767656e-05,
+      "loss": 0.7366,
+      "step": 4425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.46166493912663e-05,
+      "loss": 1.7208,
+      "step": 4426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4614373769130455e-05,
+      "loss": 0.5839,
+      "step": 4427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4612097771357932e-05,
+      "loss": 1.2804,
+      "step": 4428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.460982139803765e-05,
+      "loss": 0.9474,
+      "step": 4429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4607544649258562e-05,
+      "loss": 0.6777,
+      "step": 4430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4605267525109608e-05,
+      "loss": 0.9111,
+      "step": 4431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.460299002567976e-05,
+      "loss": 1.1202,
+      "step": 4432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4600712151058002e-05,
+      "loss": 1.2847,
+      "step": 4433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4598433901333324e-05,
+      "loss": 0.9038,
+      "step": 4434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.459615527659475e-05,
+      "loss": 0.3224,
+      "step": 4435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4593876276931303e-05,
+      "loss": 1.0306,
+      "step": 4436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4591596902432018e-05,
+      "loss": 0.8498,
+      "step": 4437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4589317153185958e-05,
+      "loss": 0.5991,
+      "step": 4438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.458703702928219e-05,
+      "loss": 0.6349,
+      "step": 4439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4584756530809797e-05,
+      "loss": 0.8914,
+      "step": 4440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4582475657857878e-05,
+      "loss": 1.2374,
+      "step": 4441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4580194410515557e-05,
+      "loss": 0.4355,
+      "step": 4442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.457791278887196e-05,
+      "loss": 0.9606,
+      "step": 4443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.457563079301622e-05,
+      "loss": 0.8521,
+      "step": 4444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4573348423037506e-05,
+      "loss": 0.5767,
+      "step": 4445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.457106567902499e-05,
+      "loss": 0.7622,
+      "step": 4446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4568782561067853e-05,
+      "loss": 0.6361,
+      "step": 4447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.45664990692553e-05,
+      "loss": 0.6535,
+      "step": 4448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4564215203676555e-05,
+      "loss": 0.5553,
+      "step": 4449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.456193096442084e-05,
+      "loss": 1.1988,
+      "step": 4450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4559646351577407e-05,
+      "loss": 0.0948,
+      "step": 4451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4557361365235514e-05,
+      "loss": 0.1257,
+      "step": 4452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4555076005484432e-05,
+      "loss": 1.3459,
+      "step": 4453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4552790272413462e-05,
+      "loss": 0.2405,
+      "step": 4454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.45505041661119e-05,
+      "loss": 0.578,
+      "step": 4455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4548217686669064e-05,
+      "loss": 0.3876,
+      "step": 4456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4545930834174293e-05,
+      "loss": 0.3207,
+      "step": 4457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4543643608716925e-05,
+      "loss": 0.7244,
+      "step": 4458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4541356010386333e-05,
+      "loss": 0.578,
+      "step": 4459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4539068039271893e-05,
+      "loss": 2.1248,
+      "step": 4460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4536779695462992e-05,
+      "loss": 0.1067,
+      "step": 4461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.453449097904904e-05,
+      "loss": 1.0011,
+      "step": 4462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4532201890119458e-05,
+      "loss": 0.1819,
+      "step": 4463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4529912428763677e-05,
+      "loss": 0.62,
+      "step": 4464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.452762259507115e-05,
+      "loss": 0.6398,
+      "step": 4465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4525332389131345e-05,
+      "loss": 1.5681,
+      "step": 4466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4523041811033736e-05,
+      "loss": 2.0153,
+      "step": 4467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4520750860867818e-05,
+      "loss": 0.5829,
+      "step": 4468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.45184595387231e-05,
+      "loss": 0.4008,
+      "step": 4469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4516167844689106e-05,
+      "loss": 0.2739,
+      "step": 4470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4513875778855368e-05,
+      "loss": 0.546,
+      "step": 4471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.451158334131144e-05,
+      "loss": 0.3474,
+      "step": 4472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4509290532146893e-05,
+      "loss": 0.5643,
+      "step": 4473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4506997351451306e-05,
+      "loss": 0.2019,
+      "step": 4474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4504703799314266e-05,
+      "loss": 1.0089,
+      "step": 4475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4502409875825392e-05,
+      "loss": 0.6585,
+      "step": 4476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4500115581074313e-05,
+      "loss": 1.9275,
+      "step": 4477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4497820915150655e-05,
+      "loss": 0.2966,
+      "step": 4478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4495525878144076e-05,
+      "loss": 0.8241,
+      "step": 4479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4493230470144245e-05,
+      "loss": 0.9589,
+      "step": 4480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.449093469124084e-05,
+      "loss": 0.9827,
+      "step": 4481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.448863854152357e-05,
+      "loss": 1.1121,
+      "step": 4482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4486342021082133e-05,
+      "loss": 0.2304,
+      "step": 4483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.448404513000626e-05,
+      "loss": 0.9052,
+      "step": 4484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4481747868385694e-05,
+      "loss": 1.335,
+      "step": 4485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4479450236310183e-05,
+      "loss": 0.945,
+      "step": 4486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4477152233869505e-05,
+      "loss": 1.631,
+      "step": 4487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.447485386115344e-05,
+      "loss": 0.5044,
+      "step": 4488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4472555118251776e-05,
+      "loss": 0.9138,
+      "step": 4489
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4470256005254347e-05,
+      "loss": 0.8607,
+      "step": 4490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4467956522250957e-05,
+      "loss": 0.7611,
+      "step": 4491
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.446565666933146e-05,
+      "loss": 0.1512,
+      "step": 4492
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.446335644658571e-05,
+      "loss": 1.1734,
+      "step": 4493
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4461055854103588e-05,
+      "loss": 0.6632,
+      "step": 4494
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.445875489197496e-05,
+      "loss": 0.5173,
+      "step": 4495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4456453560289733e-05,
+      "loss": 0.4408,
+      "step": 4496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4454151859137827e-05,
+      "loss": 0.8927,
+      "step": 4497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.445184978860916e-05,
+      "loss": 0.8736,
+      "step": 4498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4449547348793682e-05,
+      "loss": 0.6078,
+      "step": 4499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4447244539781347e-05,
+      "loss": 0.4572,
+      "step": 4500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.444494136166213e-05,
+      "loss": 0.5046,
+      "step": 4501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4442637814526008e-05,
+      "loss": 1.1542,
+      "step": 4502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4440333898462993e-05,
+      "loss": 1.4802,
+      "step": 4503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.443802961356309e-05,
+      "loss": 1.4123,
+      "step": 4504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4435724959916336e-05,
+      "loss": 0.1333,
+      "step": 4505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4433419937612768e-05,
+      "loss": 0.4348,
+      "step": 4506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4431114546742446e-05,
+      "loss": 0.6317,
+      "step": 4507
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4428808787395445e-05,
+      "loss": 0.2345,
+      "step": 4508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4426502659661852e-05,
+      "loss": 0.2255,
+      "step": 4509
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4424196163631764e-05,
+      "loss": 0.3946,
+      "step": 4510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.44218892993953e-05,
+      "loss": 0.4702,
+      "step": 4511
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4419582067042583e-05,
+      "loss": 0.5584,
+      "step": 4512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.441727446666377e-05,
+      "loss": 1.5054,
+      "step": 4513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.441496649834901e-05,
+      "loss": 0.7207,
+      "step": 4514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4412658162188476e-05,
+      "loss": 0.8834,
+      "step": 4515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.441034945827236e-05,
+      "loss": 0.9606,
+      "step": 4516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4408040386690863e-05,
+      "loss": 1.3127,
+      "step": 4517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.44057309475342e-05,
+      "loss": 0.1433,
+      "step": 4518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.44034211408926e-05,
+      "loss": 1.173,
+      "step": 4519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4401110966856308e-05,
+      "loss": 0.5798,
+      "step": 4520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.439880042551559e-05,
+      "loss": 1.5652,
+      "step": 4521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.439648951696071e-05,
+      "loss": 0.1213,
+      "step": 4522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4394178241281964e-05,
+      "loss": 1.5054,
+      "step": 4523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4391866598569642e-05,
+      "loss": 1.9572,
+      "step": 4524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.438955458891408e-05,
+      "loss": 0.3873,
+      "step": 4525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.438724221240559e-05,
+      "loss": 0.655,
+      "step": 4526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.438492946913453e-05,
+      "loss": 1.6105,
+      "step": 4527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4382616359191257e-05,
+      "loss": 0.9555,
+      "step": 4528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4380302882666136e-05,
+      "loss": 0.637,
+      "step": 4529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4377989039649567e-05,
+      "loss": 1.4764,
+      "step": 4530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4375674830231947e-05,
+      "loss": 0.3916,
+      "step": 4531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4373360254503694e-05,
+      "loss": 0.9802,
+      "step": 4532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4371045312555236e-05,
+      "loss": 1.075,
+      "step": 4533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.436873000447702e-05,
+      "loss": 1.4606,
+      "step": 4534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4366414330359515e-05,
+      "loss": 1.6421,
+      "step": 4535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4364098290293184e-05,
+      "loss": 1.382,
+      "step": 4536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4361781884368515e-05,
+      "loss": 0.757,
+      "step": 4537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.435946511267602e-05,
+      "loss": 0.3681,
+      "step": 4538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4357147975306207e-05,
+      "loss": 0.1152,
+      "step": 4539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4354830472349615e-05,
+      "loss": 0.1468,
+      "step": 4540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4352512603896782e-05,
+      "loss": 0.2526,
+      "step": 4541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4350194370038272e-05,
+      "loss": 0.2592,
+      "step": 4542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4347875770864657e-05,
+      "loss": 0.4907,
+      "step": 4543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4345556806466533e-05,
+      "loss": 0.8872,
+      "step": 4544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.434323747693449e-05,
+      "loss": 0.9363,
+      "step": 4545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4340917782359155e-05,
+      "loss": 0.7533,
+      "step": 4546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4338597722831154e-05,
+      "loss": 0.6008,
+      "step": 4547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4336277298441135e-05,
+      "loss": 0.1657,
+      "step": 4548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4333956509279756e-05,
+      "loss": 0.4426,
+      "step": 4549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4331635355437693e-05,
+      "loss": 0.3759,
+      "step": 4550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4329313837005636e-05,
+      "loss": 0.5089,
+      "step": 4551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.432699195407428e-05,
+      "loss": 0.5276,
+      "step": 4552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.432466970673435e-05,
+      "loss": 1.8436,
+      "step": 4553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.432234709507657e-05,
+      "loss": 1.2394,
+      "step": 4554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.432002411919169e-05,
+      "loss": 0.9987,
+      "step": 4555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4317700779170466e-05,
+      "loss": 1.6228,
+      "step": 4556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4315377075103675e-05,
+      "loss": 0.6882,
+      "step": 4557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4313053007082104e-05,
+      "loss": 0.5107,
+      "step": 4558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4310728575196558e-05,
+      "loss": 0.4036,
+      "step": 4559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4308403779537846e-05,
+      "loss": 0.8671,
+      "step": 4560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.43060786201968e-05,
+      "loss": 0.2464,
+      "step": 4561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4303753097264272e-05,
+      "loss": 0.2284,
+      "step": 4562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4301427210831113e-05,
+      "loss": 0.4097,
+      "step": 4563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.42991009609882e-05,
+      "loss": 1.3686,
+      "step": 4564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.429677434782642e-05,
+      "loss": 0.3047,
+      "step": 4565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.429444737143667e-05,
+      "loss": 1.4561,
+      "step": 4566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4292120031909875e-05,
+      "loss": 0.1498,
+      "step": 4567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4289792329336958e-05,
+      "loss": 0.1887,
+      "step": 4568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4287464263808864e-05,
+      "loss": 0.1751,
+      "step": 4569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4285135835416552e-05,
+      "loss": 1.2471,
+      "step": 4570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4282807044250997e-05,
+      "loss": 0.6221,
+      "step": 4571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.428047789040318e-05,
+      "loss": 1.4895,
+      "step": 4572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4278148373964103e-05,
+      "loss": 0.9283,
+      "step": 4573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4275818495024785e-05,
+      "loss": 1.6258,
+      "step": 4574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4273488253676255e-05,
+      "loss": 0.3012,
+      "step": 4575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.427115765000955e-05,
+      "loss": 0.9287,
+      "step": 4576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4268826684115736e-05,
+      "loss": 0.9562,
+      "step": 4577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4266495356085877e-05,
+      "loss": 0.1787,
+      "step": 4578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4264163666011062e-05,
+      "loss": 0.3915,
+      "step": 4579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4261831613982393e-05,
+      "loss": 0.8371,
+      "step": 4580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4259499200090982e-05,
+      "loss": 0.7887,
+      "step": 4581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.425716642442795e-05,
+      "loss": 0.8585,
+      "step": 4582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4254833287084454e-05,
+      "loss": 0.9541,
+      "step": 4583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4252499788151644e-05,
+      "loss": 0.6362,
+      "step": 4584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4250165927720684e-05,
+      "loss": 0.2006,
+      "step": 4585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4247831705882767e-05,
+      "loss": 0.8186,
+      "step": 4586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4245497122729086e-05,
+      "loss": 1.6916,
+      "step": 4587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.424316217835086e-05,
+      "loss": 0.188,
+      "step": 4588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4240826872839308e-05,
+      "loss": 1.0457,
+      "step": 4589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4238491206285683e-05,
+      "loss": 0.3286,
+      "step": 4590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.423615517878123e-05,
+      "loss": 1.3507,
+      "step": 4591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.423381879041722e-05,
+      "loss": 0.7519,
+      "step": 4592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4231482041284936e-05,
+      "loss": 0.798,
+      "step": 4593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.422914493147568e-05,
+      "loss": 0.6795,
+      "step": 4594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4226807461080758e-05,
+      "loss": 0.5742,
+      "step": 4595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4224469630191503e-05,
+      "loss": 0.1647,
+      "step": 4596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4222131438899247e-05,
+      "loss": 0.2223,
+      "step": 4597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.421979288729535e-05,
+      "loss": 0.6928,
+      "step": 4598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.421745397547118e-05,
+      "loss": 0.5796,
+      "step": 4599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4215114703518108e-05,
+      "loss": 0.5794,
+      "step": 4600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4212775071527546e-05,
+      "loss": 0.6538,
+      "step": 4601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4210435079590895e-05,
+      "loss": 0.9932,
+      "step": 4602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4208094727799585e-05,
+      "loss": 0.1934,
+      "step": 4603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.420575401624504e-05,
+      "loss": 1.3038,
+      "step": 4604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4203412945018734e-05,
+      "loss": 1.4936,
+      "step": 4605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.420107151421211e-05,
+      "loss": 1.4369,
+      "step": 4606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4198729723916673e-05,
+      "loss": 0.4515,
+      "step": 4607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4196387574223895e-05,
+      "loss": 0.2585,
+      "step": 4608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.41940450652253e-05,
+      "loss": 1.7139,
+      "step": 4609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4191702197012402e-05,
+      "loss": 1.1837,
+      "step": 4610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4189358969676746e-05,
+      "loss": 0.5462,
+      "step": 4611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4187015383309875e-05,
+      "loss": 0.2813,
+      "step": 4612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4184671438003352e-05,
+      "loss": 0.3674,
+      "step": 4613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.418232713384876e-05,
+      "loss": 0.194,
+      "step": 4614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4179982470937694e-05,
+      "loss": 1.2041,
+      "step": 4615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4177637449361758e-05,
+      "loss": 0.683,
+      "step": 4616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4175292069212573e-05,
+      "loss": 0.4647,
+      "step": 4617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4172946330581777e-05,
+      "loss": 0.7518,
+      "step": 4618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4170600233561013e-05,
+      "loss": 0.0405,
+      "step": 4619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4168253778241944e-05,
+      "loss": 0.6177,
+      "step": 4620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.416590696471625e-05,
+      "loss": 1.8796,
+      "step": 4621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4163559793075622e-05,
+      "loss": 0.3857,
+      "step": 4622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.416121226341176e-05,
+      "loss": 2.0558,
+      "step": 4623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.415886437581639e-05,
+      "loss": 1.4191,
+      "step": 4624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.415651613038124e-05,
+      "loss": 0.9758,
+      "step": 4625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.415416752719806e-05,
+      "loss": 0.843,
+      "step": 4626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4151818566358603e-05,
+      "loss": 1.332,
+      "step": 4627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4149469247954653e-05,
+      "loss": 0.7506,
+      "step": 4628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4147119572077996e-05,
+      "loss": 1.0947,
+      "step": 4629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.414476953882043e-05,
+      "loss": 0.7796,
+      "step": 4630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4142419148273774e-05,
+      "loss": 1.1934,
+      "step": 4631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.414006840052986e-05,
+      "loss": 0.6815,
+      "step": 4632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4137717295680534e-05,
+      "loss": 1.2903,
+      "step": 4633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4135365833817655e-05,
+      "loss": 0.461,
+      "step": 4634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4133014015033086e-05,
+      "loss": 0.7169,
+      "step": 4635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4130661839418723e-05,
+      "loss": 0.873,
+      "step": 4636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.412830930706646e-05,
+      "loss": 2.409,
+      "step": 4637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4125956418068217e-05,
+      "loss": 0.5637,
+      "step": 4638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.412360317251592e-05,
+      "loss": 0.1014,
+      "step": 4639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.412124957050151e-05,
+      "loss": 0.3302,
+      "step": 4640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4118895612116945e-05,
+      "loss": 0.1238,
+      "step": 4641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.411654129745419e-05,
+      "loss": 1.2219,
+      "step": 4642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4114186626605235e-05,
+      "loss": 0.6769,
+      "step": 4643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4111831599662074e-05,
+      "loss": 1.0301,
+      "step": 4644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.410947621671672e-05,
+      "loss": 0.2459,
+      "step": 4645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.41071204778612e-05,
+      "loss": 0.7804,
+      "step": 4646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.410476438318755e-05,
+      "loss": 0.4688,
+      "step": 4647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4102407932787824e-05,
+      "loss": 0.562,
+      "step": 4648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.410005112675409e-05,
+      "loss": 0.6143,
+      "step": 4649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4097693965178433e-05,
+      "loss": 0.1426,
+      "step": 4650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4095336448152943e-05,
+      "loss": 1.2988,
+      "step": 4651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.409297857576973e-05,
+      "loss": 0.1625,
+      "step": 4652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4090620348120913e-05,
+      "loss": 0.6227,
+      "step": 4653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.408826176529864e-05,
+      "loss": 1.1073,
+      "step": 4654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.408590282739505e-05,
+      "loss": 0.9424,
+      "step": 4655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4083543534502315e-05,
+      "loss": 0.2926,
+      "step": 4656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.408118388671261e-05,
+      "loss": 0.4046,
+      "step": 4657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.407882388411813e-05,
+      "loss": 0.147,
+      "step": 4658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4076463526811075e-05,
+      "loss": 0.7357,
+      "step": 4659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.407410281488367e-05,
+      "loss": 0.6234,
+      "step": 4660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4071741748428145e-05,
+      "loss": 0.7702,
+      "step": 4661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4069380327536754e-05,
+      "loss": 0.0808,
+      "step": 4662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4067018552301753e-05,
+      "loss": 0.9242,
+      "step": 4663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4064656422815418e-05,
+      "loss": 0.8059,
+      "step": 4664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4062293939170043e-05,
+      "loss": 1.2912,
+      "step": 4665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4059931101457923e-05,
+      "loss": 0.9546,
+      "step": 4666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4057567909771378e-05,
+      "loss": 0.7908,
+      "step": 4667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.405520436420274e-05,
+      "loss": 1.5633,
+      "step": 4668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4052840464844357e-05,
+      "loss": 0.7246,
+      "step": 4669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4050476211788578e-05,
+      "loss": 0.7514,
+      "step": 4670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4048111605127782e-05,
+      "loss": 0.8222,
+      "step": 4671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4045746644954354e-05,
+      "loss": 1.0226,
+      "step": 4672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4043381331360697e-05,
+      "loss": 1.6109,
+      "step": 4673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4041015664439217e-05,
+      "loss": 1.5857,
+      "step": 4674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4038649644282345e-05,
+      "loss": 0.3798,
+      "step": 4675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4036283270982522e-05,
+      "loss": 1.1164,
+      "step": 4676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.40339165446322e-05,
+      "loss": 0.5983,
+      "step": 4677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4031549465323858e-05,
+      "loss": 0.8786,
+      "step": 4678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4029182033149967e-05,
+      "loss": 1.1091,
+      "step": 4679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.402681424820303e-05,
+      "loss": 0.6933,
+      "step": 4680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4024446110575553e-05,
+      "loss": 0.674,
+      "step": 4681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.402207762036006e-05,
+      "loss": 0.3665,
+      "step": 4682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4019708777649092e-05,
+      "loss": 1.1584,
+      "step": 4683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4017339582535205e-05,
+      "loss": 0.9176,
+      "step": 4684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.401497003511095e-05,
+      "loss": 1.5531,
+      "step": 4685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.401260013546892e-05,
+      "loss": 1.1059,
+      "step": 4686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.40102298837017e-05,
+      "loss": 0.9909,
+      "step": 4687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4007859279901894e-05,
+      "loss": 0.4589,
+      "step": 4688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4005488324162134e-05,
+      "loss": 0.5863,
+      "step": 4689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4003117016575046e-05,
+      "loss": 1.1505,
+      "step": 4690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.400074535723327e-05,
+      "loss": 0.6109,
+      "step": 4691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.399837334622948e-05,
+      "loss": 1.1615,
+      "step": 4692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.399600098365635e-05,
+      "loss": 0.6955,
+      "step": 4693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.399362826960657e-05,
+      "loss": 0.5803,
+      "step": 4694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3991255204172832e-05,
+      "loss": 0.4094,
+      "step": 4695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3988881787447864e-05,
+      "loss": 0.9017,
+      "step": 4696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3986508019524384e-05,
+      "loss": 0.5627,
+      "step": 4697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.398413390049515e-05,
+      "loss": 0.4676,
+      "step": 4698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3981759430452907e-05,
+      "loss": 0.9078,
+      "step": 4699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.397938460949044e-05,
+      "loss": 0.7698,
+      "step": 4700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.397700943770052e-05,
+      "loss": 1.2302,
+      "step": 4701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.397463391517595e-05,
+      "loss": 0.281,
+      "step": 4702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.397225804200955e-05,
+      "loss": 0.4178,
+      "step": 4703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.396988181829414e-05,
+      "loss": 0.3359,
+      "step": 4704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3967505244122557e-05,
+      "loss": 1.0195,
+      "step": 4705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.396512831958766e-05,
+      "loss": 0.2849,
+      "step": 4706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3962751044782316e-05,
+      "loss": 0.1316,
+      "step": 4707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3960373419799403e-05,
+      "loss": 0.0646,
+      "step": 4708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3957995444731813e-05,
+      "loss": 0.6575,
+      "step": 4709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.395561711967246e-05,
+      "loss": 0.4359,
+      "step": 4710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3953238444714263e-05,
+      "loss": 1.4835,
+      "step": 4711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.395085941995016e-05,
+      "loss": 1.1316,
+      "step": 4712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.39484800454731e-05,
+      "loss": 0.407,
+      "step": 4713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.394610032137604e-05,
+      "loss": 1.2938,
+      "step": 4714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.394372024775196e-05,
+      "loss": 0.1735,
+      "step": 4715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.394133982469386e-05,
+      "loss": 1.2379,
+      "step": 4716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.393895905229473e-05,
+      "loss": 0.7304,
+      "step": 4717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.393657793064759e-05,
+      "loss": 1.0899,
+      "step": 4718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3934196459845474e-05,
+      "loss": 0.8193,
+      "step": 4719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3931814639981427e-05,
+      "loss": 0.2865,
+      "step": 4720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3929432471148513e-05,
+      "loss": 0.5669,
+      "step": 4721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3927049953439795e-05,
+      "loss": 1.2771,
+      "step": 4722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3924667086948357e-05,
+      "loss": 1.6063,
+      "step": 4723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.392228387176731e-05,
+      "loss": 1.3696,
+      "step": 4724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.391990030798976e-05,
+      "loss": 0.4671,
+      "step": 4725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3917516395708832e-05,
+      "loss": 1.7589,
+      "step": 4726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3915132135017673e-05,
+      "loss": 0.7852,
+      "step": 4727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3912747526009423e-05,
+      "loss": 0.2902,
+      "step": 4728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3910362568777267e-05,
+      "loss": 0.443,
+      "step": 4729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.390797726341437e-05,
+      "loss": 1.0571,
+      "step": 4730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3905591610013944e-05,
+      "loss": 0.2565,
+      "step": 4731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3903205608669177e-05,
+      "loss": 1.1615,
+      "step": 4732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.390081925947331e-05,
+      "loss": 0.7781,
+      "step": 4733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3898432562519563e-05,
+      "loss": 0.9505,
+      "step": 4734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3896045517901195e-05,
+      "loss": 1.5461,
+      "step": 4735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3893658125711463e-05,
+      "loss": 0.9175,
+      "step": 4736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.389127038604365e-05,
+      "loss": 1.0576,
+      "step": 4737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3888882298991035e-05,
+      "loss": 0.7791,
+      "step": 4738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3886493864646933e-05,
+      "loss": 0.4128,
+      "step": 4739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3884105083104653e-05,
+      "loss": 0.8825,
+      "step": 4740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3881715954457528e-05,
+      "loss": 0.5992,
+      "step": 4741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3879326478798904e-05,
+      "loss": 1.0135,
+      "step": 4742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3876936656222134e-05,
+      "loss": 0.5104,
+      "step": 4743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.387454648682059e-05,
+      "loss": 0.7094,
+      "step": 4744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.387215597068766e-05,
+      "loss": 0.2491,
+      "step": 4745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3869765107916745e-05,
+      "loss": 0.3311,
+      "step": 4746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.386737389860125e-05,
+      "loss": 0.6698,
+      "step": 4747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.38649823428346e-05,
+      "loss": 0.542,
+      "step": 4748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.386259044071024e-05,
+      "loss": 0.2792,
+      "step": 4749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.386019819232162e-05,
+      "loss": 0.1301,
+      "step": 4750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.38578055977622e-05,
+      "loss": 1.5383,
+      "step": 4751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.385541265712547e-05,
+      "loss": 0.6141,
+      "step": 4752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3853019370504912e-05,
+      "loss": 0.6516,
+      "step": 4753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3850625737994037e-05,
+      "loss": 0.5875,
+      "step": 4754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3848231759686372e-05,
+      "loss": 1.9929,
+      "step": 4755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.384583743567544e-05,
+      "loss": 2.0923,
+      "step": 4756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3843442766054793e-05,
+      "loss": 0.9899,
+      "step": 4757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3841047750917994e-05,
+      "loss": 0.1876,
+      "step": 4758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3838652390358608e-05,
+      "loss": 0.7843,
+      "step": 4759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3836256684470232e-05,
+      "loss": 0.5634,
+      "step": 4760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3833860633346463e-05,
+      "loss": 0.7009,
+      "step": 4761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3831464237080916e-05,
+      "loss": 0.6631,
+      "step": 4762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3829067495767217e-05,
+      "loss": 0.2901,
+      "step": 4763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3826670409499008e-05,
+      "loss": 1.8087,
+      "step": 4764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3824272978369947e-05,
+      "loss": 1.9423,
+      "step": 4765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3821875202473703e-05,
+      "loss": 0.7052,
+      "step": 4766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.381947708190395e-05,
+      "loss": 1.4807,
+      "step": 4767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.3817078616754395e-05,
+      "loss": 1.7425,
+      "step": 4768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3814679807118736e-05,
+      "loss": 0.6813,
+      "step": 4769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3812280653090704e-05,
+      "loss": 0.1596,
+      "step": 4770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3809881154764028e-05,
+      "loss": 1.354,
+      "step": 4771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3807481312232455e-05,
+      "loss": 0.3338,
+      "step": 4772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.380508112558976e-05,
+      "loss": 1.4098,
+      "step": 4773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.380268059492971e-05,
+      "loss": 0.3594,
+      "step": 4774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3800279720346094e-05,
+      "loss": 0.6478,
+      "step": 4775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3797878501932718e-05,
+      "loss": 0.748,
+      "step": 4776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3795476939783394e-05,
+      "loss": 2.603,
+      "step": 4777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3793075033991957e-05,
+      "loss": 1.2452,
+      "step": 4778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.379067278465225e-05,
+      "loss": 0.7024,
+      "step": 4779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3788270191858125e-05,
+      "loss": 2.257,
+      "step": 4780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3785867255703457e-05,
+      "loss": 0.573,
+      "step": 4781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.378346397628213e-05,
+      "loss": 1.0416,
+      "step": 4782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3781060353688035e-05,
+      "loss": 1.1399,
+      "step": 4783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3778656388015086e-05,
+      "loss": 0.9244,
+      "step": 4784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.377625207935721e-05,
+      "loss": 0.615,
+      "step": 4785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3773847427808337e-05,
+      "loss": 1.2497,
+      "step": 4786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.377144243346242e-05,
+      "loss": 0.9801,
+      "step": 4787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3769037096413427e-05,
+      "loss": 0.3035,
+      "step": 4788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.376663141675533e-05,
+      "loss": 0.7176,
+      "step": 4789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3764225394582123e-05,
+      "loss": 0.4407,
+      "step": 4790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3761819029987808e-05,
+      "loss": 0.9973,
+      "step": 4791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3759412323066404e-05,
+      "loss": 0.7287,
+      "step": 4792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3757005273911942e-05,
+      "loss": 0.6746,
+      "step": 4793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.375459788261846e-05,
+      "loss": 0.1849,
+      "step": 4794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.375219014928003e-05,
+      "loss": 0.534,
+      "step": 4795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.374978207399071e-05,
+      "loss": 0.7371,
+      "step": 4796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.374737365684459e-05,
+      "loss": 0.4205,
+      "step": 4797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.374496489793576e-05,
+      "loss": 0.8215,
+      "step": 4798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.374255579735834e-05,
+      "loss": 0.1795,
+      "step": 4799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3740146355206454e-05,
+      "loss": 0.5163,
+      "step": 4800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3737736571574235e-05,
+      "loss": 1.2625,
+      "step": 4801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3735326446555835e-05,
+      "loss": 0.4148,
+      "step": 4802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3732915980245418e-05,
+      "loss": 1.3725,
+      "step": 4803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3730505172737167e-05,
+      "loss": 0.428,
+      "step": 4804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3728094024125265e-05,
+      "loss": 0.8706,
+      "step": 4805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3725682534503922e-05,
+      "loss": 0.7391,
+      "step": 4806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3723270703967353e-05,
+      "loss": 1.1406,
+      "step": 4807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.372085853260979e-05,
+      "loss": 0.2612,
+      "step": 4808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3718446020525477e-05,
+      "loss": 0.1658,
+      "step": 4809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3716033167808674e-05,
+      "loss": 1.3037,
+      "step": 4810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3713619974553647e-05,
+      "loss": 0.2658,
+      "step": 4811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3711206440854683e-05,
+      "loss": 0.3567,
+      "step": 4812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.370879256680608e-05,
+      "loss": 0.3316,
+      "step": 4813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3706378352502153e-05,
+      "loss": 0.3983,
+      "step": 4814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.370396379803722e-05,
+      "loss": 1.6537,
+      "step": 4815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3701548903505614e-05,
+      "loss": 0.2215,
+      "step": 4816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3699133669001697e-05,
+      "loss": 0.315,
+      "step": 4817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3696718094619823e-05,
+      "loss": 0.6139,
+      "step": 4818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.369430218045438e-05,
+      "loss": 1.3045,
+      "step": 4819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.369188592659975e-05,
+      "loss": 1.1272,
+      "step": 4820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3689469333150333e-05,
+      "loss": 1.3578,
+      "step": 4821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.368705240020056e-05,
+      "loss": 0.468,
+      "step": 4822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3684635127844854e-05,
+      "loss": 1.886,
+      "step": 4823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3682217516177658e-05,
+      "loss": 0.4351,
+      "step": 4824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3679799565293425e-05,
+      "loss": 0.6637,
+      "step": 4825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3677381275286632e-05,
+      "loss": 1.0258,
+      "step": 4826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3674962646251758e-05,
+      "loss": 0.678,
+      "step": 4827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3672543678283305e-05,
+      "loss": 0.2382,
+      "step": 4828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3670124371475777e-05,
+      "loss": 1.078,
+      "step": 4829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3667704725923697e-05,
+      "loss": 0.9983,
+      "step": 4830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3665284741721607e-05,
+      "loss": 1.311,
+      "step": 4831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.366286441896405e-05,
+      "loss": 1.3664,
+      "step": 4832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3660443757745597e-05,
+      "loss": 0.8314,
+      "step": 4833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3658022758160814e-05,
+      "loss": 1.332,
+      "step": 4834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3655601420304298e-05,
+      "loss": 2.0625,
+      "step": 4835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3653179744270644e-05,
+      "loss": 0.804,
+      "step": 4836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.365075773015448e-05,
+      "loss": 1.2225,
+      "step": 4837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.364833537805042e-05,
+      "loss": 0.9457,
+      "step": 4838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3645912688053116e-05,
+      "loss": 0.5339,
+      "step": 4839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.364348966025722e-05,
+      "loss": 0.8552,
+      "step": 4840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3641066294757406e-05,
+      "loss": 0.871,
+      "step": 4841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3638642591648342e-05,
+      "loss": 0.231,
+      "step": 4842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3636218551024733e-05,
+      "loss": 0.8745,
+      "step": 4843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.363379417298129e-05,
+      "loss": 1.0569,
+      "step": 4844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3631369457612732e-05,
+      "loss": 1.0514,
+      "step": 4845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3628944405013785e-05,
+      "loss": 0.4972,
+      "step": 4846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3626519015279204e-05,
+      "loss": 0.8197,
+      "step": 4847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3624093288503752e-05,
+      "loss": 0.5273,
+      "step": 4848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3621667224782196e-05,
+      "loss": 0.6614,
+      "step": 4849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3619240824209328e-05,
+      "loss": 0.4646,
+      "step": 4850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3616814086879945e-05,
+      "loss": 0.357,
+      "step": 4851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3614387012888863e-05,
+      "loss": 1.2874,
+      "step": 4852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3611959602330907e-05,
+      "loss": 1.0394,
+      "step": 4853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3609531855300924e-05,
+      "loss": 0.2536,
+      "step": 4854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3607103771893754e-05,
+      "loss": 1.2053,
+      "step": 4855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3604675352204268e-05,
+      "loss": 0.1354,
+      "step": 4856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.360224659632735e-05,
+      "loss": 0.4193,
+      "step": 4857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3599817504357888e-05,
+      "loss": 0.336,
+      "step": 4858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.359738807639079e-05,
+      "loss": 1.0149,
+      "step": 4859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.359495831252097e-05,
+      "loss": 0.1336,
+      "step": 4860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3592528212843365e-05,
+      "loss": 1.1493,
+      "step": 4861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3590097777452916e-05,
+      "loss": 0.1525,
+      "step": 4862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3587667006444577e-05,
+      "loss": 1.2012,
+      "step": 4863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.358523589991333e-05,
+      "loss": 0.5975,
+      "step": 4864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.358280445795415e-05,
+      "loss": 1.3539,
+      "step": 4865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3580372680662038e-05,
+      "loss": 0.3389,
+      "step": 4866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3577940568132007e-05,
+      "loss": 0.2558,
+      "step": 4867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3575508120459076e-05,
+      "loss": 1.2206,
+      "step": 4868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3573075337738282e-05,
+      "loss": 0.2318,
+      "step": 4869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3570642220064675e-05,
+      "loss": 2.1175,
+      "step": 4870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3568208767533316e-05,
+      "loss": 1.5012,
+      "step": 4871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3565774980239286e-05,
+      "loss": 1.4618,
+      "step": 4872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3563340858277667e-05,
+      "loss": 0.2796,
+      "step": 4873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3560906401743566e-05,
+      "loss": 0.4083,
+      "step": 4874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.35584716107321e-05,
+      "loss": 1.2434,
+      "step": 4875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3556036485338388e-05,
+      "loss": 0.2592,
+      "step": 4876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.355360102565758e-05,
+      "loss": 0.7262,
+      "step": 4877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3551165231784827e-05,
+      "loss": 1.6565,
+      "step": 4878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3548729103815295e-05,
+      "loss": 0.9142,
+      "step": 4879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3546292641844162e-05,
+      "loss": 1.8869,
+      "step": 4880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3543855845966628e-05,
+      "loss": 0.6677,
+      "step": 4881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3541418716277895e-05,
+      "loss": 0.4677,
+      "step": 4882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.353898125287319e-05,
+      "loss": 0.529,
+      "step": 4883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.353654345584773e-05,
+      "loss": 0.6469,
+      "step": 4884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.353410532529678e-05,
+      "loss": 0.4677,
+      "step": 4885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3531666861315577e-05,
+      "loss": 0.8374,
+      "step": 4886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3529228063999406e-05,
+      "loss": 1.0874,
+      "step": 4887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3526788933443556e-05,
+      "loss": 1.3154,
+      "step": 4888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3524349469743312e-05,
+      "loss": 0.7254,
+      "step": 4889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3521909672993993e-05,
+      "loss": 0.7713,
+      "step": 4890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3519469543290916e-05,
+      "loss": 0.6603,
+      "step": 4891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.351702908072943e-05,
+      "loss": 1.1025,
+      "step": 4892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3514588285404872e-05,
+      "loss": 0.7378,
+      "step": 4893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.3512147157412606e-05,
+      "loss": 0.4568,
+      "step": 4894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.350970569684802e-05,
+      "loss": 0.3892,
+      "step": 4895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3507263903806486e-05,
+      "loss": 0.6574,
+      "step": 4896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.350482177838341e-05,
+      "loss": 0.7681,
+      "step": 4897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.350237932067422e-05,
+      "loss": 0.5913,
+      "step": 4898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3499936530774328e-05,
+      "loss": 0.6356,
+      "step": 4899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.349749340877918e-05,
+      "loss": 0.0786,
+      "step": 4900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3495049954784227e-05,
+      "loss": 0.2183,
+      "step": 4901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3492606168884938e-05,
+      "loss": 1.7788,
+      "step": 4902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3490162051176796e-05,
+      "loss": 1.0876,
+      "step": 4903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.348771760175529e-05,
+      "loss": 1.743,
+      "step": 4904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3485272820715925e-05,
+      "loss": 1.6821,
+      "step": 4905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.348282770815422e-05,
+      "loss": 1.3326,
+      "step": 4906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3480382264165704e-05,
+      "loss": 0.3216,
+      "step": 4907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3477936488845922e-05,
+      "loss": 1.5581,
+      "step": 4908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3475490382290437e-05,
+      "loss": 0.0435,
+      "step": 4909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3473043944594817e-05,
+      "loss": 0.8734,
+      "step": 4910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.347059717585464e-05,
+      "loss": 0.7184,
+      "step": 4911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.34681500761655e-05,
+      "loss": 0.3505,
+      "step": 4912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.346570264562302e-05,
+      "loss": 0.9233,
+      "step": 4913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3463254884322804e-05,
+      "loss": 0.143,
+      "step": 4914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.34608067923605e-05,
+      "loss": 1.73,
+      "step": 4915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3458358369831753e-05,
+      "loss": 1.546,
+      "step": 4916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.345590961683222e-05,
+      "loss": 0.1518,
+      "step": 4917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3453460533457575e-05,
+      "loss": 1.3613,
+      "step": 4918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.345101111980351e-05,
+      "loss": 0.3045,
+      "step": 4919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.344856137596572e-05,
+      "loss": 0.756,
+      "step": 4920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3446111302039913e-05,
+      "loss": 0.7507,
+      "step": 4921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3443660898121822e-05,
+      "loss": 1.272,
+      "step": 4922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3441210164307186e-05,
+      "loss": 0.3299,
+      "step": 4923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3438759100691747e-05,
+      "loss": 0.2472,
+      "step": 4924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3436307707371275e-05,
+      "loss": 1.5725,
+      "step": 4925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.343385598444155e-05,
+      "loss": 0.5183,
+      "step": 4926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3431403931998347e-05,
+      "loss": 0.204,
+      "step": 4927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.342895155013749e-05,
+      "loss": 0.7411,
+      "step": 4928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3426498838954775e-05,
+      "loss": 0.6499,
+      "step": 4929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.342404579854604e-05,
+      "loss": 1.788,
+      "step": 4930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.342159242900713e-05,
+      "loss": 0.6236,
+      "step": 4931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3419138730433886e-05,
+      "loss": 0.2231,
+      "step": 4932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3416684702922182e-05,
+      "loss": 0.2741,
+      "step": 4933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3414230346567903e-05,
+      "loss": 0.9094,
+      "step": 4934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3411775661466934e-05,
+      "loss": 0.0962,
+      "step": 4935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.340932064771518e-05,
+      "loss": 1.4825,
+      "step": 4936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3406865305408563e-05,
+      "loss": 0.9475,
+      "step": 4937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3404409634643014e-05,
+      "loss": 1.5601,
+      "step": 4938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3401953635514477e-05,
+      "loss": 0.5793,
+      "step": 4939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.33994973081189e-05,
+      "loss": 0.7484,
+      "step": 4940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3397040652552267e-05,
+      "loss": 0.2437,
+      "step": 4941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3394583668910553e-05,
+      "loss": 1.4615,
+      "step": 4942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.339212635728975e-05,
+      "loss": 0.5857,
+      "step": 4943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3389668717785867e-05,
+      "loss": 0.5383,
+      "step": 4944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.338721075049493e-05,
+      "loss": 0.8929,
+      "step": 4945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3384752455512973e-05,
+      "loss": 0.5726,
+      "step": 4946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.338229383293603e-05,
+      "loss": 0.5206,
+      "step": 4947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3379834882860177e-05,
+      "loss": 0.3756,
+      "step": 4948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3377375605381473e-05,
+      "loss": 0.3826,
+      "step": 4949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3374916000596012e-05,
+      "loss": 0.8419,
+      "step": 4950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3372456068599887e-05,
+      "loss": 0.4088,
+      "step": 4951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3369995809489206e-05,
+      "loss": 0.3185,
+      "step": 4952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3367535223360096e-05,
+      "loss": 0.3032,
+      "step": 4953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.336507431030869e-05,
+      "loss": 0.2922,
+      "step": 4954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.336261307043114e-05,
+      "loss": 0.6168,
+      "step": 4955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3360151503823606e-05,
+      "loss": 0.2094,
+      "step": 4956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3357689610582265e-05,
+      "loss": 1.322,
+      "step": 4957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3355227390803292e-05,
+      "loss": 0.4401,
+      "step": 4958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3352764844582904e-05,
+      "loss": 0.4067,
+      "step": 4959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3350301972017302e-05,
+      "loss": 0.8143,
+      "step": 4960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3347838773202714e-05,
+      "loss": 0.157,
+      "step": 4961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.334537524823538e-05,
+      "loss": 0.1733,
+      "step": 4962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3342911397211547e-05,
+      "loss": 0.5641,
+      "step": 4963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3340447220227478e-05,
+      "loss": 0.6284,
+      "step": 4964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.333798271737946e-05,
+      "loss": 0.6631,
+      "step": 4965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3335517888763767e-05,
+      "loss": 0.389,
+      "step": 4966
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3333052734476706e-05,
+      "loss": 0.3055,
+      "step": 4967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3330587254614598e-05,
+      "loss": 0.1784,
+      "step": 4968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.332812144927376e-05,
+      "loss": 1.331,
+      "step": 4969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.332565531855054e-05,
+      "loss": 1.1985,
+      "step": 4970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.332318886254128e-05,
+      "loss": 0.4698,
+      "step": 4971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.332072208134236e-05,
+      "loss": 0.8027,
+      "step": 4972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.331825497505015e-05,
+      "loss": 1.0638,
+      "step": 4973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3315787543761043e-05,
+      "loss": 1.298,
+      "step": 4974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3313319787571435e-05,
+      "loss": 0.248,
+      "step": 4975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3310851706577744e-05,
+      "loss": 1.4132,
+      "step": 4976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.330838330087641e-05,
+      "loss": 0.6437,
+      "step": 4977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.330591457056386e-05,
+      "loss": 0.6801,
+      "step": 4978
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3303445515736557e-05,
+      "loss": 0.628,
+      "step": 4979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3300976136490964e-05,
+      "loss": 1.1619,
+      "step": 4980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3298506432923564e-05,
+      "loss": 0.7163,
+      "step": 4981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.329603640513084e-05,
+      "loss": 0.613,
+      "step": 4982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3293566053209312e-05,
+      "loss": 0.188,
+      "step": 4983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3291095377255486e-05,
+      "loss": 0.5187,
+      "step": 4984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3288624377365892e-05,
+      "loss": 1.6206,
+      "step": 4985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3286153053637083e-05,
+      "loss": 0.948,
+      "step": 4986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.32836814061656e-05,
+      "loss": 0.9797,
+      "step": 4987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3281209435048022e-05,
+      "loss": 0.3635,
+      "step": 4988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3278737140380924e-05,
+      "loss": 0.7708,
+      "step": 4989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3276264522260902e-05,
+      "loss": 0.4264,
+      "step": 4990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3273791580784562e-05,
+      "loss": 0.7087,
+      "step": 4991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3271318316048523e-05,
+      "loss": 0.6106,
+      "step": 4992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3268844728149414e-05,
+      "loss": 0.1614,
+      "step": 4993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3266370817183878e-05,
+      "loss": 0.6049,
+      "step": 4994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3263896583248573e-05,
+      "loss": 0.7136,
+      "step": 4995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3261422026440175e-05,
+      "loss": 0.3432,
+      "step": 4996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3258947146855357e-05,
+      "loss": 0.7211,
+      "step": 4997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.325647194459081e-05,
+      "loss": 0.406,
+      "step": 4998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3253996419743254e-05,
+      "loss": 0.2072,
+      "step": 4999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3251520572409396e-05,
+      "loss": 0.3241,
+      "step": 5000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.324904440268598e-05,
+      "loss": 0.9907,
+      "step": 5001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3246567910669735e-05,
+      "loss": 0.5542,
+      "step": 5002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3244091096457434e-05,
+      "loss": 1.4018,
+      "step": 5003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3241613960145836e-05,
+      "loss": 0.3366,
+      "step": 5004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3239136501831735e-05,
+      "loss": 0.6437,
+      "step": 5005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.323665872161191e-05,
+      "loss": 1.5159,
+      "step": 5006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.323418061958318e-05,
+      "loss": 0.0919,
+      "step": 5007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3231702195842368e-05,
+      "loss": 0.5434,
+      "step": 5008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.32292234504863e-05,
+      "loss": 0.4015,
+      "step": 5009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3226744383611826e-05,
+      "loss": 0.9828,
+      "step": 5010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.322426499531579e-05,
+      "loss": 0.9747,
+      "step": 5011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3221785285695086e-05,
+      "loss": 1.1459,
+      "step": 5012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.321930525484658e-05,
+      "loss": 0.1089,
+      "step": 5013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3216824902867167e-05,
+      "loss": 0.3825,
+      "step": 5014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3214344229853767e-05,
+      "loss": 0.6382,
+      "step": 5015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.321186323590329e-05,
+      "loss": 0.6742,
+      "step": 5016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3209381921112677e-05,
+      "loss": 0.1899,
+      "step": 5017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.320690028557887e-05,
+      "loss": 0.3058,
+      "step": 5018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3204418329398825e-05,
+      "loss": 1.5254,
+      "step": 5019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3201936052669516e-05,
+      "loss": 0.5136,
+      "step": 5020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.3199453455487927e-05,
+      "loss": 1.4292,
+      "step": 5021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.319697053795105e-05,
+      "loss": 0.5956,
+      "step": 5022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3194487300155897e-05,
+      "loss": 0.6803,
+      "step": 5023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3192003742199487e-05,
+      "loss": 1.5897,
+      "step": 5024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3189519864178857e-05,
+      "loss": 1.1097,
+      "step": 5025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3187035666191046e-05,
+      "loss": 0.9407,
+      "step": 5026
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.318455114833312e-05,
+      "loss": 0.4953,
+      "step": 5027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.318206631070214e-05,
+      "loss": 0.1278,
+      "step": 5028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3179581153395202e-05,
+      "loss": 0.6947,
+      "step": 5029
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3177095676509394e-05,
+      "loss": 0.2686,
+      "step": 5030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3174609880141827e-05,
+      "loss": 1.7641,
+      "step": 5031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.317212376438962e-05,
+      "loss": 0.6972,
+      "step": 5032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3169637329349907e-05,
+      "loss": 0.9222,
+      "step": 5033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3167150575119834e-05,
+      "loss": 0.8548,
+      "step": 5034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.316466350179656e-05,
+      "loss": 0.0984,
+      "step": 5035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3162176109477254e-05,
+      "loss": 0.5665,
+      "step": 5036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.31596883982591e-05,
+      "loss": 0.1995,
+      "step": 5037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3157200368239296e-05,
+      "loss": 0.2519,
+      "step": 5038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3154712019515046e-05,
+      "loss": 0.1995,
+      "step": 5039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.315222335218358e-05,
+      "loss": 0.9936,
+      "step": 5040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.314973436634211e-05,
+      "loss": 0.6256,
+      "step": 5041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3147245062087905e-05,
+      "loss": 1.2903,
+      "step": 5042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.314475543951821e-05,
+      "loss": 0.5252,
+      "step": 5043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3142265498730304e-05,
+      "loss": 0.6865,
+      "step": 5044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3139775239821458e-05,
+      "loss": 0.3495,
+      "step": 5045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3137284662888973e-05,
+      "loss": 0.6012,
+      "step": 5046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3134793768030162e-05,
+      "loss": 0.5279,
+      "step": 5047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3132302555342336e-05,
+      "loss": 0.2259,
+      "step": 5048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3129811024922837e-05,
+      "loss": 0.2835,
+      "step": 5049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3127319176869e-05,
+      "loss": 0.7214,
+      "step": 5050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3124827011278186e-05,
+      "loss": 1.4196,
+      "step": 5051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3122334528247768e-05,
+      "loss": 0.3885,
+      "step": 5052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3119841727875128e-05,
+      "loss": 1.5993,
+      "step": 5053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3117348610257654e-05,
+      "loss": 1.0524,
+      "step": 5054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.311485517549276e-05,
+      "loss": 0.4133,
+      "step": 5055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.311236142367786e-05,
+      "loss": 1.2983,
+      "step": 5056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3109867354910386e-05,
+      "loss": 1.5825,
+      "step": 5057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3107372969287793e-05,
+      "loss": 1.257,
+      "step": 5058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.310487826690752e-05,
+      "loss": 0.553,
+      "step": 5059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.310238324786705e-05,
+      "loss": 0.4448,
+      "step": 5060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3099887912263856e-05,
+      "loss": 1.7516,
+      "step": 5061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3097392260195435e-05,
+      "loss": 0.5776,
+      "step": 5062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.309489629175929e-05,
+      "loss": 0.4991,
+      "step": 5063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3092400007052945e-05,
+      "loss": 0.0664,
+      "step": 5064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3089903406173923e-05,
+      "loss": 0.3207,
+      "step": 5065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3087406489219774e-05,
+      "loss": 0.5958,
+      "step": 5066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.308490925628805e-05,
+      "loss": 1.9441,
+      "step": 5067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3082411707476322e-05,
+      "loss": 0.0721,
+      "step": 5068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3079913842882162e-05,
+      "loss": 0.8567,
+      "step": 5069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3077415662603166e-05,
+      "loss": 0.159,
+      "step": 5070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.307491716673695e-05,
+      "loss": 1.4745,
+      "step": 5071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.307241835538111e-05,
+      "loss": 0.7252,
+      "step": 5072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3069919228633298e-05,
+      "loss": 0.5459,
+      "step": 5073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3067419786591133e-05,
+      "loss": 0.7878,
+      "step": 5074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.306492002935229e-05,
+      "loss": 0.6398,
+      "step": 5075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.306241995701442e-05,
+      "loss": 0.3589,
+      "step": 5076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3059919569675208e-05,
+      "loss": 0.5169,
+      "step": 5077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3057418867432346e-05,
+      "loss": 0.9719,
+      "step": 5078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3054917850383533e-05,
+      "loss": 0.3096,
+      "step": 5079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.305241651862649e-05,
+      "loss": 0.7888,
+      "step": 5080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3049914872258943e-05,
+      "loss": 1.0084,
+      "step": 5081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3047412911378623e-05,
+      "loss": 0.5285,
+      "step": 5082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3044910636083296e-05,
+      "loss": 0.6846,
+      "step": 5083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3042408046470725e-05,
+      "loss": 0.6619,
+      "step": 5084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.303990514263868e-05,
+      "loss": 0.7619,
+      "step": 5085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.303740192468495e-05,
+      "loss": 0.5174,
+      "step": 5086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.303489839270734e-05,
+      "loss": 1.006,
+      "step": 5087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3032394546803668e-05,
+      "loss": 0.2925,
+      "step": 5088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3029890387071753e-05,
+      "loss": 0.7293,
+      "step": 5089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3027385913609437e-05,
+      "loss": 1.1752,
+      "step": 5090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3024881126514567e-05,
+      "loss": 1.2166,
+      "step": 5091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.302237602588501e-05,
+      "loss": 0.9306,
+      "step": 5092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3019870611818648e-05,
+      "loss": 0.7691,
+      "step": 5093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.301736488441335e-05,
+      "loss": 0.5144,
+      "step": 5094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.301485884376703e-05,
+      "loss": 0.8315,
+      "step": 5095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.301235248997759e-05,
+      "loss": 0.8253,
+      "step": 5096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3009845823142967e-05,
+      "loss": 1.6944,
+      "step": 5097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3007338843361083e-05,
+      "loss": 0.1094,
+      "step": 5098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.300483155072989e-05,
+      "loss": 0.2005,
+      "step": 5099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3002323945347358e-05,
+      "loss": 0.507,
+      "step": 5100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.299981602731146e-05,
+      "loss": 0.3588,
+      "step": 5101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2997307796720166e-05,
+      "loss": 1.1771,
+      "step": 5102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2994799253671486e-05,
+      "loss": 1.2347,
+      "step": 5103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2992290398263422e-05,
+      "loss": 1.3641,
+      "step": 5104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2989781230594006e-05,
+      "loss": 1.7695,
+      "step": 5105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.298727175076126e-05,
+      "loss": 0.1627,
+      "step": 5106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.298476195886324e-05,
+      "loss": 1.5456,
+      "step": 5107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2982251854997996e-05,
+      "loss": 0.814,
+      "step": 5108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2979741439263607e-05,
+      "loss": 0.0737,
+      "step": 5109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.297723071175815e-05,
+      "loss": 2.0019,
+      "step": 5110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.297471967257972e-05,
+      "loss": 0.4556,
+      "step": 5111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2972208321826426e-05,
+      "loss": 0.2045,
+      "step": 5112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.296969665959639e-05,
+      "loss": 1.8874,
+      "step": 5113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2967184685987737e-05,
+      "loss": 0.1523,
+      "step": 5114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2964672401098616e-05,
+      "loss": 0.2957,
+      "step": 5115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.296215980502718e-05,
+      "loss": 0.2355,
+      "step": 5116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.29596468978716e-05,
+      "loss": 0.1793,
+      "step": 5117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.295713367973005e-05,
+      "loss": 0.4325,
+      "step": 5118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2954620150700728e-05,
+      "loss": 1.2912,
+      "step": 5119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2952106310881834e-05,
+      "loss": 0.1459,
+      "step": 5120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.294959216037159e-05,
+      "loss": 0.7508,
+      "step": 5121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2947077699268222e-05,
+      "loss": 0.9814,
+      "step": 5122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.294456292766997e-05,
+      "loss": 0.651,
+      "step": 5123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2942047845675086e-05,
+      "loss": 1.4064,
+      "step": 5124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2939532453381836e-05,
+      "loss": 0.8131,
+      "step": 5125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.29370167508885e-05,
+      "loss": 1.0792,
+      "step": 5126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2934500738293363e-05,
+      "loss": 1.0024,
+      "step": 5127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2931984415694727e-05,
+      "loss": 0.3067,
+      "step": 5128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.292946778319091e-05,
+      "loss": 0.4889,
+      "step": 5129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.292695084088023e-05,
+      "loss": 0.6733,
+      "step": 5130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2924433588861036e-05,
+      "loss": 0.547,
+      "step": 5131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2921916027231667e-05,
+      "loss": 0.9151,
+      "step": 5132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2919398156090494e-05,
+      "loss": 0.991,
+      "step": 5133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2916879975535882e-05,
+      "loss": 0.6401,
+      "step": 5134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2914361485666214e-05,
+      "loss": 1.463,
+      "step": 5135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2911842686579904e-05,
+      "loss": 0.6435,
+      "step": 5136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2909323578375358e-05,
+      "loss": 0.6961,
+      "step": 5137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2906804161150986e-05,
+      "loss": 0.8383,
+      "step": 5138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2904284435005232e-05,
+      "loss": 0.1199,
+      "step": 5139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2901764400036544e-05,
+      "loss": 0.8191,
+      "step": 5140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2899244056343374e-05,
+      "loss": 1.2287,
+      "step": 5141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.28967234040242e-05,
+      "loss": 0.5335,
+      "step": 5142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2894202443177498e-05,
+      "loss": 0.419,
+      "step": 5143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2891681173901764e-05,
+      "loss": 0.5784,
+      "step": 5144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.288915959629551e-05,
+      "loss": 0.5452,
+      "step": 5145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2886637710457253e-05,
+      "loss": 0.6134,
+      "step": 5146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2884115516485523e-05,
+      "loss": 0.1832,
+      "step": 5147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2881593014478855e-05,
+      "loss": 0.5237,
+      "step": 5148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.2879070204535818e-05,
+      "loss": 0.1748,
+      "step": 5149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2876547086754963e-05,
+      "loss": 0.431,
+      "step": 5150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.287402366123489e-05,
+      "loss": 1.4577,
+      "step": 5151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.287149992807417e-05,
+      "loss": 0.3579,
+      "step": 5152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.286897588737142e-05,
+      "loss": 0.4038,
+      "step": 5153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2866451539225245e-05,
+      "loss": 0.3559,
+      "step": 5154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.286392688373428e-05,
+      "loss": 1.03,
+      "step": 5155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.286140192099716e-05,
+      "loss": 0.4566,
+      "step": 5156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.285887665111253e-05,
+      "loss": 0.2222,
+      "step": 5157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2856351074179066e-05,
+      "loss": 0.0384,
+      "step": 5158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2853825190295438e-05,
+      "loss": 0.1329,
+      "step": 5159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.285129899956033e-05,
+      "loss": 0.4363,
+      "step": 5160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2848772502072437e-05,
+      "loss": 0.3715,
+      "step": 5161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2846245697930484e-05,
+      "loss": 0.5925,
+      "step": 5162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2843718587233186e-05,
+      "loss": 0.8695,
+      "step": 5163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.284119117007928e-05,
+      "loss": 1.8294,
+      "step": 5164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2838663446567507e-05,
+      "loss": 1.5256,
+      "step": 5165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2836135416796625e-05,
+      "loss": 0.5479,
+      "step": 5166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2833607080865423e-05,
+      "loss": 0.2973,
+      "step": 5167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.283107843887266e-05,
+      "loss": 1.7877,
+      "step": 5168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.282854949091715e-05,
+      "loss": 0.3908,
+      "step": 5169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.282602023709768e-05,
+      "loss": 0.9266,
+      "step": 5170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.282349067751309e-05,
+      "loss": 0.459,
+      "step": 5171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2820960812262205e-05,
+      "loss": 0.4333,
+      "step": 5172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.281843064144386e-05,
+      "loss": 1.1808,
+      "step": 5173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2815900165156906e-05,
+      "loss": 0.7034,
+      "step": 5174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2813369383500228e-05,
+      "loss": 1.8591,
+      "step": 5175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2810838296572685e-05,
+      "loss": 0.519,
+      "step": 5176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2808306904473178e-05,
+      "loss": 1.1295,
+      "step": 5177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2805775207300607e-05,
+      "loss": 0.8909,
+      "step": 5178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.280324320515389e-05,
+      "loss": 0.9668,
+      "step": 5179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2800710898131942e-05,
+      "loss": 0.8299,
+      "step": 5180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.279817828633372e-05,
+      "loss": 0.9798,
+      "step": 5181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2795645369858148e-05,
+      "loss": 0.7285,
+      "step": 5182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.279311214880421e-05,
+      "loss": 0.8275,
+      "step": 5183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2790578623270873e-05,
+      "loss": 0.4468,
+      "step": 5184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.278804479335712e-05,
+      "loss": 1.0844,
+      "step": 5185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2785510659161952e-05,
+      "loss": 0.1377,
+      "step": 5186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2782976220784372e-05,
+      "loss": 0.5647,
+      "step": 5187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.278044147832341e-05,
+      "loss": 1.222,
+      "step": 5188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2777906431878096e-05,
+      "loss": 0.7334,
+      "step": 5189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2775371081547473e-05,
+      "loss": 0.5867,
+      "step": 5190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2772835427430598e-05,
+      "loss": 0.9998,
+      "step": 5191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2770299469626543e-05,
+      "loss": 1.0401,
+      "step": 5192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2767763208234388e-05,
+      "loss": 0.9232,
+      "step": 5193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2765226643353224e-05,
+      "loss": 0.6984,
+      "step": 5194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.276268977508215e-05,
+      "loss": 0.6007,
+      "step": 5195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2760152603520294e-05,
+      "loss": 0.115,
+      "step": 5196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2757615128766777e-05,
+      "loss": 0.1192,
+      "step": 5197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2755077350920738e-05,
+      "loss": 0.2415,
+      "step": 5198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2752539270081334e-05,
+      "loss": 0.3699,
+      "step": 5199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.275000088634772e-05,
+      "loss": 0.6852,
+      "step": 5200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2747462199819085e-05,
+      "loss": 1.3334,
+      "step": 5201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.27449232105946e-05,
+      "loss": 0.4911,
+      "step": 5202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2742383918773478e-05,
+      "loss": 0.2192,
+      "step": 5203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2739844324454923e-05,
+      "loss": 0.1926,
+      "step": 5204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2737304427738158e-05,
+      "loss": 0.8708,
+      "step": 5205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.273476422872242e-05,
+      "loss": 0.6542,
+      "step": 5206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.273222372750695e-05,
+      "loss": 1.6165,
+      "step": 5207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2729682924191016e-05,
+      "loss": 0.1964,
+      "step": 5208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2727141818873883e-05,
+      "loss": 0.3489,
+      "step": 5209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2724600411654827e-05,
+      "loss": 0.3249,
+      "step": 5210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.272205870263315e-05,
+      "loss": 0.732,
+      "step": 5211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2719516691908152e-05,
+      "loss": 0.1728,
+      "step": 5212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2716974379579155e-05,
+      "loss": 0.4631,
+      "step": 5213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.271443176574548e-05,
+      "loss": 0.5708,
+      "step": 5214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2711888850506476e-05,
+      "loss": 0.7284,
+      "step": 5215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2709345633961497e-05,
+      "loss": 1.2364,
+      "step": 5216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2706802116209898e-05,
+      "loss": 0.2773,
+      "step": 5217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.270425829735106e-05,
+      "loss": 0.1753,
+      "step": 5218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2701714177484368e-05,
+      "loss": 1.1666,
+      "step": 5219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.269916975670923e-05,
+      "loss": 1.8994,
+      "step": 5220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2696625035125047e-05,
+      "loss": 0.1197,
+      "step": 5221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.269408001283125e-05,
+      "loss": 0.5548,
+      "step": 5222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2691534689927266e-05,
+      "loss": 0.884,
+      "step": 5223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.268898906651255e-05,
+      "loss": 0.3083,
+      "step": 5224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2686443142686553e-05,
+      "loss": 0.7032,
+      "step": 5225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2683896918548756e-05,
+      "loss": 0.7678,
+      "step": 5226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2681350394198626e-05,
+      "loss": 0.2593,
+      "step": 5227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2678803569735664e-05,
+      "loss": 0.6766,
+      "step": 5228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2676256445259377e-05,
+      "loss": 0.4154,
+      "step": 5229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2673709020869283e-05,
+      "loss": 1.3399,
+      "step": 5230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2671161296664905e-05,
+      "loss": 0.5407,
+      "step": 5231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2668613272745782e-05,
+      "loss": 0.7666,
+      "step": 5232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2666064949211475e-05,
+      "loss": 0.41,
+      "step": 5233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.266351632616154e-05,
+      "loss": 0.7317,
+      "step": 5234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.266096740369556e-05,
+      "loss": 0.6061,
+      "step": 5235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2658418181913115e-05,
+      "loss": 0.0731,
+      "step": 5236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.26558686609138e-05,
+      "loss": 0.8888,
+      "step": 5237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2653318840797248e-05,
+      "loss": 0.7894,
+      "step": 5238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2650768721663054e-05,
+      "loss": 0.2709,
+      "step": 5239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.264821830361087e-05,
+      "loss": 0.9495,
+      "step": 5240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2645667586740328e-05,
+      "loss": 0.8729,
+      "step": 5241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2643116571151098e-05,
+      "loss": 1.1502,
+      "step": 5242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2640565256942846e-05,
+      "loss": 0.9001,
+      "step": 5243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2638013644215243e-05,
+      "loss": 0.5722,
+      "step": 5244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.263546173306799e-05,
+      "loss": 0.8809,
+      "step": 5245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.263290952360079e-05,
+      "loss": 0.4839,
+      "step": 5246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2630357015913362e-05,
+      "loss": 0.8927,
+      "step": 5247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2627804210105426e-05,
+      "loss": 0.473,
+      "step": 5248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2625251106276727e-05,
+      "loss": 0.5436,
+      "step": 5249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.262269770452701e-05,
+      "loss": 0.6317,
+      "step": 5250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.262014400495604e-05,
+      "loss": 0.5889,
+      "step": 5251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2617590007663593e-05,
+      "loss": 0.3584,
+      "step": 5252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.261503571274945e-05,
+      "loss": 0.3632,
+      "step": 5253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2612481120313416e-05,
+      "loss": 0.5371,
+      "step": 5254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2609926230455293e-05,
+      "loss": 1.8613,
+      "step": 5255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.26073710432749e-05,
+      "loss": 1.4895,
+      "step": 5256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2604815558872073e-05,
+      "loss": 1.6158,
+      "step": 5257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2602259777346656e-05,
+      "loss": 0.5734,
+      "step": 5258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2599703698798503e-05,
+      "loss": 1.5041,
+      "step": 5259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2597147323327475e-05,
+      "loss": 0.0679,
+      "step": 5260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.259459065103346e-05,
+      "loss": 0.3248,
+      "step": 5261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2592033682016343e-05,
+      "loss": 0.0927,
+      "step": 5262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2589476416376025e-05,
+      "loss": 0.154,
+      "step": 5263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2586918854212424e-05,
+      "loss": 0.1828,
+      "step": 5264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.258436099562546e-05,
+      "loss": 0.0703,
+      "step": 5265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2581802840715074e-05,
+      "loss": 0.8407,
+      "step": 5266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2579244389581202e-05,
+      "loss": 0.4544,
+      "step": 5267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2576685642323823e-05,
+      "loss": 0.0791,
+      "step": 5268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.257412659904289e-05,
+      "loss": 1.2286,
+      "step": 5269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2571567259838394e-05,
+      "loss": 0.892,
+      "step": 5270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.256900762481033e-05,
+      "loss": 2.1774,
+      "step": 5271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2566447694058702e-05,
+      "loss": 0.6915,
+      "step": 5272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2563887467683526e-05,
+      "loss": 1.4564,
+      "step": 5273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2561326945784834e-05,
+      "loss": 1.0143,
+      "step": 5274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2558766128462662e-05,
+      "loss": 1.092,
+      "step": 5275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.2556205015817063e-05,
+      "loss": 0.1952,
+      "step": 5276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2553643607948104e-05,
+      "loss": 1.601,
+      "step": 5277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.255108190495586e-05,
+      "loss": 1.7709,
+      "step": 5278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2548519906940414e-05,
+      "loss": 1.1228,
+      "step": 5279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.254595761400187e-05,
+      "loss": 0.9637,
+      "step": 5280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2543395026240327e-05,
+      "loss": 1.3899,
+      "step": 5281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.254083214375591e-05,
+      "loss": 0.4859,
+      "step": 5282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.253826896664876e-05,
+      "loss": 0.7612,
+      "step": 5283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.253570549501902e-05,
+      "loss": 1.0266,
+      "step": 5284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2533141728966832e-05,
+      "loss": 0.7948,
+      "step": 5285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2530577668592373e-05,
+      "loss": 0.7739,
+      "step": 5286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2528013313995822e-05,
+      "loss": 1.8245,
+      "step": 5287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2525448665277374e-05,
+      "loss": 0.7223,
+      "step": 5288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.252288372253722e-05,
+      "loss": 0.3037,
+      "step": 5289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2520318485875572e-05,
+      "loss": 0.5464,
+      "step": 5290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2517752955392664e-05,
+      "loss": 0.5381,
+      "step": 5291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.251518713118873e-05,
+      "loss": 1.6607,
+      "step": 5292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2512621013364018e-05,
+      "loss": 0.6282,
+      "step": 5293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2510054602018778e-05,
+      "loss": 0.5192,
+      "step": 5294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2507487897253292e-05,
+      "loss": 0.9782,
+      "step": 5295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2504920899167842e-05,
+      "loss": 0.6833,
+      "step": 5296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.250235360786271e-05,
+      "loss": 0.7073,
+      "step": 5297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2499786023438207e-05,
+      "loss": 0.5786,
+      "step": 5298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.249721814599465e-05,
+      "loss": 0.653,
+      "step": 5299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2494649975632376e-05,
+      "loss": 0.5958,
+      "step": 5300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.249208151245171e-05,
+      "loss": 0.1791,
+      "step": 5301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2489512756553005e-05,
+      "loss": 0.9551,
+      "step": 5302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2486943708036627e-05,
+      "loss": 1.1317,
+      "step": 5303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2484374367002954e-05,
+      "loss": 0.121,
+      "step": 5304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2481804733552362e-05,
+      "loss": 2.0516,
+      "step": 5305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2479234807785248e-05,
+      "loss": 0.4988,
+      "step": 5306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2476664589802024e-05,
+      "loss": 0.8509,
+      "step": 5307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2474094079703113e-05,
+      "loss": 0.8504,
+      "step": 5308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2471523277588936e-05,
+      "loss": 0.9192,
+      "step": 5309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.246895218355994e-05,
+      "loss": 0.9617,
+      "step": 5310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2466380797716576e-05,
+      "loss": 1.0311,
+      "step": 5311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.246380912015932e-05,
+      "loss": 0.7675,
+      "step": 5312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2461237150988633e-05,
+      "loss": 0.9272,
+      "step": 5313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2458664890305005e-05,
+      "loss": 1.2146,
+      "step": 5314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2456092338208944e-05,
+      "loss": 0.5167,
+      "step": 5315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2453519494800957e-05,
+      "loss": 0.6394,
+      "step": 5316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.245094636018156e-05,
+      "loss": 0.1985,
+      "step": 5317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2448372934451294e-05,
+      "loss": 1.2898,
+      "step": 5318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2445799217710698e-05,
+      "loss": 1.1342,
+      "step": 5319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2443225210060333e-05,
+      "loss": 0.5117,
+      "step": 5320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2440650911600765e-05,
+      "loss": 1.1779,
+      "step": 5321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2438076322432563e-05,
+      "loss": 1.4216,
+      "step": 5322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2435501442656333e-05,
+      "loss": 1.6734,
+      "step": 5323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2432926272372664e-05,
+      "loss": 1.027,
+      "step": 5324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.243035081168217e-05,
+      "loss": 1.7392,
+      "step": 5325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2427775060685486e-05,
+      "loss": 1.1951,
+      "step": 5326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2425199019483236e-05,
+      "loss": 0.3162,
+      "step": 5327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.242262268817607e-05,
+      "loss": 1.7297,
+      "step": 5328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.242004606686465e-05,
+      "loss": 0.6885,
+      "step": 5329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2417469155649638e-05,
+      "loss": 0.6533,
+      "step": 5330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.241489195463172e-05,
+      "loss": 1.3985,
+      "step": 5331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2412314463911587e-05,
+      "loss": 0.943,
+      "step": 5332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2409736683589942e-05,
+      "loss": 1.5778,
+      "step": 5333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2407158613767497e-05,
+      "loss": 0.8479,
+      "step": 5334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2404580254544986e-05,
+      "loss": 1.3263,
+      "step": 5335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.240200160602314e-05,
+      "loss": 0.6108,
+      "step": 5336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2399422668302705e-05,
+      "loss": 0.4226,
+      "step": 5337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2396843441484448e-05,
+      "loss": 0.3103,
+      "step": 5338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2394263925669137e-05,
+      "loss": 0.8202,
+      "step": 5339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2391684120957552e-05,
+      "loss": 1.309,
+      "step": 5340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2389104027450494e-05,
+      "loss": 1.7026,
+      "step": 5341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2386523645248763e-05,
+      "loss": 0.741,
+      "step": 5342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.238394297445317e-05,
+      "loss": 1.1733,
+      "step": 5343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.238136201516455e-05,
+      "loss": 0.3985,
+      "step": 5344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.237878076748374e-05,
+      "loss": 1.0734,
+      "step": 5345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.23761992315116e-05,
+      "loss": 0.9352,
+      "step": 5346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.237361740734897e-05,
+      "loss": 0.3094,
+      "step": 5347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.237103529509674e-05,
+      "loss": 0.5727,
+      "step": 5348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2368452894855787e-05,
+      "loss": 0.0903,
+      "step": 5349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2365870206727014e-05,
+      "loss": 0.3718,
+      "step": 5350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2363287230811316e-05,
+      "loss": 0.2401,
+      "step": 5351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.236070396720961e-05,
+      "loss": 1.2837,
+      "step": 5352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2358120416022838e-05,
+      "loss": 0.6006,
+      "step": 5353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2355536577351933e-05,
+      "loss": 0.2008,
+      "step": 5354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.235295245129785e-05,
+      "loss": 0.8916,
+      "step": 5355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2350368037961538e-05,
+      "loss": 0.305,
+      "step": 5356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2347783337443984e-05,
+      "loss": 1.7502,
+      "step": 5357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2345198349846175e-05,
+      "loss": 0.8058,
+      "step": 5358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2342613075269103e-05,
+      "loss": 0.0943,
+      "step": 5359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2340027513813767e-05,
+      "loss": 1.1122,
+      "step": 5360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.23374416655812e-05,
+      "loss": 0.0558,
+      "step": 5361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2334855530672425e-05,
+      "loss": 0.9639,
+      "step": 5362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.233226910918848e-05,
+      "loss": 0.6135,
+      "step": 5363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2329682401230425e-05,
+      "loss": 1.7351,
+      "step": 5364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2327095406899315e-05,
+      "loss": 1.3926,
+      "step": 5365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2324508126296232e-05,
+      "loss": 1.3412,
+      "step": 5366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2321920559522262e-05,
+      "loss": 0.8097,
+      "step": 5367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2319332706678496e-05,
+      "loss": 1.3569,
+      "step": 5368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2316744567866043e-05,
+      "loss": 0.8914,
+      "step": 5369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2314156143186027e-05,
+      "loss": 0.3527,
+      "step": 5370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2311567432739576e-05,
+      "loss": 0.9869,
+      "step": 5371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2308978436627835e-05,
+      "loss": 0.8925,
+      "step": 5372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2306389154951952e-05,
+      "loss": 0.2408,
+      "step": 5373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.230379958781309e-05,
+      "loss": 1.7165,
+      "step": 5374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2301209735312433e-05,
+      "loss": 0.5958,
+      "step": 5375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.229861959755116e-05,
+      "loss": 0.5566,
+      "step": 5376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.229602917463047e-05,
+      "loss": 0.5864,
+      "step": 5377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2293438466651565e-05,
+      "loss": 1.1681,
+      "step": 5378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.229084747371568e-05,
+      "loss": 0.7142,
+      "step": 5379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2288256195924034e-05,
+      "loss": 1.0815,
+      "step": 5380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2285664633377876e-05,
+      "loss": 1.3985,
+      "step": 5381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.228307278617845e-05,
+      "loss": 0.3974,
+      "step": 5382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.228048065442703e-05,
+      "loss": 0.5527,
+      "step": 5383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.227788823822489e-05,
+      "loss": 0.8425,
+      "step": 5384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.227529553767331e-05,
+      "loss": 0.3662,
+      "step": 5385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2272702552873594e-05,
+      "loss": 1.0903,
+      "step": 5386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2270109283927043e-05,
+      "loss": 1.6833,
+      "step": 5387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2267515730934987e-05,
+      "loss": 1.2217,
+      "step": 5388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2264921893998754e-05,
+      "loss": 0.657,
+      "step": 5389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2262327773219683e-05,
+      "loss": 0.404,
+      "step": 5390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2259733368699127e-05,
+      "loss": 0.8828,
+      "step": 5391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2257138680538455e-05,
+      "loss": 0.9006,
+      "step": 5392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2254543708839033e-05,
+      "loss": 0.6481,
+      "step": 5393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.225194845370226e-05,
+      "loss": 0.9536,
+      "step": 5394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.224935291522952e-05,
+      "loss": 0.7293,
+      "step": 5395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2246757093522233e-05,
+      "loss": 0.9604,
+      "step": 5396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.224416098868181e-05,
+      "loss": 0.9882,
+      "step": 5397
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.224156460080969e-05,
+      "loss": 0.1346,
+      "step": 5398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2238967930007307e-05,
+      "loss": 0.3425,
+      "step": 5399
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2236370976376122e-05,
+      "loss": 0.6681,
+      "step": 5400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2233773740017588e-05,
+      "loss": 1.1709,
+      "step": 5401
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.223117622103319e-05,
+      "loss": 2.0114,
+      "step": 5402
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.2228578419524406e-05,
+      "loss": 1.2395,
+      "step": 5403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2225980335592738e-05,
+      "loss": 0.4026,
+      "step": 5404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2223381969339696e-05,
+      "loss": 0.4714,
+      "step": 5405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.222078332086679e-05,
+      "loss": 1.4984,
+      "step": 5406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.221818439027556e-05,
+      "loss": 0.6776,
+      "step": 5407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2215585177667543e-05,
+      "loss": 0.2382,
+      "step": 5408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2212985683144284e-05,
+      "loss": 2.0006,
+      "step": 5409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.221038590680736e-05,
+      "loss": 0.1706,
+      "step": 5410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2207785848758333e-05,
+      "loss": 0.7666,
+      "step": 5411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2205185509098797e-05,
+      "loss": 0.429,
+      "step": 5412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.220258488793034e-05,
+      "loss": 0.1256,
+      "step": 5413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2199983985354577e-05,
+      "loss": 0.3137,
+      "step": 5414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.219738280147312e-05,
+      "loss": 0.6231,
+      "step": 5415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.21947813363876e-05,
+      "loss": 0.7182,
+      "step": 5416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2192179590199662e-05,
+      "loss": 0.6538,
+      "step": 5417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2189577563010944e-05,
+      "loss": 0.8769,
+      "step": 5418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2186975254923125e-05,
+      "loss": 0.5124,
+      "step": 5419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2184372666037864e-05,
+      "loss": 0.4805,
+      "step": 5420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2181769796456857e-05,
+      "loss": 0.9904,
+      "step": 5421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2179166646281788e-05,
+      "loss": 1.2533,
+      "step": 5422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2176563215614367e-05,
+      "loss": 0.2008,
+      "step": 5423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2173959504556313e-05,
+      "loss": 0.5218,
+      "step": 5424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.217135551320935e-05,
+      "loss": 0.9695,
+      "step": 5425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.216875124167522e-05,
+      "loss": 1.1922,
+      "step": 5426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2166146690055672e-05,
+      "loss": 0.8977,
+      "step": 5427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2163541858452463e-05,
+      "loss": 1.4686,
+      "step": 5428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2160936746967377e-05,
+      "loss": 1.0287,
+      "step": 5429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.215833135570218e-05,
+      "loss": 1.1385,
+      "step": 5430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2155725684758674e-05,
+      "loss": 0.8778,
+      "step": 5431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2153119734238664e-05,
+      "loss": 0.9365,
+      "step": 5432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2150513504243965e-05,
+      "loss": 0.6144,
+      "step": 5433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2147906994876397e-05,
+      "loss": 0.6278,
+      "step": 5434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2145300206237804e-05,
+      "loss": 0.2779,
+      "step": 5435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2142693138430034e-05,
+      "loss": 0.5126,
+      "step": 5436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2140085791554943e-05,
+      "loss": 0.6895,
+      "step": 5437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2137478165714406e-05,
+      "loss": 0.5887,
+      "step": 5438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2134870261010294e-05,
+      "loss": 1.2841,
+      "step": 5439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2132262077544502e-05,
+      "loss": 0.9576,
+      "step": 5440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2129653615418944e-05,
+      "loss": 0.6409,
+      "step": 5441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.212704487473552e-05,
+      "loss": 0.9003,
+      "step": 5442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2124435855596156e-05,
+      "loss": 0.4865,
+      "step": 5443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2121826558102792e-05,
+      "loss": 1.2296,
+      "step": 5444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2119216982357374e-05,
+      "loss": 0.2346,
+      "step": 5445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2116607128461857e-05,
+      "loss": 0.7341,
+      "step": 5446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2113996996518207e-05,
+      "loss": 0.557,
+      "step": 5447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2111386586628407e-05,
+      "loss": 0.3328,
+      "step": 5448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2108775898894443e-05,
+      "loss": 0.6276,
+      "step": 5449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2106164933418314e-05,
+      "loss": 0.5787,
+      "step": 5450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2103553690302037e-05,
+      "loss": 1.4524,
+      "step": 5451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.210094216964763e-05,
+      "loss": 1.5397,
+      "step": 5452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2098330371557128e-05,
+      "loss": 1.2607,
+      "step": 5453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2095718296132573e-05,
+      "loss": 0.6123,
+      "step": 5454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.209310594347602e-05,
+      "loss": 1.4735,
+      "step": 5455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2090493313689533e-05,
+      "loss": 0.199,
+      "step": 5456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2087880406875194e-05,
+      "loss": 1.103,
+      "step": 5457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2085267223135085e-05,
+      "loss": 0.5002,
+      "step": 5458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2082653762571307e-05,
+      "loss": 1.8637,
+      "step": 5459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.208004002528596e-05,
+      "loss": 1.7403,
+      "step": 5460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2077426011381176e-05,
+      "loss": 1.697,
+      "step": 5461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.207481172095908e-05,
+      "loss": 0.1241,
+      "step": 5462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.207219715412181e-05,
+      "loss": 1.2313,
+      "step": 5463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2069582310971525e-05,
+      "loss": 0.4326,
+      "step": 5464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.206696719161038e-05,
+      "loss": 1.3529,
+      "step": 5465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2064351796140552e-05,
+      "loss": 0.6801,
+      "step": 5466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2061736124664227e-05,
+      "loss": 1.0849,
+      "step": 5467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.20591201772836e-05,
+      "loss": 0.6166,
+      "step": 5468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2056503954100873e-05,
+      "loss": 0.425,
+      "step": 5469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2053887455218267e-05,
+      "loss": 0.4298,
+      "step": 5470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2051270680738006e-05,
+      "loss": 0.0664,
+      "step": 5471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.204865363076233e-05,
+      "loss": 0.7923,
+      "step": 5472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.204603630539349e-05,
+      "loss": 0.5826,
+      "step": 5473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2043418704733745e-05,
+      "loss": 0.1087,
+      "step": 5474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2040800828885362e-05,
+      "loss": 0.4281,
+      "step": 5475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2038182677950625e-05,
+      "loss": 1.2848,
+      "step": 5476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2035564252031825e-05,
+      "loss": 0.7266,
+      "step": 5477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2032945551231265e-05,
+      "loss": 0.336,
+      "step": 5478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2030326575651264e-05,
+      "loss": 0.5889,
+      "step": 5479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2027707325394132e-05,
+      "loss": 1.1507,
+      "step": 5480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2025087800562217e-05,
+      "loss": 0.1228,
+      "step": 5481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.202246800125786e-05,
+      "loss": 0.2683,
+      "step": 5482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2019847927583423e-05,
+      "loss": 0.7328,
+      "step": 5483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2017227579641258e-05,
+      "loss": 1.9007,
+      "step": 5484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.201460695753376e-05,
+      "loss": 1.6289,
+      "step": 5485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.201198606136331e-05,
+      "loss": 0.6693,
+      "step": 5486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.200936489123231e-05,
+      "loss": 1.3806,
+      "step": 5487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2006743447243162e-05,
+      "loss": 1.3718,
+      "step": 5488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2004121729498292e-05,
+      "loss": 0.1071,
+      "step": 5489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.2001499738100136e-05,
+      "loss": 0.7186,
+      "step": 5490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.199887747315113e-05,
+      "loss": 1.0732,
+      "step": 5491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.199625493475373e-05,
+      "loss": 0.4153,
+      "step": 5492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1993632123010397e-05,
+      "loss": 0.7337,
+      "step": 5493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1991009038023602e-05,
+      "loss": 1.0596,
+      "step": 5494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.198838567989584e-05,
+      "loss": 0.5111,
+      "step": 5495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1985762048729597e-05,
+      "loss": 1.2737,
+      "step": 5496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1983138144627386e-05,
+      "loss": 0.2327,
+      "step": 5497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1980513967691713e-05,
+      "loss": 0.9373,
+      "step": 5498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1977889518025115e-05,
+      "loss": 0.5748,
+      "step": 5499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.197526479573013e-05,
+      "loss": 0.1542,
+      "step": 5500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1972639800909303e-05,
+      "loss": 1.7143,
+      "step": 5501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1970014533665193e-05,
+      "loss": 0.2672,
+      "step": 5502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1967388994100377e-05,
+      "loss": 0.1964,
+      "step": 5503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.196476318231743e-05,
+      "loss": 1.2451,
+      "step": 5504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1962137098418942e-05,
+      "loss": 1.6845,
+      "step": 5505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.195951074250751e-05,
+      "loss": 0.4625,
+      "step": 5506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1956884114685763e-05,
+      "loss": 0.1771,
+      "step": 5507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1954257215056312e-05,
+      "loss": 0.4277,
+      "step": 5508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1951630043721788e-05,
+      "loss": 0.9244,
+      "step": 5509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.194900260078485e-05,
+      "loss": 0.2845,
+      "step": 5510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1946374886348137e-05,
+      "loss": 0.6179,
+      "step": 5511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1943746900514323e-05,
+      "loss": 0.542,
+      "step": 5512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1941118643386083e-05,
+      "loss": 0.3005,
+      "step": 5513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.19384901150661e-05,
+      "loss": 0.4338,
+      "step": 5514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.193586131565708e-05,
+      "loss": 1.9219,
+      "step": 5515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1933232245261727e-05,
+      "loss": 0.3112,
+      "step": 5516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1930602903982755e-05,
+      "loss": 0.6382,
+      "step": 5517
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1927973291922895e-05,
+      "loss": 0.6398,
+      "step": 5518
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1925343409184892e-05,
+      "loss": 1.1513,
+      "step": 5519
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1922713255871494e-05,
+      "loss": 0.854,
+      "step": 5520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.192008283208546e-05,
+      "loss": 1.1639,
+      "step": 5521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.191745213792956e-05,
+      "loss": 0.9672,
+      "step": 5522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.191482117350658e-05,
+      "loss": 0.224,
+      "step": 5523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1912189938919312e-05,
+      "loss": 0.8653,
+      "step": 5524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.190955843427056e-05,
+      "loss": 1.2158,
+      "step": 5525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1906926659663133e-05,
+      "loss": 0.9852,
+      "step": 5526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.190429461519986e-05,
+      "loss": 0.9764,
+      "step": 5527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1901662300983573e-05,
+      "loss": 1.4142,
+      "step": 5528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.189902971711712e-05,
+      "loss": 1.4224,
+      "step": 5529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.189639686370335e-05,
+      "loss": 1.8239,
+      "step": 5530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.1893763740845142e-05,
+      "loss": 0.9189,
+      "step": 5531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.189113034864536e-05,
+      "loss": 1.1474,
+      "step": 5532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1888496687206905e-05,
+      "loss": 0.8934,
+      "step": 5533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.188586275663266e-05,
+      "loss": 1.202,
+      "step": 5534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1883228557025543e-05,
+      "loss": 0.322,
+      "step": 5535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1880594088488474e-05,
+      "loss": 0.5933,
+      "step": 5536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1877959351124377e-05,
+      "loss": 0.6858,
+      "step": 5537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.187532434503619e-05,
+      "loss": 0.7138,
+      "step": 5538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1872689070326874e-05,
+      "loss": 0.6278,
+      "step": 5539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.187005352709938e-05,
+      "loss": 0.5051,
+      "step": 5540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1867417715456692e-05,
+      "loss": 0.3398,
+      "step": 5541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1864781635501775e-05,
+      "loss": 0.9678,
+      "step": 5542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1862145287337633e-05,
+      "loss": 0.6361,
+      "step": 5543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.185950867106726e-05,
+      "loss": 0.551,
+      "step": 5544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1856871786793687e-05,
+      "loss": 0.543,
+      "step": 5545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1854234634619915e-05,
+      "loss": 0.5495,
+      "step": 5546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1851597214648995e-05,
+      "loss": 0.4464,
+      "step": 5547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1848959526983965e-05,
+      "loss": 0.7341,
+      "step": 5548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.184632157172789e-05,
+      "loss": 0.1391,
+      "step": 5549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1843683348983818e-05,
+      "loss": 0.1482,
+      "step": 5550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1841044858854834e-05,
+      "loss": 1.361,
+      "step": 5551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.183840610144403e-05,
+      "loss": 1.5905,
+      "step": 5552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1835767076854503e-05,
+      "loss": 1.3239,
+      "step": 5553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1833127785189347e-05,
+      "loss": 1.4811,
+      "step": 5554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1830488226551692e-05,
+      "loss": 0.3123,
+      "step": 5555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1827848401044665e-05,
+      "loss": 1.373,
+      "step": 5556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1825208308771405e-05,
+      "loss": 0.232,
+      "step": 5557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1822567949835056e-05,
+      "loss": 2.1052,
+      "step": 5558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.181992732433878e-05,
+      "loss": 0.4769,
+      "step": 5559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1817286432385753e-05,
+      "loss": 0.5693,
+      "step": 5560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1814645274079153e-05,
+      "loss": 0.9823,
+      "step": 5561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1812003849522166e-05,
+      "loss": 0.3267,
+      "step": 5562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.180936215881799e-05,
+      "loss": 0.8492,
+      "step": 5563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1806720202069844e-05,
+      "loss": 0.5871,
+      "step": 5564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1804077979380955e-05,
+      "loss": 1.4401,
+      "step": 5565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.180143549085455e-05,
+      "loss": 0.2649,
+      "step": 5566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1798792736593868e-05,
+      "loss": 0.8657,
+      "step": 5567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1796149716702163e-05,
+      "loss": 0.5272,
+      "step": 5568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.179350643128271e-05,
+      "loss": 0.2398,
+      "step": 5569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1790862880438766e-05,
+      "loss": 1.3635,
+      "step": 5570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1788219064273627e-05,
+      "loss": 0.7314,
+      "step": 5571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.178557498289058e-05,
+      "loss": 0.0574,
+      "step": 5572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.178293063639294e-05,
+      "loss": 0.8535,
+      "step": 5573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.178028602488401e-05,
+      "loss": 0.2368,
+      "step": 5574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.177764114846713e-05,
+      "loss": 0.8248,
+      "step": 5575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1774996007245628e-05,
+      "loss": 0.179,
+      "step": 5576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1772350601322847e-05,
+      "loss": 0.2699,
+      "step": 5577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.176970493080215e-05,
+      "loss": 0.7005,
+      "step": 5578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1767058995786904e-05,
+      "loss": 0.3115,
+      "step": 5579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1764412796380483e-05,
+      "loss": 0.5819,
+      "step": 5580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1761766332686277e-05,
+      "loss": 1.1323,
+      "step": 5581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1759119604807686e-05,
+      "loss": 1.2519,
+      "step": 5582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1756472612848115e-05,
+      "loss": 0.4997,
+      "step": 5583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1753825356910987e-05,
+      "loss": 0.6713,
+      "step": 5584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1751177837099727e-05,
+      "loss": 0.5551,
+      "step": 5585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1748530053517773e-05,
+      "loss": 0.5058,
+      "step": 5586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.174588200626858e-05,
+      "loss": 0.7285,
+      "step": 5587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1743233695455604e-05,
+      "loss": 0.6194,
+      "step": 5588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.174058512118232e-05,
+      "loss": 0.7365,
+      "step": 5589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1737936283552206e-05,
+      "loss": 0.5449,
+      "step": 5590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.173528718266875e-05,
+      "loss": 0.2023,
+      "step": 5591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.173263781863546e-05,
+      "loss": 0.1362,
+      "step": 5592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.172998819155584e-05,
+      "loss": 0.7173,
+      "step": 5593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1727338301533418e-05,
+      "loss": 1.078,
+      "step": 5594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1724688148671717e-05,
+      "loss": 1.3571,
+      "step": 5595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.172203773307429e-05,
+      "loss": 0.7415,
+      "step": 5596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1719387054844687e-05,
+      "loss": 0.6349,
+      "step": 5597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1716736114086468e-05,
+      "loss": 0.847,
+      "step": 5598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1714084910903208e-05,
+      "loss": 1.1497,
+      "step": 5599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1711433445398486e-05,
+      "loss": 0.5115,
+      "step": 5600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1708781717675904e-05,
+      "loss": 0.2926,
+      "step": 5601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.170612972783906e-05,
+      "loss": 0.9687,
+      "step": 5602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1703477475991568e-05,
+      "loss": 2.0837,
+      "step": 5603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.170082496223705e-05,
+      "loss": 1.8166,
+      "step": 5604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1698172186679148e-05,
+      "loss": 1.8325,
+      "step": 5605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1695519149421502e-05,
+      "loss": 0.7393,
+      "step": 5606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1692865850567766e-05,
+      "loss": 0.6978,
+      "step": 5607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.169021229022161e-05,
+      "loss": 0.3454,
+      "step": 5608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1687558468486704e-05,
+      "loss": 0.4486,
+      "step": 5609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1684904385466738e-05,
+      "loss": 0.725,
+      "step": 5610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1682250041265403e-05,
+      "loss": 0.1736,
+      "step": 5611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1679595435986416e-05,
+      "loss": 0.1109,
+      "step": 5612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1676940569733478e-05,
+      "loss": 1.3117,
+      "step": 5613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.167428544261032e-05,
+      "loss": 0.536,
+      "step": 5614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.167163005472069e-05,
+      "loss": 0.1293,
+      "step": 5615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1668974406168325e-05,
+      "loss": 1.5105,
+      "step": 5616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1666318497056977e-05,
+      "loss": 1.6481,
+      "step": 5617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1663662327490425e-05,
+      "loss": 1.6884,
+      "step": 5618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.166100589757244e-05,
+      "loss": 0.4192,
+      "step": 5619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.165834920740681e-05,
+      "loss": 1.4185,
+      "step": 5620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1655692257097334e-05,
+      "loss": 0.2771,
+      "step": 5621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1653035046747816e-05,
+      "loss": 0.2354,
+      "step": 5622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1650377576462085e-05,
+      "loss": 0.315,
+      "step": 5623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1647719846343962e-05,
+      "loss": 0.3918,
+      "step": 5624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.164506185649728e-05,
+      "loss": 0.9742,
+      "step": 5625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.164240360702589e-05,
+      "loss": 0.2912,
+      "step": 5626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1639745098033654e-05,
+      "loss": 1.7285,
+      "step": 5627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1637086329624445e-05,
+      "loss": 1.229,
+      "step": 5628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1634427301902134e-05,
+      "loss": 0.9144,
+      "step": 5629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.163176801497061e-05,
+      "loss": 0.8479,
+      "step": 5630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.162910846893378e-05,
+      "loss": 0.5799,
+      "step": 5631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1626448663895545e-05,
+      "loss": 1.0656,
+      "step": 5632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1623788599959832e-05,
+      "loss": 1.0441,
+      "step": 5633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1621128277230565e-05,
+      "loss": 0.808,
+      "step": 5634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.161846769581168e-05,
+      "loss": 0.6308,
+      "step": 5635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.161580685580714e-05,
+      "loss": 0.8981,
+      "step": 5636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.161314575732089e-05,
+      "loss": 0.4908,
+      "step": 5637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1610484400456912e-05,
+      "loss": 1.0598,
+      "step": 5638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1607822785319177e-05,
+      "loss": 0.7601,
+      "step": 5639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1605160912011682e-05,
+      "loss": 1.0569,
+      "step": 5640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1602498780638422e-05,
+      "loss": 1.0688,
+      "step": 5641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.159983639130341e-05,
+      "loss": 0.5831,
+      "step": 5642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1597173744110665e-05,
+      "loss": 0.7326,
+      "step": 5643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.159451083916422e-05,
+      "loss": 0.8875,
+      "step": 5644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1591847676568113e-05,
+      "loss": 0.882,
+      "step": 5645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1589184256426397e-05,
+      "loss": 1.1122,
+      "step": 5646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.158652057884313e-05,
+      "loss": 0.4155,
+      "step": 5647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1583856643922386e-05,
+      "loss": 0.4743,
+      "step": 5648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1581192451768242e-05,
+      "loss": 0.3237,
+      "step": 5649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1578528002484788e-05,
+      "loss": 0.3743,
+      "step": 5650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1575863296176136e-05,
+      "loss": 1.499,
+      "step": 5651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.157319833294638e-05,
+      "loss": 0.2415,
+      "step": 5652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1570533112899656e-05,
+      "loss": 1.6654,
+      "step": 5653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1567867636140084e-05,
+      "loss": 1.4847,
+      "step": 5654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.156520190277181e-05,
+      "loss": 1.2742,
+      "step": 5655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1562535912898986e-05,
+      "loss": 0.3356,
+      "step": 5656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.155986966662577e-05,
+      "loss": 0.3632,
+      "step": 5657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.1557203164056335e-05,
+      "loss": 0.9667,
+      "step": 5658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1554536405294864e-05,
+      "loss": 0.1744,
+      "step": 5659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1551869390445544e-05,
+      "loss": 1.0575,
+      "step": 5660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1549202119612578e-05,
+      "loss": 0.3053,
+      "step": 5661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.154653459290018e-05,
+      "loss": 0.304,
+      "step": 5662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1543866810412565e-05,
+      "loss": 0.3021,
+      "step": 5663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.154119877225397e-05,
+      "loss": 1.395,
+      "step": 5664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.153853047852863e-05,
+      "loss": 0.5124,
+      "step": 5665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1535861929340804e-05,
+      "loss": 0.0441,
+      "step": 5666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1533193124794746e-05,
+      "loss": 1.6807,
+      "step": 5667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.153052406499473e-05,
+      "loss": 1.775,
+      "step": 5668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1527854750045037e-05,
+      "loss": 1.1187,
+      "step": 5669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1525185180049962e-05,
+      "loss": 0.3237,
+      "step": 5670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.15225153551138e-05,
+      "loss": 0.2527,
+      "step": 5671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.151984527534086e-05,
+      "loss": 0.0898,
+      "step": 5672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1517174940835466e-05,
+      "loss": 0.2827,
+      "step": 5673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1514504351701954e-05,
+      "loss": 0.938,
+      "step": 5674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.151183350804466e-05,
+      "loss": 1.7878,
+      "step": 5675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.150916240996793e-05,
+      "loss": 0.3281,
+      "step": 5676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1506491057576136e-05,
+      "loss": 0.8331,
+      "step": 5677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.150381945097364e-05,
+      "loss": 1.7686,
+      "step": 5678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1501147590264825e-05,
+      "loss": 2.1492,
+      "step": 5679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.149847547555408e-05,
+      "loss": 0.7046,
+      "step": 5680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1495803106945804e-05,
+      "loss": 0.5671,
+      "step": 5681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1493130484544416e-05,
+      "loss": 0.5391,
+      "step": 5682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1490457608454326e-05,
+      "loss": 1.2402,
+      "step": 5683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1487784478779967e-05,
+      "loss": 1.0247,
+      "step": 5684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1485111095625783e-05,
+      "loss": 1.6424,
+      "step": 5685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1482437459096222e-05,
+      "loss": 0.8235,
+      "step": 5686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1479763569295742e-05,
+      "loss": 0.807,
+      "step": 5687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1477089426328807e-05,
+      "loss": 0.9708,
+      "step": 5688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1474415030299913e-05,
+      "loss": 0.9176,
+      "step": 5689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1471740381313537e-05,
+      "loss": 0.9426,
+      "step": 5690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1469065479474182e-05,
+      "loss": 0.6677,
+      "step": 5691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1466390324886352e-05,
+      "loss": 0.5404,
+      "step": 5692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1463714917654578e-05,
+      "loss": 0.4364,
+      "step": 5693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.146103925788338e-05,
+      "loss": 0.8737,
+      "step": 5694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1458363345677293e-05,
+      "loss": 0.8161,
+      "step": 5695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.145568718114088e-05,
+      "loss": 0.5384,
+      "step": 5696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.145301076437868e-05,
+      "loss": 0.4357,
+      "step": 5697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.145033409549528e-05,
+      "loss": 0.3746,
+      "step": 5698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1447657174595253e-05,
+      "loss": 0.6295,
+      "step": 5699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.144498000178318e-05,
+      "loss": 0.2278,
+      "step": 5700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.144230257716366e-05,
+      "loss": 0.6805,
+      "step": 5701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1439624900841318e-05,
+      "loss": 1.0016,
+      "step": 5702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1436946972920746e-05,
+      "loss": 0.6155,
+      "step": 5703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1434268793506592e-05,
+      "loss": 1.6011,
+      "step": 5704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1431590362703478e-05,
+      "loss": 0.5929,
+      "step": 5705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1428911680616066e-05,
+      "loss": 0.1618,
+      "step": 5706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1426232747349e-05,
+      "loss": 0.7303,
+      "step": 5707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.142355356300695e-05,
+      "loss": 0.3144,
+      "step": 5708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.14208741276946e-05,
+      "loss": 0.067,
+      "step": 5709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1418194441516633e-05,
+      "loss": 0.3419,
+      "step": 5710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.141551450457774e-05,
+      "loss": 0.2531,
+      "step": 5711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.141283431698263e-05,
+      "loss": 0.3954,
+      "step": 5712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.141015387883602e-05,
+      "loss": 0.8216,
+      "step": 5713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.140747319024264e-05,
+      "loss": 0.2561,
+      "step": 5714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1404792251307208e-05,
+      "loss": 1.9077,
+      "step": 5715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.140211106213449e-05,
+      "loss": 0.421,
+      "step": 5716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.139942962282923e-05,
+      "loss": 0.3588,
+      "step": 5717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1396747933496197e-05,
+      "loss": 0.2675,
+      "step": 5718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1394065994240162e-05,
+      "loss": 0.1397,
+      "step": 5719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1391383805165908e-05,
+      "loss": 0.3728,
+      "step": 5720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1388701366378237e-05,
+      "loss": 0.3356,
+      "step": 5721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.138601867798194e-05,
+      "loss": 0.5178,
+      "step": 5722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1383335740081842e-05,
+      "loss": 0.8044,
+      "step": 5723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1380652552782763e-05,
+      "loss": 0.2413,
+      "step": 5724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1377969116189535e-05,
+      "loss": 1.3175,
+      "step": 5725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1375285430407003e-05,
+      "loss": 0.4128,
+      "step": 5726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.137260149554001e-05,
+      "loss": 0.6966,
+      "step": 5727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.136991731169343e-05,
+      "loss": 0.442,
+      "step": 5728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1367232878972132e-05,
+      "loss": 0.5674,
+      "step": 5729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1364548197480995e-05,
+      "loss": 0.5997,
+      "step": 5730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.136186326732491e-05,
+      "loss": 0.8829,
+      "step": 5731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.135917808860878e-05,
+      "loss": 1.2146,
+      "step": 5732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1356492661437514e-05,
+      "loss": 0.4586,
+      "step": 5733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.135380698591604e-05,
+      "loss": 1.0181,
+      "step": 5734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1351121062149274e-05,
+      "loss": 0.2467,
+      "step": 5735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1348434890242168e-05,
+      "loss": 0.4446,
+      "step": 5736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1345748470299668e-05,
+      "loss": 0.6776,
+      "step": 5737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.134306180242673e-05,
+      "loss": 0.8076,
+      "step": 5738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.134037488672833e-05,
+      "loss": 0.4821,
+      "step": 5739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.133768772330944e-05,
+      "loss": 0.4662,
+      "step": 5740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.133500031227505e-05,
+      "loss": 0.9165,
+      "step": 5741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.133231265373016e-05,
+      "loss": 1.1212,
+      "step": 5742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1329624747779777e-05,
+      "loss": 1.3597,
+      "step": 5743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1326936594528915e-05,
+      "loss": 0.7501,
+      "step": 5744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1324248194082607e-05,
+      "loss": 0.5137,
+      "step": 5745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1321559546545885e-05,
+      "loss": 0.3187,
+      "step": 5746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.13188706520238e-05,
+      "loss": 0.5612,
+      "step": 5747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.13161815106214e-05,
+      "loss": 0.6199,
+      "step": 5748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.131349212244376e-05,
+      "loss": 0.149,
+      "step": 5749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1310802487595947e-05,
+      "loss": 0.4624,
+      "step": 5750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.130811260618305e-05,
+      "loss": 0.5717,
+      "step": 5751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1305422478310165e-05,
+      "loss": 1.4336,
+      "step": 5752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1302732104082395e-05,
+      "loss": 1.2663,
+      "step": 5753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.130004148360485e-05,
+      "loss": 0.3056,
+      "step": 5754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1297350616982658e-05,
+      "loss": 1.6246,
+      "step": 5755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.129465950432095e-05,
+      "loss": 1.4426,
+      "step": 5756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1291968145724874e-05,
+      "loss": 1.3672,
+      "step": 5757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.128927654129957e-05,
+      "loss": 0.096,
+      "step": 5758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.128658469115021e-05,
+      "loss": 0.0436,
+      "step": 5759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1283892595381962e-05,
+      "loss": 0.6396,
+      "step": 5760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.128120025410001e-05,
+      "loss": 0.3574,
+      "step": 5761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1278507667409543e-05,
+      "loss": 0.5319,
+      "step": 5762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1275814835415757e-05,
+      "loss": 0.5476,
+      "step": 5763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1273121758223866e-05,
+      "loss": 0.3387,
+      "step": 5764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1270428435939095e-05,
+      "loss": 0.3578,
+      "step": 5765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1267734868666663e-05,
+      "loss": 0.2358,
+      "step": 5766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1265041056511808e-05,
+      "loss": 1.3177,
+      "step": 5767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1262346999579784e-05,
+      "loss": 1.3598,
+      "step": 5768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.125965269797585e-05,
+      "loss": 0.9003,
+      "step": 5769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.125695815180527e-05,
+      "loss": 0.4661,
+      "step": 5770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.125426336117332e-05,
+      "loss": 0.2268,
+      "step": 5771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1251568326185287e-05,
+      "loss": 0.3024,
+      "step": 5772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.124887304694647e-05,
+      "loss": 0.6536,
+      "step": 5773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.124617752356217e-05,
+      "loss": 0.3617,
+      "step": 5774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.124348175613771e-05,
+      "loss": 0.9614,
+      "step": 5775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1240785744778395e-05,
+      "loss": 0.2732,
+      "step": 5776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1238089489589586e-05,
+      "loss": 0.9497,
+      "step": 5777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1235392990676604e-05,
+      "loss": 1.5049,
+      "step": 5778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.123269624814481e-05,
+      "loss": 1.5132,
+      "step": 5779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1229999262099572e-05,
+      "loss": 1.2863,
+      "step": 5780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1227302032646257e-05,
+      "loss": 0.0868,
+      "step": 5781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1224604559890246e-05,
+      "loss": 0.4174,
+      "step": 5782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1221906843936934e-05,
+      "loss": 0.7107,
+      "step": 5783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.121920888489172e-05,
+      "loss": 0.7691,
+      "step": 5784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1216510682860008e-05,
+      "loss": 0.9099,
+      "step": 5785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1213812237947225e-05,
+      "loss": 0.7631,
+      "step": 5786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.12111135502588e-05,
+      "loss": 1.0711,
+      "step": 5787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1208414619900165e-05,
+      "loss": 1.0746,
+      "step": 5788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1205715446976776e-05,
+      "loss": 0.374,
+      "step": 5789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1203016031594086e-05,
+      "loss": 0.6406,
+      "step": 5790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1200316373857563e-05,
+      "loss": 0.9013,
+      "step": 5791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1197616473872683e-05,
+      "loss": 0.7897,
+      "step": 5792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1194916331744933e-05,
+      "loss": 0.4561,
+      "step": 5793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1192215947579808e-05,
+      "loss": 0.7012,
+      "step": 5794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.118951532148282e-05,
+      "loss": 0.9713,
+      "step": 5795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1186814453559465e-05,
+      "loss": 1.2381,
+      "step": 5796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1184113343915283e-05,
+      "loss": 0.4753,
+      "step": 5797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1181411992655804e-05,
+      "loss": 0.5883,
+      "step": 5798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1178710399886567e-05,
+      "loss": 0.5822,
+      "step": 5799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1176008565713126e-05,
+      "loss": 0.1098,
+      "step": 5800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.117330649024104e-05,
+      "loss": 0.701,
+      "step": 5801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1170604173575887e-05,
+      "loss": 0.8494,
+      "step": 5802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.116790161582324e-05,
+      "loss": 1.3069,
+      "step": 5803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1165198817088692e-05,
+      "loss": 0.2642,
+      "step": 5804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1162495777477843e-05,
+      "loss": 0.5136,
+      "step": 5805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.11597924970963e-05,
+      "loss": 0.2319,
+      "step": 5806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.115708897604968e-05,
+      "loss": 0.0579,
+      "step": 5807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1154385214443613e-05,
+      "loss": 0.7117,
+      "step": 5808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1151681212383738e-05,
+      "loss": 1.3088,
+      "step": 5809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1148976969975695e-05,
+      "loss": 0.1468,
+      "step": 5810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1146272487325144e-05,
+      "loss": 0.165,
+      "step": 5811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1143567764537744e-05,
+      "loss": 0.4276,
+      "step": 5812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.114086280171918e-05,
+      "loss": 0.3305,
+      "step": 5813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1138157598975126e-05,
+      "loss": 0.7349,
+      "step": 5814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1135452156411284e-05,
+      "loss": 1.7,
+      "step": 5815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1132746474133344e-05,
+      "loss": 0.8663,
+      "step": 5816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1130040552247032e-05,
+      "loss": 0.2496,
+      "step": 5817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1127334390858058e-05,
+      "loss": 1.4882,
+      "step": 5818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.112462799007216e-05,
+      "loss": 0.5438,
+      "step": 5819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1121921349995074e-05,
+      "loss": 1.4103,
+      "step": 5820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1119214470732554e-05,
+      "loss": 0.1779,
+      "step": 5821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.111650735239035e-05,
+      "loss": 1.4348,
+      "step": 5822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.111379999507424e-05,
+      "loss": 0.2247,
+      "step": 5823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1111092398889994e-05,
+      "loss": 1.5595,
+      "step": 5824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.11083845639434e-05,
+      "loss": 0.8174,
+      "step": 5825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1105676490340253e-05,
+      "loss": 0.2447,
+      "step": 5826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1102968178186364e-05,
+      "loss": 0.4781,
+      "step": 5827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1100259627587544e-05,
+      "loss": 1.2489,
+      "step": 5828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.109755083864961e-05,
+      "loss": 1.894,
+      "step": 5829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.109484181147841e-05,
+      "loss": 1.2141,
+      "step": 5830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1092132546179782e-05,
+      "loss": 0.7768,
+      "step": 5831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.108942304285957e-05,
+      "loss": 0.6983,
+      "step": 5832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1086713301623636e-05,
+      "loss": 0.9766,
+      "step": 5833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1084003322577856e-05,
+      "loss": 1.2159,
+      "step": 5834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1081293105828112e-05,
+      "loss": 0.6316,
+      "step": 5835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1078582651480286e-05,
+      "loss": 1.0696,
+      "step": 5836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.107587195964028e-05,
+      "loss": 0.5567,
+      "step": 5837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1073161030414e-05,
+      "loss": 0.2569,
+      "step": 5838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1070449863907363e-05,
+      "loss": 0.6321,
+      "step": 5839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.10677384602263e-05,
+      "loss": 0.829,
+      "step": 5840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1065026819476743e-05,
+      "loss": 0.8337,
+      "step": 5841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.106231494176463e-05,
+      "loss": 1.5014,
+      "step": 5842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1059602827195923e-05,
+      "loss": 0.5935,
+      "step": 5843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1056890475876585e-05,
+      "loss": 0.6893,
+      "step": 5844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1054177887912586e-05,
+      "loss": 0.1203,
+      "step": 5845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.105146506340991e-05,
+      "loss": 0.7978,
+      "step": 5846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1048752002474545e-05,
+      "loss": 0.7936,
+      "step": 5847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1046038705212488e-05,
+      "loss": 1.3459,
+      "step": 5848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.104332517172976e-05,
+      "loss": 0.4376,
+      "step": 5849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.104061140213237e-05,
+      "loss": 0.536,
+      "step": 5850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1037897396526344e-05,
+      "loss": 1.1863,
+      "step": 5851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1035183155017727e-05,
+      "loss": 1.5886,
+      "step": 5852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1032468677712563e-05,
+      "loss": 1.3276,
+      "step": 5853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1029753964716904e-05,
+      "loss": 1.0689,
+      "step": 5854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.102703901613682e-05,
+      "loss": 1.4647,
+      "step": 5855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1024323832078375e-05,
+      "loss": 1.3173,
+      "step": 5856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1021608412647666e-05,
+      "loss": 0.8157,
+      "step": 5857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1018892757950775e-05,
+      "loss": 1.1293,
+      "step": 5858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.101617686809381e-05,
+      "loss": 0.9148,
+      "step": 5859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1013460743182874e-05,
+      "loss": 0.2351,
+      "step": 5860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1010744383324092e-05,
+      "loss": 0.1335,
+      "step": 5861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1008027788623595e-05,
+      "loss": 0.6728,
+      "step": 5862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1005310959187514e-05,
+      "loss": 0.0679,
+      "step": 5863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1002593895122005e-05,
+      "loss": 0.5701,
+      "step": 5864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.099987659653322e-05,
+      "loss": 1.5715,
+      "step": 5865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0997159063527323e-05,
+      "loss": 0.1043,
+      "step": 5866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0994441296210494e-05,
+      "loss": 1.5792,
+      "step": 5867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.099172329468891e-05,
+      "loss": 1.8252,
+      "step": 5868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.098900505906877e-05,
+      "loss": 0.2013,
+      "step": 5869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0986286589456277e-05,
+      "loss": 1.242,
+      "step": 5870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0983567885957638e-05,
+      "loss": 0.6771,
+      "step": 5871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.098084894867908e-05,
+      "loss": 0.3352,
+      "step": 5872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0978129777726825e-05,
+      "loss": 1.418,
+      "step": 5873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0975410373207114e-05,
+      "loss": 0.8788,
+      "step": 5874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.09726907352262e-05,
+      "loss": 1.41,
+      "step": 5875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0969970863890335e-05,
+      "loss": 1.3608,
+      "step": 5876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.096725075930579e-05,
+      "loss": 0.6626,
+      "step": 5877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.096453042157883e-05,
+      "loss": 0.2527,
+      "step": 5878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0961809850815753e-05,
+      "loss": 0.279,
+      "step": 5879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0959089047122842e-05,
+      "loss": 0.2979,
+      "step": 5880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0956368010606408e-05,
+      "loss": 1.8411,
+      "step": 5881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0953646741372754e-05,
+      "loss": 0.2746,
+      "step": 5882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.095092523952821e-05,
+      "loss": 0.8888,
+      "step": 5883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0948203505179098e-05,
+      "loss": 1.0542,
+      "step": 5884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.094548153843177e-05,
+      "loss": 0.7318,
+      "step": 5885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0942759339392553e-05,
+      "loss": 1.2604,
+      "step": 5886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.094003690816782e-05,
+      "loss": 1.3757,
+      "step": 5887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.093731424486393e-05,
+      "loss": 0.3098,
+      "step": 5888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0934591349587272e-05,
+      "loss": 0.7095,
+      "step": 5889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.093186822244421e-05,
+      "loss": 0.8534,
+      "step": 5890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0929144863541148e-05,
+      "loss": 0.7573,
+      "step": 5891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0926421272984486e-05,
+      "loss": 0.5753,
+      "step": 5892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0923697450880648e-05,
+      "loss": 0.9218,
+      "step": 5893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0920973397336035e-05,
+      "loss": 0.6727,
+      "step": 5894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0918249112457088e-05,
+      "loss": 0.5298,
+      "step": 5895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.091552459635024e-05,
+      "loss": 0.4531,
+      "step": 5896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.091279984912195e-05,
+      "loss": 1.0233,
+      "step": 5897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.091007487087866e-05,
+      "loss": 0.995,
+      "step": 5898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.090734966172684e-05,
+      "loss": 1.5688,
+      "step": 5899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0904624221772974e-05,
+      "loss": 0.3021,
+      "step": 5900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0901898551123537e-05,
+      "loss": 0.5693,
+      "step": 5901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0899172649885024e-05,
+      "loss": 0.2046,
+      "step": 5902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0896446518163936e-05,
+      "loss": 0.1435,
+      "step": 5903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0893720156066787e-05,
+      "loss": 0.3604,
+      "step": 5904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0890993563700096e-05,
+      "loss": 1.2849,
+      "step": 5905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0888266741170385e-05,
+      "loss": 0.3237,
+      "step": 5906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.08855396885842e-05,
+      "loss": 1.471,
+      "step": 5907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0882812406048085e-05,
+      "loss": 0.1348,
+      "step": 5908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0880084893668596e-05,
+      "loss": 0.7021,
+      "step": 5909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0877357151552297e-05,
+      "loss": 0.25,
+      "step": 5910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0874629179805763e-05,
+      "loss": 0.2678,
+      "step": 5911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0871900978535574e-05,
+      "loss": 0.3945,
+      "step": 5912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.086917254784833e-05,
+      "loss": 0.1473,
+      "step": 5913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0866443887850616e-05,
+      "loss": 0.4071,
+      "step": 5914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.086371499864906e-05,
+      "loss": 1.0254,
+      "step": 5915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0860985880350268e-05,
+      "loss": 0.8477,
+      "step": 5916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.085825653306087e-05,
+      "loss": 1.9803,
+      "step": 5917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0855526956887508e-05,
+      "loss": 1.425,
+      "step": 5918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.085279715193682e-05,
+      "loss": 0.3097,
+      "step": 5919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.085006711831546e-05,
+      "loss": 0.6458,
+      "step": 5920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.08473368561301e-05,
+      "loss": 0.57,
+      "step": 5921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0844606365487405e-05,
+      "loss": 1.2206,
+      "step": 5922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0841875646494058e-05,
+      "loss": 1.9324,
+      "step": 5923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.083914469925675e-05,
+      "loss": 1.3569,
+      "step": 5924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.083641352388218e-05,
+      "loss": 1.3394,
+      "step": 5925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0833682120477055e-05,
+      "loss": 0.6292,
+      "step": 5926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.083095048914809e-05,
+      "loss": 0.7619,
+      "step": 5927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0828218630002012e-05,
+      "loss": 0.6211,
+      "step": 5928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.082548654314556e-05,
+      "loss": 0.4352,
+      "step": 5929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0822754228685473e-05,
+      "loss": 1.1693,
+      "step": 5930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.08200216867285e-05,
+      "loss": 0.8704,
+      "step": 5931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.081728891738141e-05,
+      "loss": 0.3939,
+      "step": 5932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.081455592075097e-05,
+      "loss": 1.9916,
+      "step": 5933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.081182269694396e-05,
+      "loss": 0.478,
+      "step": 5934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0809089246067164e-05,
+      "loss": 1.2165,
+      "step": 5935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0806355568227382e-05,
+      "loss": 0.7848,
+      "step": 5936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0803621663531417e-05,
+      "loss": 0.5561,
+      "step": 5937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0800887532086088e-05,
+      "loss": 0.2263,
+      "step": 5938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0798153173998218e-05,
+      "loss": 0.8288,
+      "step": 5939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0795418589374635e-05,
+      "loss": 1.6172,
+      "step": 5940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.079268377832218e-05,
+      "loss": 0.5537,
+      "step": 5941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.078994874094771e-05,
+      "loss": 0.5642,
+      "step": 5942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0787213477358074e-05,
+      "loss": 1.0415,
+      "step": 5943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.078447798766015e-05,
+      "loss": 0.7512,
+      "step": 5944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0781742271960805e-05,
+      "loss": 0.7063,
+      "step": 5945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.077900633036693e-05,
+      "loss": 1.3984,
+      "step": 5946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.077627016298542e-05,
+      "loss": 1.1472,
+      "step": 5947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0773533769923175e-05,
+      "loss": 0.9428,
+      "step": 5948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0770797151287104e-05,
+      "loss": 0.7494,
+      "step": 5949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.076806030718413e-05,
+      "loss": 0.3067,
+      "step": 5950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0765323237721188e-05,
+      "loss": 0.6525,
+      "step": 5951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0762585943005215e-05,
+      "loss": 1.1028,
+      "step": 5952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0759848423143148e-05,
+      "loss": 0.615,
+      "step": 5953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0757110678241946e-05,
+      "loss": 0.6882,
+      "step": 5954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.075437270840858e-05,
+      "loss": 0.7185,
+      "step": 5955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.075163451375003e-05,
+      "loss": 0.5901,
+      "step": 5956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0748896094373256e-05,
+      "loss": 0.5685,
+      "step": 5957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0746157450385264e-05,
+      "loss": 0.9947,
+      "step": 5958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0743418581893052e-05,
+      "loss": 1.339,
+      "step": 5959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0740679489003632e-05,
+      "loss": 0.528,
+      "step": 5960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0737940171824007e-05,
+      "loss": 0.1892,
+      "step": 5961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0735200630461216e-05,
+      "loss": 1.3899,
+      "step": 5962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0732460865022293e-05,
+      "loss": 0.2421,
+      "step": 5963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0729720875614282e-05,
+      "loss": 0.2535,
+      "step": 5964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.072698066234423e-05,
+      "loss": 1.6797,
+      "step": 5965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0724240225319194e-05,
+      "loss": 0.418,
+      "step": 5966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0721499564646253e-05,
+      "loss": 0.1498,
+      "step": 5967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0718758680432487e-05,
+      "loss": 0.1243,
+      "step": 5968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0716017572784976e-05,
+      "loss": 0.0869,
+      "step": 5969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.071327624181082e-05,
+      "loss": 0.2077,
+      "step": 5970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.071053468761712e-05,
+      "loss": 0.2106,
+      "step": 5971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0707792910310994e-05,
+      "loss": 0.0924,
+      "step": 5972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.070505090999956e-05,
+      "loss": 0.2021,
+      "step": 5973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0702308686789952e-05,
+      "loss": 0.3199,
+      "step": 5974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0699566240789303e-05,
+      "loss": 1.0013,
+      "step": 5975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0696823572104776e-05,
+      "loss": 1.711,
+      "step": 5976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0694080680843512e-05,
+      "loss": 0.1056,
+      "step": 5977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0691337567112686e-05,
+      "loss": 0.2872,
+      "step": 5978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.068859423101947e-05,
+      "loss": 1.2114,
+      "step": 5979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.068585067267104e-05,
+      "loss": 1.298,
+      "step": 5980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.06831068921746e-05,
+      "loss": 0.5436,
+      "step": 5981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0680362889637345e-05,
+      "loss": 0.3042,
+      "step": 5982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0677618665166482e-05,
+      "loss": 1.8288,
+      "step": 5983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.067487421886923e-05,
+      "loss": 0.7156,
+      "step": 5984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0672129550852814e-05,
+      "loss": 0.6605,
+      "step": 5985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0669384661224475e-05,
+      "loss": 0.4151,
+      "step": 5986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.066663955009145e-05,
+      "loss": 0.2598,
+      "step": 5987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0663894217560996e-05,
+      "loss": 1.2389,
+      "step": 5988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0661148663740366e-05,
+      "loss": 0.479,
+      "step": 5989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0658402888736842e-05,
+      "loss": 1.4233,
+      "step": 5990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.065565689265769e-05,
+      "loss": 1.2685,
+      "step": 5991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.065291067561021e-05,
+      "loss": 0.6017,
+      "step": 5992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0650164237701686e-05,
+      "loss": 0.9448,
+      "step": 5993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.064741757903943e-05,
+      "loss": 0.7675,
+      "step": 5994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.064467069973075e-05,
+      "loss": 0.6002,
+      "step": 5995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0641923599882966e-05,
+      "loss": 1.0091,
+      "step": 5996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0639176279603412e-05,
+      "loss": 1.6116,
+      "step": 5997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.063642873899943e-05,
+      "loss": 0.4637,
+      "step": 5998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.063368097817836e-05,
+      "loss": 0.4659,
+      "step": 5999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.063093299724756e-05,
+      "loss": 0.4703,
+      "step": 6000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0628184796314396e-05,
+      "loss": 0.4998,
+      "step": 6001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.062543637548624e-05,
+      "loss": 0.0463,
+      "step": 6002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0622687734870477e-05,
+      "loss": 1.1784,
+      "step": 6003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0619938874574494e-05,
+      "loss": 0.1952,
+      "step": 6004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0617189794705687e-05,
+      "loss": 0.1448,
+      "step": 6005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0614440495371468e-05,
+      "loss": 0.2853,
+      "step": 6006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.061169097667925e-05,
+      "loss": 0.2529,
+      "step": 6007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0608941238736464e-05,
+      "loss": 0.073,
+      "step": 6008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.060619128165054e-05,
+      "loss": 0.2527,
+      "step": 6009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0603441105528916e-05,
+      "loss": 0.9398,
+      "step": 6010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.060069071047904e-05,
+      "loss": 0.1847,
+      "step": 6011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0597940096608378e-05,
+      "loss": 0.1686,
+      "step": 6012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0595189264024397e-05,
+      "loss": 1.5157,
+      "step": 6013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.059243821283457e-05,
+      "loss": 1.8119,
+      "step": 6014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0589686943146378e-05,
+      "loss": 1.0574,
+      "step": 6015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.058693545506732e-05,
+      "loss": 1.8285,
+      "step": 6016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0584183748704896e-05,
+      "loss": 0.532,
+      "step": 6017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.058143182416662e-05,
+      "loss": 0.266,
+      "step": 6018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0578679681560002e-05,
+      "loss": 0.117,
+      "step": 6019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.057592732099257e-05,
+      "loss": 0.4387,
+      "step": 6020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0573174742571864e-05,
+      "loss": 0.3779,
+      "step": 6021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0570421946405436e-05,
+      "loss": 1.9478,
+      "step": 6022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.056766893260082e-05,
+      "loss": 0.7444,
+      "step": 6023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.056491570126559e-05,
+      "loss": 0.8271,
+      "step": 6024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0562162252507312e-05,
+      "loss": 0.2895,
+      "step": 6025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0559408586433568e-05,
+      "loss": 0.991,
+      "step": 6026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0556654703151936e-05,
+      "loss": 1.7337,
+      "step": 6027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0553900602770014e-05,
+      "loss": 0.7299,
+      "step": 6028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0551146285395415e-05,
+      "loss": 0.7214,
+      "step": 6029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0548391751135743e-05,
+      "loss": 0.5539,
+      "step": 6030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0545637000098618e-05,
+      "loss": 0.7153,
+      "step": 6031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0542882032391672e-05,
+      "loss": 0.2774,
+      "step": 6032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0540126848122534e-05,
+      "loss": 0.6363,
+      "step": 6033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0537371447398867e-05,
+      "loss": 1.49,
+      "step": 6034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0534615830328308e-05,
+      "loss": 0.5397,
+      "step": 6035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.053185999701853e-05,
+      "loss": 0.1902,
+      "step": 6036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0529103947577196e-05,
+      "loss": 0.5229,
+      "step": 6037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.0526347682112003e-05,
+      "loss": 0.5291,
+      "step": 6038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.052359120073062e-05,
+      "loss": 0.8294,
+      "step": 6039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.052083450354075e-05,
+      "loss": 0.5566,
+      "step": 6040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.05180775906501e-05,
+      "loss": 0.8639,
+      "step": 6041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.051532046216638e-05,
+      "loss": 0.5889,
+      "step": 6042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.051256311819732e-05,
+      "loss": 1.4536,
+      "step": 6043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0509805558850643e-05,
+      "loss": 1.0353,
+      "step": 6044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.050704778423409e-05,
+      "loss": 0.6858,
+      "step": 6045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0504289794455403e-05,
+      "loss": 0.4346,
+      "step": 6046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0501531589622347e-05,
+      "loss": 0.6111,
+      "step": 6047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0498773169842676e-05,
+      "loss": 0.7303,
+      "step": 6048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.049601453522417e-05,
+      "loss": 0.2322,
+      "step": 6049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.049325568587461e-05,
+      "loss": 0.2933,
+      "step": 6050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.049049662190178e-05,
+      "loss": 0.5447,
+      "step": 6051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.048773734341348e-05,
+      "loss": 2.0231,
+      "step": 6052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0484977850517514e-05,
+      "loss": 0.3837,
+      "step": 6053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0482218143321703e-05,
+      "loss": 0.3185,
+      "step": 6054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.047945822193386e-05,
+      "loss": 1.7401,
+      "step": 6055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.047669808646182e-05,
+      "loss": 0.2587,
+      "step": 6056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.047393773701343e-05,
+      "loss": 1.0619,
+      "step": 6057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0471177173696522e-05,
+      "loss": 0.2352,
+      "step": 6058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0468416396618962e-05,
+      "loss": 0.3027,
+      "step": 6059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.046565540588862e-05,
+      "loss": 1.7645,
+      "step": 6060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0462894201613358e-05,
+      "loss": 1.2588,
+      "step": 6061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.046013278390106e-05,
+      "loss": 0.0577,
+      "step": 6062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.045737115285962e-05,
+      "loss": 1.0647,
+      "step": 6063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.045460930859693e-05,
+      "loss": 0.3782,
+      "step": 6064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0451847251220897e-05,
+      "loss": 0.4873,
+      "step": 6065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.044908498083944e-05,
+      "loss": 0.037,
+      "step": 6066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0446322497560477e-05,
+      "loss": 0.0306,
+      "step": 6067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0443559801491943e-05,
+      "loss": 0.0776,
+      "step": 6068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0440796892741774e-05,
+      "loss": 1.1984,
+      "step": 6069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0438033771417923e-05,
+      "loss": 0.2844,
+      "step": 6070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.043527043762834e-05,
+      "loss": 1.3254,
+      "step": 6071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0432506891480992e-05,
+      "loss": 0.7204,
+      "step": 6072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0429743133083848e-05,
+      "loss": 0.5637,
+      "step": 6073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0426979162544893e-05,
+      "loss": 1.3145,
+      "step": 6074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.042421497997212e-05,
+      "loss": 1.8703,
+      "step": 6075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0421450585473518e-05,
+      "loss": 1.2812,
+      "step": 6076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0418685979157094e-05,
+      "loss": 0.6658,
+      "step": 6077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.041592116113087e-05,
+      "loss": 0.8046,
+      "step": 6078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0413156131502864e-05,
+      "loss": 0.5659,
+      "step": 6079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0410390890381105e-05,
+      "loss": 0.7389,
+      "step": 6080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0407625437873627e-05,
+      "loss": 0.9973,
+      "step": 6081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0404859774088485e-05,
+      "loss": 0.6329,
+      "step": 6082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0402093899133734e-05,
+      "loss": 0.6804,
+      "step": 6083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0399327813117437e-05,
+      "loss": 0.6555,
+      "step": 6084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0396561516147662e-05,
+      "loss": 1.233,
+      "step": 6085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0393795008332488e-05,
+      "loss": 0.2727,
+      "step": 6086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0391028289780012e-05,
+      "loss": 0.6491,
+      "step": 6087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0388261360598327e-05,
+      "loss": 1.2651,
+      "step": 6088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0385494220895534e-05,
+      "loss": 1.0333,
+      "step": 6089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0382726870779748e-05,
+      "loss": 1.2573,
+      "step": 6090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0379959310359088e-05,
+      "loss": 0.9448,
+      "step": 6091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.037719153974169e-05,
+      "loss": 0.3759,
+      "step": 6092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.037442355903569e-05,
+      "loss": 0.6554,
+      "step": 6093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0371655368349228e-05,
+      "loss": 0.9442,
+      "step": 6094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.036888696779046e-05,
+      "loss": 0.5678,
+      "step": 6095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0366118357467556e-05,
+      "loss": 0.0935,
+      "step": 6096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0363349537488676e-05,
+      "loss": 0.6537,
+      "step": 6097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0360580507962008e-05,
+      "loss": 0.4935,
+      "step": 6098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0357811268995726e-05,
+      "loss": 0.688,
+      "step": 6099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0355041820698042e-05,
+      "loss": 0.4929,
+      "step": 6100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0352272163177145e-05,
+      "loss": 1.4892,
+      "step": 6101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0349502296541255e-05,
+      "loss": 1.07,
+      "step": 6102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0346732220898587e-05,
+      "loss": 0.4116,
+      "step": 6103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0343961936357366e-05,
+      "loss": 0.5506,
+      "step": 6104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0341191443025838e-05,
+      "loss": 0.3728,
+      "step": 6105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0338420741012242e-05,
+      "loss": 0.3458,
+      "step": 6106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0335649830424826e-05,
+      "loss": 0.1396,
+      "step": 6107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0332878711371855e-05,
+      "loss": 1.3652,
+      "step": 6108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0330107383961596e-05,
+      "loss": 0.287,
+      "step": 6109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0327335848302324e-05,
+      "loss": 1.8197,
+      "step": 6110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.032456410450233e-05,
+      "loss": 0.2613,
+      "step": 6111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.03217921526699e-05,
+      "loss": 1.3162,
+      "step": 6112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.031901999291334e-05,
+      "loss": 1.3283,
+      "step": 6113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0316247625340954e-05,
+      "loss": 0.1844,
+      "step": 6114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0313475050061063e-05,
+      "loss": 0.3224,
+      "step": 6115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0310702267181992e-05,
+      "loss": 0.2128,
+      "step": 6116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0307929276812073e-05,
+      "loss": 1.9267,
+      "step": 6117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0305156079059653e-05,
+      "loss": 0.5471,
+      "step": 6118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.030238267403307e-05,
+      "loss": 0.3998,
+      "step": 6119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0299609061840696e-05,
+      "loss": 0.3358,
+      "step": 6120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0296835242590888e-05,
+      "loss": 1.7359,
+      "step": 6121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0294061216392018e-05,
+      "loss": 0.1719,
+      "step": 6122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.029128698335248e-05,
+      "loss": 0.9597,
+      "step": 6123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0288512543580653e-05,
+      "loss": 0.5819,
+      "step": 6124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.028573789718494e-05,
+      "loss": 1.0306,
+      "step": 6125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0282963044273742e-05,
+      "loss": 0.7618,
+      "step": 6126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0280187984955482e-05,
+      "loss": 0.6393,
+      "step": 6127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0277412719338573e-05,
+      "loss": 1.5162,
+      "step": 6128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0274637247531456e-05,
+      "loss": 0.1228,
+      "step": 6129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0271861569642564e-05,
+      "loss": 1.1806,
+      "step": 6130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.026908568578034e-05,
+      "loss": 0.2022,
+      "step": 6131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.026630959605324e-05,
+      "loss": 0.3062,
+      "step": 6132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0263533300569737e-05,
+      "loss": 0.3457,
+      "step": 6133
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.026075679943829e-05,
+      "loss": 1.7599,
+      "step": 6134
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0257980092767377e-05,
+      "loss": 0.6367,
+      "step": 6135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0255203180665497e-05,
+      "loss": 0.7068,
+      "step": 6136
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0252426063241134e-05,
+      "loss": 0.7011,
+      "step": 6137
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0249648740602795e-05,
+      "loss": 0.6363,
+      "step": 6138
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0246871212858988e-05,
+      "loss": 1.0218,
+      "step": 6139
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0244093480118236e-05,
+      "loss": 1.117,
+      "step": 6140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.024131554248906e-05,
+      "loss": 0.7399,
+      "step": 6141
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0238537400080004e-05,
+      "loss": 0.8992,
+      "step": 6142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0235759052999602e-05,
+      "loss": 0.8806,
+      "step": 6143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.023298050135641e-05,
+      "loss": 0.453,
+      "step": 6144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0230201745258985e-05,
+      "loss": 0.5292,
+      "step": 6145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0227422784815896e-05,
+      "loss": 0.7461,
+      "step": 6146
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0224643620135718e-05,
+      "loss": 0.2414,
+      "step": 6147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.022186425132703e-05,
+      "loss": 0.6182,
+      "step": 6148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0219084678498423e-05,
+      "loss": 0.3767,
+      "step": 6149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.02163049017585e-05,
+      "loss": 0.3497,
+      "step": 6150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0213524921215867e-05,
+      "loss": 0.9093,
+      "step": 6151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0210744736979132e-05,
+      "loss": 1.1207,
+      "step": 6152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0207964349156927e-05,
+      "loss": 0.3285,
+      "step": 6153
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0205183757857875e-05,
+      "loss": 0.6166,
+      "step": 6154
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0202402963190625e-05,
+      "loss": 1.8385,
+      "step": 6155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.019962196526381e-05,
+      "loss": 1.0556,
+      "step": 6156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0196840764186097e-05,
+      "loss": 1.3538,
+      "step": 6157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.019405936006614e-05,
+      "loss": 0.0828,
+      "step": 6158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.019127775301261e-05,
+      "loss": 1.2154,
+      "step": 6159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0188495943134193e-05,
+      "loss": 0.1368,
+      "step": 6160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0185713930539562e-05,
+      "loss": 1.8495,
+      "step": 6161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.018293171533742e-05,
+      "loss": 0.3852,
+      "step": 6162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0180149297636476e-05,
+      "loss": 0.3054,
+      "step": 6163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0177366677545427e-05,
+      "loss": 0.1377,
+      "step": 6164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0174583855172993e-05,
+      "loss": 0.2235,
+      "step": 6165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.0171800830627903e-05,
+      "loss": 1.5115,
+      "step": 6166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0169017604018892e-05,
+      "loss": 1.2774,
+      "step": 6167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.01662341754547e-05,
+      "loss": 0.845,
+      "step": 6168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0163450545044073e-05,
+      "loss": 0.8094,
+      "step": 6169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.016066671289577e-05,
+      "loss": 1.4855,
+      "step": 6170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0157882679118562e-05,
+      "loss": 0.5595,
+      "step": 6171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0155098443821213e-05,
+      "loss": 1.5551,
+      "step": 6172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.015231400711251e-05,
+      "loss": 0.6805,
+      "step": 6173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.014952936910124e-05,
+      "loss": 0.9272,
+      "step": 6174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0146744529896204e-05,
+      "loss": 0.7128,
+      "step": 6175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.01439594896062e-05,
+      "loss": 1.0997,
+      "step": 6176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0141174248340043e-05,
+      "loss": 1.1456,
+      "step": 6177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.013838880620656e-05,
+      "loss": 1.1261,
+      "step": 6178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.013560316331456e-05,
+      "loss": 1.9386,
+      "step": 6179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0132817319772903e-05,
+      "loss": 0.3602,
+      "step": 6180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0130031275690414e-05,
+      "loss": 0.413,
+      "step": 6181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0127245031175955e-05,
+      "loss": 0.8805,
+      "step": 6182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.012445858633838e-05,
+      "loss": 1.1057,
+      "step": 6183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.012167194128656e-05,
+      "loss": 0.4076,
+      "step": 6184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0118885096129366e-05,
+      "loss": 0.4309,
+      "step": 6185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.011609805097569e-05,
+      "loss": 1.2119,
+      "step": 6186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0113310805934414e-05,
+      "loss": 0.3422,
+      "step": 6187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0110523361114438e-05,
+      "loss": 0.4577,
+      "step": 6188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0107735716624666e-05,
+      "loss": 0.9338,
+      "step": 6189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0104947872574023e-05,
+      "loss": 1.1945,
+      "step": 6190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0102159829071417e-05,
+      "loss": 1.0082,
+      "step": 6191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0099371586225787e-05,
+      "loss": 0.7232,
+      "step": 6192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.009658314414607e-05,
+      "loss": 0.76,
+      "step": 6193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0093794502941206e-05,
+      "loss": 1.1773,
+      "step": 6194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.009100566272015e-05,
+      "loss": 1.0688,
+      "step": 6195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.008821662359187e-05,
+      "loss": 0.4312,
+      "step": 6196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0085427385665325e-05,
+      "loss": 0.4723,
+      "step": 6197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0082637949049495e-05,
+      "loss": 0.3297,
+      "step": 6198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0079848313853363e-05,
+      "loss": 0.4428,
+      "step": 6199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0077058480185925e-05,
+      "loss": 0.8922,
+      "step": 6200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.007426844815618e-05,
+      "loss": 0.3337,
+      "step": 6201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.007147821787313e-05,
+      "loss": 0.2607,
+      "step": 6202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0068687789445794e-05,
+      "loss": 0.3143,
+      "step": 6203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0065897162983192e-05,
+      "loss": 0.4806,
+      "step": 6204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0063106338594364e-05,
+      "loss": 0.3995,
+      "step": 6205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0060315316388337e-05,
+      "loss": 1.6922,
+      "step": 6206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0057524096474164e-05,
+      "loss": 1.8461,
+      "step": 6207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0054732678960896e-05,
+      "loss": 0.1992,
+      "step": 6208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0051941063957597e-05,
+      "loss": 0.4968,
+      "step": 6209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.004914925157333e-05,
+      "loss": 0.2285,
+      "step": 6210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0046357241917178e-05,
+      "loss": 0.746,
+      "step": 6211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0043565035098226e-05,
+      "loss": 0.3382,
+      "step": 6212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.004077263122557e-05,
+      "loss": 1.2843,
+      "step": 6213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.00379800304083e-05,
+      "loss": 0.2788,
+      "step": 6214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0035187232755525e-05,
+      "loss": 0.7508,
+      "step": 6215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0032394238376368e-05,
+      "loss": 1.1428,
+      "step": 6216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0029601047379948e-05,
+      "loss": 0.8595,
+      "step": 6217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.00268076598754e-05,
+      "loss": 0.4424,
+      "step": 6218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.002401407597185e-05,
+      "loss": 1.1017,
+      "step": 6219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.002122029577846e-05,
+      "loss": 0.3287,
+      "step": 6220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0018426319404377e-05,
+      "loss": 0.3098,
+      "step": 6221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.001563214695876e-05,
+      "loss": 1.4268,
+      "step": 6222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0012837778550778e-05,
+      "loss": 1.5768,
+      "step": 6223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0010043214289616e-05,
+      "loss": 0.269,
+      "step": 6224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0007248454284453e-05,
+      "loss": 1.0246,
+      "step": 6225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0004453498644477e-05,
+      "loss": 0.8297,
+      "step": 6226
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0001658347478892e-05,
+      "loss": 0.7599,
+      "step": 6227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9998863000896908e-05,
+      "loss": 0.5559,
+      "step": 6228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9996067459007736e-05,
+      "loss": 0.3943,
+      "step": 6229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9993271721920605e-05,
+      "loss": 0.1876,
+      "step": 6230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9990475789744737e-05,
+      "loss": 0.9178,
+      "step": 6231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.998767966258937e-05,
+      "loss": 1.1142,
+      "step": 6232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.998488334056376e-05,
+      "loss": 1.1446,
+      "step": 6233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9982086823777154e-05,
+      "loss": 0.5404,
+      "step": 6234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9979290112338805e-05,
+      "loss": 0.5371,
+      "step": 6235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.997649320635799e-05,
+      "loss": 0.4468,
+      "step": 6236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9973696105943988e-05,
+      "loss": 0.9507,
+      "step": 6237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.997089881120608e-05,
+      "loss": 0.5985,
+      "step": 6238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.996810132225355e-05,
+      "loss": 0.5317,
+      "step": 6239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9965303639195702e-05,
+      "loss": 0.3398,
+      "step": 6240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9962505762141846e-05,
+      "loss": 0.6088,
+      "step": 6241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9959707691201295e-05,
+      "loss": 1.2125,
+      "step": 6242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9956909426483362e-05,
+      "loss": 0.6874,
+      "step": 6243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9954110968097384e-05,
+      "loss": 0.9712,
+      "step": 6244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.99513123161527e-05,
+      "loss": 0.3717,
+      "step": 6245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9948513470758648e-05,
+      "loss": 0.5455,
+      "step": 6246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.994571443202458e-05,
+      "loss": 1.187,
+      "step": 6247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9942915200059857e-05,
+      "loss": 0.2991,
+      "step": 6248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9940115774973852e-05,
+      "loss": 0.7988,
+      "step": 6249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9937316156875926e-05,
+      "loss": 0.1512,
+      "step": 6250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9934516345875473e-05,
+      "loss": 0.2441,
+      "step": 6251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9931716342081877e-05,
+      "loss": 0.939,
+      "step": 6252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9928916145604535e-05,
+      "loss": 0.3279,
+      "step": 6253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9926115756552854e-05,
+      "loss": 1.255,
+      "step": 6254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.992331517503625e-05,
+      "loss": 0.1266,
+      "step": 6255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.992051440116413e-05,
+      "loss": 0.1341,
+      "step": 6256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.991771343504593e-05,
+      "loss": 0.135,
+      "step": 6257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9914912276791083e-05,
+      "loss": 0.254,
+      "step": 6258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9912110926509035e-05,
+      "loss": 0.5483,
+      "step": 6259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9909309384309227e-05,
+      "loss": 0.9133,
+      "step": 6260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9906507650301125e-05,
+      "loss": 0.4149,
+      "step": 6261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.990370572459419e-05,
+      "loss": 0.4542,
+      "step": 6262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9900903607297893e-05,
+      "loss": 0.1863,
+      "step": 6263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9898101298521713e-05,
+      "loss": 0.1572,
+      "step": 6264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9895298798375143e-05,
+      "loss": 1.7566,
+      "step": 6265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9892496106967673e-05,
+      "loss": 2.0633,
+      "step": 6266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.98896932244088e-05,
+      "loss": 0.4831,
+      "step": 6267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9886890150808043e-05,
+      "loss": 0.2759,
+      "step": 6268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9884086886274913e-05,
+      "loss": 1.5272,
+      "step": 6269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9881283430918938e-05,
+      "loss": 0.4877,
+      "step": 6270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.987847978484965e-05,
+      "loss": 0.6369,
+      "step": 6271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.987567594817658e-05,
+      "loss": 0.9366,
+      "step": 6272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.987287192100929e-05,
+      "loss": 0.0496,
+      "step": 6273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.987006770345732e-05,
+      "loss": 0.6729,
+      "step": 6274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.986726329563024e-05,
+      "loss": 1.284,
+      "step": 6275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9864458697637614e-05,
+      "loss": 0.4619,
+      "step": 6276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9861653909589022e-05,
+      "loss": 0.2288,
+      "step": 6277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9858848931594047e-05,
+      "loss": 0.2251,
+      "step": 6278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9856043763762282e-05,
+      "loss": 0.2264,
+      "step": 6279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9853238406203326e-05,
+      "loss": 0.7308,
+      "step": 6280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9850432859026778e-05,
+      "loss": 0.9015,
+      "step": 6281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9847627122342257e-05,
+      "loss": 0.7509,
+      "step": 6282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9844821196259385e-05,
+      "loss": 0.8508,
+      "step": 6283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.98420150808878e-05,
+      "loss": 1.0026,
+      "step": 6284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983920877633711e-05,
+      "loss": 0.6936,
+      "step": 6285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9836402282716986e-05,
+      "loss": 0.6562,
+      "step": 6286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9833595600137065e-05,
+      "loss": 0.6091,
+      "step": 6287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9830788728707016e-05,
+      "loss": 1.153,
+      "step": 6288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9827981668536483e-05,
+      "loss": 0.8048,
+      "step": 6289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.982517441973516e-05,
+      "loss": 1.4253,
+      "step": 6290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9822366982412718e-05,
+      "loss": 0.2483,
+      "step": 6291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9819559356678852e-05,
+      "loss": 0.9152,
+      "step": 6292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.981675154264324e-05,
+      "loss": 0.8481,
+      "step": 6293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.98139435404156e-05,
+      "loss": 0.1905,
+      "step": 6294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9811135350105636e-05,
+      "loss": 0.8092,
+      "step": 6295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9808326971823068e-05,
+      "loss": 1.0791,
+      "step": 6296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9805518405677615e-05,
+      "loss": 0.6626,
+      "step": 6297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.980270965177901e-05,
+      "loss": 0.5544,
+      "step": 6298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9799900710236994e-05,
+      "loss": 0.4541,
+      "step": 6299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9797091581161315e-05,
+      "loss": 0.3385,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.979428226466172e-05,
+      "loss": 1.2116,
+      "step": 6301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9791472760847977e-05,
+      "loss": 0.4964,
+      "step": 6302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9788663069829848e-05,
+      "loss": 1.1324,
+      "step": 6303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9785853191717117e-05,
+      "loss": 1.5009,
+      "step": 6304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9783043126619558e-05,
+      "loss": 0.7627,
+      "step": 6305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9780232874646965e-05,
+      "loss": 0.3191,
+      "step": 6306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.977742243590913e-05,
+      "loss": 0.2024,
+      "step": 6307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.977461181051587e-05,
+      "loss": 1.2301,
+      "step": 6308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9771800998576987e-05,
+      "loss": 0.9237,
+      "step": 6309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9768990000202305e-05,
+      "loss": 0.4824,
+      "step": 6310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9766178815501648e-05,
+      "loss": 0.4456,
+      "step": 6311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9763367444584847e-05,
+      "loss": 0.5689,
+      "step": 6312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.976055588756175e-05,
+      "loss": 2.1123,
+      "step": 6313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.97577441445422e-05,
+      "loss": 1.6936,
+      "step": 6314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975493221563606e-05,
+      "loss": 0.1846,
+      "step": 6315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9752120100953183e-05,
+      "loss": 0.372,
+      "step": 6316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749307800603446e-05,
+      "loss": 0.793,
+      "step": 6317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746495314696722e-05,
+      "loss": 0.5115,
+      "step": 6318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743682643342902e-05,
+      "loss": 0.3431,
+      "step": 6319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740869786651873e-05,
+      "loss": 1.7475,
+      "step": 6320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738056744733535e-05,
+      "loss": 0.1672,
+      "step": 6321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9735243517697794e-05,
+      "loss": 0.5477,
+      "step": 6322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9732430105654566e-05,
+      "loss": 0.223,
+      "step": 6323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9729616508713774e-05,
+      "loss": 1.9489,
+      "step": 6324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9726802726985337e-05,
+      "loss": 1.2067,
+      "step": 6325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9723988760579198e-05,
+      "loss": 0.7323,
+      "step": 6326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.97211746096053e-05,
+      "loss": 0.5636,
+      "step": 6327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.971836027417359e-05,
+      "loss": 0.7601,
+      "step": 6328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9715545754394025e-05,
+      "loss": 0.863,
+      "step": 6329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9712731050376566e-05,
+      "loss": 1.0326,
+      "step": 6330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.970991616223119e-05,
+      "loss": 0.5154,
+      "step": 6331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9707101090067875e-05,
+      "loss": 0.6388,
+      "step": 6332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9704285833996605e-05,
+      "loss": 2.329,
+      "step": 6333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9701470394127374e-05,
+      "loss": 0.4527,
+      "step": 6334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9698654770570178e-05,
+      "loss": 0.6863,
+      "step": 6335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9695838963435033e-05,
+      "loss": 0.6692,
+      "step": 6336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9693022972831942e-05,
+      "loss": 1.2586,
+      "step": 6337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9690206798870938e-05,
+      "loss": 1.0349,
+      "step": 6338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9687390441662047e-05,
+      "loss": 1.4681,
+      "step": 6339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9684573901315295e-05,
+      "loss": 1.0833,
+      "step": 6340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.968175717794074e-05,
+      "loss": 1.0801,
+      "step": 6341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9678940271648425e-05,
+      "loss": 0.3669,
+      "step": 6342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.96761231825484e-05,
+      "loss": 1.1533,
+      "step": 6343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9673305910750744e-05,
+      "loss": 0.6822,
+      "step": 6344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.967048845636552e-05,
+      "loss": 0.847,
+      "step": 6345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.966767081950281e-05,
+      "loss": 0.3103,
+      "step": 6346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9664853000272694e-05,
+      "loss": 0.2578,
+      "step": 6347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9662034998785275e-05,
+      "loss": 0.6862,
+      "step": 6348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9659216815150648e-05,
+      "loss": 0.489,
+      "step": 6349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9656398449478923e-05,
+      "loss": 1.023,
+      "step": 6350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9653579901880204e-05,
+      "loss": 1.6268,
+      "step": 6351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9650761172464624e-05,
+      "loss": 0.3273,
+      "step": 6352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9647942261342306e-05,
+      "loss": 1.2635,
+      "step": 6353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9645123168623395e-05,
+      "loss": 1.4016,
+      "step": 6354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9642303894418018e-05,
+      "loss": 0.302,
+      "step": 6355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.963948443883634e-05,
+      "loss": 0.2826,
+      "step": 6356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9636664801988507e-05,
+      "loss": 1.6541,
+      "step": 6357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.963384498398469e-05,
+      "loss": 0.6308,
+      "step": 6358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.963102498493506e-05,
+      "loss": 1.4009,
+      "step": 6359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.962820480494979e-05,
+      "loss": 1.3011,
+      "step": 6360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.962538444413907e-05,
+      "loss": 1.8584,
+      "step": 6361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9622563902613094e-05,
+      "loss": 0.545,
+      "step": 6362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9619743180482056e-05,
+      "loss": 0.0403,
+      "step": 6363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9616922277856164e-05,
+      "loss": 0.2871,
+      "step": 6364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9614101194845633e-05,
+      "loss": 0.6393,
+      "step": 6365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9611279931560685e-05,
+      "loss": 0.3574,
+      "step": 6366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.960845848811154e-05,
+      "loss": 0.4586,
+      "step": 6367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9605636864608443e-05,
+      "loss": 0.454,
+      "step": 6368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.960281506116163e-05,
+      "loss": 1.2501,
+      "step": 6369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9599993077881355e-05,
+      "loss": 1.6507,
+      "step": 6370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.959717091487787e-05,
+      "loss": 0.4822,
+      "step": 6371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9594348572261437e-05,
+      "loss": 0.4758,
+      "step": 6372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9591526050142323e-05,
+      "loss": 0.0813,
+      "step": 6373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.958870334863081e-05,
+      "loss": 1.8823,
+      "step": 6374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.958588046783718e-05,
+      "loss": 0.9792,
+      "step": 6375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9583057407871722e-05,
+      "loss": 1.4232,
+      "step": 6376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9580234168844738e-05,
+      "loss": 0.7026,
+      "step": 6377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9577410750866533e-05,
+      "loss": 0.8046,
+      "step": 6378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9574587154047414e-05,
+      "loss": 0.3656,
+      "step": 6379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9571763378497706e-05,
+      "loss": 0.6798,
+      "step": 6380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9568939424327726e-05,
+      "loss": 0.4631,
+      "step": 6381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9566115291647816e-05,
+      "loss": 0.5531,
+      "step": 6382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.956329098056831e-05,
+      "loss": 1.1458,
+      "step": 6383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.956046649119956e-05,
+      "loss": 0.9215,
+      "step": 6384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.955764182365191e-05,
+      "loss": 0.5136,
+      "step": 6385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9554816978035735e-05,
+      "loss": 0.7908,
+      "step": 6386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.955199195446139e-05,
+      "loss": 0.2639,
+      "step": 6387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9549166753039252e-05,
+      "loss": 0.5199,
+      "step": 6388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.954634137387971e-05,
+      "loss": 1.4521,
+      "step": 6389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9543515817093146e-05,
+      "loss": 1.1446,
+      "step": 6390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9540690082789952e-05,
+      "loss": 0.4624,
+      "step": 6391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9537864171080543e-05,
+      "loss": 0.7691,
+      "step": 6392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9535038082075306e-05,
+      "loss": 0.7723,
+      "step": 6393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9532211815884685e-05,
+      "loss": 0.5981,
+      "step": 6394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9529385372619086e-05,
+      "loss": 0.6139,
+      "step": 6395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9526558752388947e-05,
+      "loss": 0.6855,
+      "step": 6396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9523731955304692e-05,
+      "loss": 0.5778,
+      "step": 6397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.952090498147678e-05,
+      "loss": 0.3886,
+      "step": 6398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.951807783101565e-05,
+      "loss": 0.3541,
+      "step": 6399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9515250504031772e-05,
+      "loss": 0.6475,
+      "step": 6400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.95124230006356e-05,
+      "loss": 1.1411,
+      "step": 6401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9509595320937602e-05,
+      "loss": 0.1093,
+      "step": 6402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9506767465048272e-05,
+      "loss": 0.0966,
+      "step": 6403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9503939433078088e-05,
+      "loss": 1.2197,
+      "step": 6404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.950111122513753e-05,
+      "loss": 0.9986,
+      "step": 6405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9498282841337115e-05,
+      "loss": 0.8094,
+      "step": 6406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9495454281787347e-05,
+      "loss": 0.1023,
+      "step": 6407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9492625546598728e-05,
+      "loss": 1.4802,
+      "step": 6408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9489796635881782e-05,
+      "loss": 1.2835,
+      "step": 6409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9486967549747037e-05,
+      "loss": 0.2466,
+      "step": 6410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9484138288305024e-05,
+      "loss": 1.5353,
+      "step": 6411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9481308851666286e-05,
+      "loss": 0.1489,
+      "step": 6412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9478479239941367e-05,
+      "loss": 0.0634,
+      "step": 6413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9475649453240826e-05,
+      "loss": 1.4309,
+      "step": 6414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.947281949167522e-05,
+      "loss": 0.9299,
+      "step": 6415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9469989355355117e-05,
+      "loss": 1.2599,
+      "step": 6416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9467159044391093e-05,
+      "loss": 0.2999,
+      "step": 6417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9464328558893724e-05,
+      "loss": 0.1326,
+      "step": 6418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.94614978989736e-05,
+      "loss": 1.8983,
+      "step": 6419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9458667064741317e-05,
+      "loss": 0.19,
+      "step": 6420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9455836056307477e-05,
+      "loss": 0.573,
+      "step": 6421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.945300487378269e-05,
+      "loss": 0.5039,
+      "step": 6422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.945017351727757e-05,
+      "loss": 0.1761,
+      "step": 6423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.944734198690274e-05,
+      "loss": 0.3246,
+      "step": 6424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9444510282768824e-05,
+      "loss": 0.598,
+      "step": 6425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9441678404986463e-05,
+      "loss": 0.2818,
+      "step": 6426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9438846353666295e-05,
+      "loss": 0.2803,
+      "step": 6427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9436014128918973e-05,
+      "loss": 1.8483,
+      "step": 6428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9433181730855142e-05,
+      "loss": 1.1114,
+      "step": 6429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9430349159585483e-05,
+      "loss": 1.0178,
+      "step": 6430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9427516415220657e-05,
+      "loss": 1.6814,
+      "step": 6431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.942468349787134e-05,
+      "loss": 0.5522,
+      "step": 6432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9421850407648214e-05,
+      "loss": 0.6077,
+      "step": 6433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9419017144661967e-05,
+      "loss": 1.25,
+      "step": 6434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.94161837090233e-05,
+      "loss": 0.5305,
+      "step": 6435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.941335010084291e-05,
+      "loss": 0.9199,
+      "step": 6436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9410516320231516e-05,
+      "loss": 0.5214,
+      "step": 6437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9407682367299827e-05,
+      "loss": 0.3875,
+      "step": 6438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9404848242158567e-05,
+      "loss": 0.6078,
+      "step": 6439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9402013944918474e-05,
+      "loss": 1.2972,
+      "step": 6440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.939917947569028e-05,
+      "loss": 1.2249,
+      "step": 6441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9396344834584724e-05,
+      "loss": 0.9925,
+      "step": 6442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.939351002171256e-05,
+      "loss": 0.8085,
+      "step": 6443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9390675037184545e-05,
+      "loss": 0.9651,
+      "step": 6444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9387839881111445e-05,
+      "loss": 1.0016,
+      "step": 6445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9385004553604023e-05,
+      "loss": 0.168,
+      "step": 6446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9382169054773068e-05,
+      "loss": 1.2841,
+      "step": 6447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.937933338472936e-05,
+      "loss": 0.2276,
+      "step": 6448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9376497543583684e-05,
+      "loss": 0.3271,
+      "step": 6449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.937366153144684e-05,
+      "loss": 0.6283,
+      "step": 6450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.937082534842963e-05,
+      "loss": 1.3088,
+      "step": 6451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.936798899464287e-05,
+      "loss": 0.2236,
+      "step": 6452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9365152470197375e-05,
+      "loss": 1.2336,
+      "step": 6453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9362315775203964e-05,
+      "loss": 0.3128,
+      "step": 6454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.935947890977347e-05,
+      "loss": 1.4147,
+      "step": 6455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.935664187401674e-05,
+      "loss": 0.2438,
+      "step": 6456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9353804668044604e-05,
+      "loss": 0.1417,
+      "step": 6457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.935096729196792e-05,
+      "loss": 1.1286,
+      "step": 6458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9348129745897543e-05,
+      "loss": 0.0971,
+      "step": 6459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9345292029944337e-05,
+      "loss": 0.2226,
+      "step": 6460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.934245414421917e-05,
+      "loss": 0.1848,
+      "step": 6461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.933961608883293e-05,
+      "loss": 1.6297,
+      "step": 6462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.933677786389649e-05,
+      "loss": 0.2663,
+      "step": 6463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.933393946952074e-05,
+      "loss": 1.2104,
+      "step": 6464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.933110090581658e-05,
+      "loss": 0.6459,
+      "step": 6465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9328262172894914e-05,
+      "loss": 1.0222,
+      "step": 6466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9325423270866655e-05,
+      "loss": 0.5157,
+      "step": 6467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9322584199842712e-05,
+      "loss": 0.3665,
+      "step": 6468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9319744959934014e-05,
+      "loss": 1.1398,
+      "step": 6469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.931690555125149e-05,
+      "loss": 1.7954,
+      "step": 6470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9314065973906082e-05,
+      "loss": 0.2522,
+      "step": 6471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9311226228008723e-05,
+      "loss": 0.5731,
+      "step": 6472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9308386313670366e-05,
+      "loss": 2.1828,
+      "step": 6473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.930554623100197e-05,
+      "loss": 0.0843,
+      "step": 6474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9302705980114503e-05,
+      "loss": 0.797,
+      "step": 6475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.929986556111892e-05,
+      "loss": 0.2876,
+      "step": 6476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.929702497412621e-05,
+      "loss": 0.2435,
+      "step": 6477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9294184219247352e-05,
+      "loss": 0.4565,
+      "step": 6478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9291343296593334e-05,
+      "loss": 1.4376,
+      "step": 6479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.928850220627515e-05,
+      "loss": 1.1188,
+      "step": 6480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9285660948403805e-05,
+      "loss": 0.5839,
+      "step": 6481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.928281952309031e-05,
+      "loss": 0.4275,
+      "step": 6482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.927997793044567e-05,
+      "loss": 0.2058,
+      "step": 6483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9277136170580923e-05,
+      "loss": 1.0338,
+      "step": 6484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.927429424360709e-05,
+      "loss": 0.8132,
+      "step": 6485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.92714521496352e-05,
+      "loss": 0.6248,
+      "step": 6486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9268609888776305e-05,
+      "loss": 1.087,
+      "step": 6487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.926576746114144e-05,
+      "loss": 0.8058,
+      "step": 6488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9262924866841677e-05,
+      "loss": 0.7608,
+      "step": 6489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.926008210598806e-05,
+      "loss": 0.729,
+      "step": 6490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9257239178691662e-05,
+      "loss": 0.8792,
+      "step": 6491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9254396085063567e-05,
+      "loss": 1.3805,
+      "step": 6492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.925155282521484e-05,
+      "loss": 0.5908,
+      "step": 6493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.924870939925658e-05,
+      "loss": 0.3456,
+      "step": 6494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.924586580729988e-05,
+      "loss": 0.5671,
+      "step": 6495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.924302204945583e-05,
+      "loss": 0.4714,
+      "step": 6496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.924017812583555e-05,
+      "loss": 0.0745,
+      "step": 6497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.923733403655014e-05,
+      "loss": 0.1264,
+      "step": 6498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9234489781710722e-05,
+      "loss": 0.505,
+      "step": 6499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9231645361428427e-05,
+      "loss": 0.535,
+      "step": 6500
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 12715,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 5041457742741504.0,
+  "trial_name": null,
+  "trial_params": null
+}