diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9037 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999375871179812,
+  "global_step": 1502,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.7799,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 8.5819,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 8.3491,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.4053,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 8.6403,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 8.8499,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 8.7314,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 9.183,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.5262,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.9182,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.3289,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.887,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.9669,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 9.0928,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.4936,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.919,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.8329,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.8242,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.9946,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 8.7148,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 9.2945,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999978096767827e-05,
+      "loss": 9.3696,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999991238716726e-05,
+      "loss": 4.8219,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999802871486138e-05,
+      "loss": 3.7635,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999964955020422e-05,
+      "loss": 3.4165,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999452423993156e-05,
+      "loss": 2.8996,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999921149371648e-05,
+      "loss": 2.7341,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998926760429626e-05,
+      "loss": 2.194,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998598225379905e-05,
+      "loss": 2.0475,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998225890006516e-05,
+      "loss": 2.1878,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997809755940525e-05,
+      "loss": 2.0967,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997349825004876e-05,
+      "loss": 1.99,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996846099214352e-05,
+      "loss": 1.8566,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999629858077561e-05,
+      "loss": 1.7685,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995707272087127e-05,
+      "loss": 1.7896,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995072175739226e-05,
+      "loss": 1.7578,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994393294514024e-05,
+      "loss": 2.0856,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993670631385477e-05,
+      "loss": 1.7727,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992904189519303e-05,
+      "loss": 2.0062,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999209397227302e-05,
+      "loss": 1.7943,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991239983195902e-05,
+      "loss": 1.6883,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990342226028972e-05,
+      "loss": 1.8946,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998940070470499e-05,
+      "loss": 1.8493,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988415423348423e-05,
+      "loss": 1.5857,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9987386386275443e-05,
+      "loss": 1.6156,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9986313597993898e-05,
+      "loss": 1.6102,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985197063203293e-05,
+      "loss": 1.4862,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9984036786794775e-05,
+      "loss": 1.6467,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99828327738511e-05,
+      "loss": 1.5734,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981585029646625e-05,
+      "loss": 1.4468,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998029355964728e-05,
+      "loss": 1.469,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978958369510532e-05,
+      "loss": 1.6436,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.997757946508538e-05,
+      "loss": 1.4381,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9976156852412318e-05,
+      "loss": 1.6637,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9974690537723308e-05,
+      "loss": 1.9004,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9973180527441757e-05,
+      "loss": 1.5168,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.997162682818248e-05,
+      "loss": 1.6769,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9970029446751695e-05,
+      "loss": 1.8059,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996838839014696e-05,
+      "loss": 1.8391,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996670366555716e-05,
+      "loss": 1.2964,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9964975280362486e-05,
+      "loss": 1.3924,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9963203242134374e-05,
+      "loss": 1.6445,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9961387558635497e-05,
+      "loss": 1.4033,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959528237819734e-05,
+      "loss": 1.5729,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957625287832097e-05,
+      "loss": 1.5994,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995567871700875e-05,
+      "loss": 1.592,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953688533876924e-05,
+      "loss": 1.4148,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9951654747154906e-05,
+      "loss": 1.6324,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9949577365752005e-05,
+      "loss": 1.4515,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9947456398768486e-05,
+      "loss": 1.2683,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.994529185549556e-05,
+      "loss": 1.4222,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9943083745415325e-05,
+      "loss": 1.3702,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9940832078200733e-05,
+      "loss": 1.6536,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9938536863715533e-05,
+      "loss": 1.5127,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9936198112014254e-05,
+      "loss": 1.4665,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9933815833342147e-05,
+      "loss": 1.4374,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993139003813512e-05,
+      "loss": 1.5101,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9928920737019735e-05,
+      "loss": 1.5986,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992640794081312e-05,
+      "loss": 1.3795,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992385166052295e-05,
+      "loss": 1.0694,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921251907347386e-05,
+      "loss": 1.3281,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991860869267502e-05,
+      "loss": 1.5592,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.991592202808485e-05,
+      "loss": 1.167,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99131919253462e-05,
+      "loss": 1.104,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9910418396418688e-05,
+      "loss": 1.1629,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9907601453452158e-05,
+      "loss": 1.2083,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.990474110878665e-05,
+      "loss": 1.4332,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9901837374952308e-05,
+      "loss": 1.4257,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9898890264669376e-05,
+      "loss": 1.3028,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9895899790848097e-05,
+      "loss": 1.4226,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989286596658868e-05,
+      "loss": 1.4529,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9889788805181237e-05,
+      "loss": 1.3159,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9886668320105724e-05,
+      "loss": 1.2511,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9883504525031886e-05,
+      "loss": 1.2383,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9880297433819185e-05,
+      "loss": 1.6267,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9877047060516755e-05,
+      "loss": 1.3208,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9873753419363336e-05,
+      "loss": 1.0784,
+      "step": 97
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98704165247872e-05,
+      "loss": 1.2762,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.986703639140611e-05,
+      "loss": 1.5412,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 1.3009,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9860146467647072e-05,
+      "loss": 1.2991,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9856636707451446e-05,
+      "loss": 1.5869,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9853083768815372e-05,
+      "loss": 1.633,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9849487667303013e-05,
+      "loss": 1.7519,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.984584841866762e-05,
+      "loss": 1.4797,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9842166038851455e-05,
+      "loss": 1.3602,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9838440543985723e-05,
+      "loss": 1.3682,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9834671950390502e-05,
+      "loss": 1.301,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983086027457466e-05,
+      "loss": 1.5226,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9827005533235804e-05,
+      "loss": 1.4096,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9823107743260196e-05,
+      "loss": 1.3706,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981916692172267e-05,
+      "loss": 1.2882,
+      "step": 112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9815183085886583e-05,
+      "loss": 1.3456,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9811156253203702e-05,
+      "loss": 1.3662,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.980708644131416e-05,
+      "loss": 1.6255,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9802973668046364e-05,
+      "loss": 1.17,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9798817951416914e-05,
+      "loss": 1.081,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9794619309630546e-05,
+      "loss": 1.2337,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.979037776108002e-05,
+      "loss": 1.3838,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9786093324346057e-05,
+      "loss": 1.498,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.978176601819726e-05,
+      "loss": 1.5018,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9777395861590036e-05,
+      "loss": 1.3046,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9772982873668487e-05,
+      "loss": 1.1584,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9768527073764355e-05,
+      "loss": 1.3485,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9764028481396928e-05,
+      "loss": 1.4488,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9759487116272938e-05,
+      "loss": 1.1664,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975490299828651e-05,
+      "loss": 1.2325,
+      "step": 127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9750276147519045e-05,
+      "loss": 1.1022,
+      "step": 128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.974560658423913e-05,
+      "loss": 1.4822,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9740894328902483e-05,
+      "loss": 1.1994,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.973613940215182e-05,
+      "loss": 1.2343,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97313418248168e-05,
+      "loss": 1.7081,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9726501617913904e-05,
+      "loss": 1.4261,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9721618802646375e-05,
+      "loss": 1.2681,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97166934004041e-05,
+      "loss": 1.2279,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9711725432763524e-05,
+      "loss": 1.225,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9706714921487554e-05,
+      "loss": 1.4773,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970166188852547e-05,
+      "loss": 1.451,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969656635601282e-05,
+      "loss": 1.0577,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9691428346271334e-05,
+      "loss": 1.0459,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968624788180882e-05,
+      "loss": 1.3105,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9681024985319054e-05,
+      "loss": 1.3013,
+      "step": 142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9675759679681697e-05,
+      "loss": 1.4053,
+      "step": 143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.96704519879622e-05,
+      "loss": 1.2314,
+      "step": 144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9665101933411682e-05,
+      "loss": 1.4979,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9659709539466833e-05,
+      "loss": 1.2871,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9654274829749828e-05,
+      "loss": 1.4446,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9648797828068214e-05,
+      "loss": 1.3122,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9643278558414794e-05,
+      "loss": 1.6317,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9637717044967536e-05,
+      "loss": 1.342,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9632113312089466e-05,
+      "loss": 1.1897,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9626467384328558e-05,
+      "loss": 1.3822,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.962077928641762e-05,
+      "loss": 1.1728,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9615049043274207e-05,
+      "loss": 1.1171,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960927668000048e-05,
+      "loss": 1.257,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9603462221883122e-05,
+      "loss": 1.181,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9597605694393223e-05,
+      "loss": 1.3059,
+      "step": 157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9591707123186156e-05,
+      "loss": 1.0049,
+      "step": 158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9585766534101475e-05,
+      "loss": 1.0794,
+      "step": 159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9579783953162803e-05,
+      "loss": 1.3177,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9573759406577712e-05,
+      "loss": 1.2834,
+      "step": 161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9567692920737608e-05,
+      "loss": 1.2902,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9561584522217624e-05,
+      "loss": 1.178,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.955543423777649e-05,
+      "loss": 1.3826,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.954924209435643e-05,
+      "loss": 1.2984,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9543008119083035e-05,
+      "loss": 1.267,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9536732339265145e-05,
+      "loss": 1.2156,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9530414782394735e-05,
+      "loss": 1.0258,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952405547614679e-05,
+      "loss": 1.5041,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9517654448379173e-05,
+      "loss": 1.2333,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9511211727132534e-05,
+      "loss": 1.172,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950472734063015e-05,
+      "loss": 1.2654,
+      "step": 172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.949820131727783e-05,
+      "loss": 1.6377,
+      "step": 173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.949163368566377e-05,
+      "loss": 1.0229,
+      "step": 174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9485024474558445e-05,
+      "loss": 1.2945,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9478373712914475e-05,
+      "loss": 1.0461,
+      "step": 176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.947168142986649e-05,
+      "loss": 1.2937,
+      "step": 177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.946494765473102e-05,
+      "loss": 1.1789,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 1.09,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94513557463724e-05,
+      "loss": 1.1847,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94444976726906e-05,
+      "loss": 1.2865,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9437598226003738e-05,
+      "loss": 1.1802,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9430657436535857e-05,
+      "loss": 1.3919,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94236753346921e-05,
+      "loss": 1.0321,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9416651951058584e-05,
+      "loss": 1.0071,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9409587316402272e-05,
+      "loss": 1.1726,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940248146167083e-05,
+      "loss": 0.9987,
+      "step": 187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9395334417992494e-05,
+      "loss": 1.1561,
+      "step": 188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9388146216675938e-05,
+      "loss": 1.0527,
+      "step": 189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.938091688921013e-05,
+      "loss": 1.2099,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9373646467264193e-05,
+      "loss": 1.0293,
+      "step": 191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.936633498268728e-05,
+      "loss": 1.159,
+      "step": 192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.935898246750842e-05,
+      "loss": 1.0963,
+      "step": 193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.935158895393638e-05,
+      "loss": 1.1177,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.934415447435953e-05,
+      "loss": 1.1974,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9336679061345694e-05,
+      "loss": 1.075,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9329162747642017e-05,
+      "loss": 1.0858,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9321605566174808e-05,
+      "loss": 1.2001,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.931400755004941e-05,
+      "loss": 0.9978,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9306368732550045e-05,
+      "loss": 1.2024,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9298689147139674e-05,
+      "loss": 1.297,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9290968827459843e-05,
+      "loss": 1.1893,
+      "step": 202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.928320780733054e-05,
+      "loss": 1.2689,
+      "step": 203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.927540612075005e-05,
+      "loss": 1.1269,
+      "step": 204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.926756380189481e-05,
+      "loss": 1.0516,
+      "step": 205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.925968088511924e-05,
+      "loss": 0.9208,
+      "step": 206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9251757404955617e-05,
+      "loss": 1.0471,
+      "step": 207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9243793396113902e-05,
+      "loss": 1.3811,
+      "step": 208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9235788893481602e-05,
+      "loss": 0.9613,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9227743932123614e-05,
+      "loss": 1.0434,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.921965854728207e-05,
+      "loss": 0.9956,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9211532774376177e-05,
+      "loss": 1.0143,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9203366649002082e-05,
+      "loss": 0.9994,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9195160206932683e-05,
+      "loss": 1.1395,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918691348411751e-05,
+      "loss": 1.0882,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9178626516682536e-05,
+      "loss": 1.3267,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9170299340930035e-05,
+      "loss": 1.0497,
+      "step": 217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9161931993338422e-05,
+      "loss": 1.1867,
+      "step": 218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9153524510562087e-05,
+      "loss": 1.2729,
+      "step": 219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.914507692943124e-05,
+      "loss": 0.9416,
+      "step": 220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9136589286951742e-05,
+      "loss": 1.1378,
+      "step": 221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9128061620304962e-05,
+      "loss": 1.3341,
+      "step": 222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.911949396684759e-05,
+      "loss": 1.2821,
+      "step": 223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9110886364111483e-05,
+      "loss": 1.0449,
+      "step": 224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9102238849803506e-05,
+      "loss": 1.1662,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9093551461805367e-05,
+      "loss": 1.0909,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9084824238173433e-05,
+      "loss": 1.0458,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9076057217138594e-05,
+      "loss": 0.9827,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9067250437106063e-05,
+      "loss": 1.0685,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9058403936655235e-05,
+      "loss": 1.2948,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904951775453949e-05,
+      "loss": 0.9489,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9040591929686064e-05,
+      "loss": 1.1723,
+      "step": 232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9031626501195833e-05,
+      "loss": 0.9746,
+      "step": 233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9022621508343168e-05,
+      "loss": 1.1142,
+      "step": 234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.901357699057576e-05,
+      "loss": 1.2316,
+      "step": 235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.900449298751444e-05,
+      "loss": 0.9651,
+      "step": 236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8995369538953024e-05,
+      "loss": 1.125,
+      "step": 237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8986206684858106e-05,
+      "loss": 1.3784,
+      "step": 238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.897700446536891e-05,
+      "loss": 1.1435,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8967762920797107e-05,
+      "loss": 1.385,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8958482091626643e-05,
+      "loss": 1.0251,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.894916201851354e-05,
+      "loss": 1.1461,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8939802742285746e-05,
+      "loss": 1.2473,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.893040430394294e-05,
+      "loss": 1.3738,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8920966744656362e-05,
+      "loss": 1.0929,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8911490105768617e-05,
+      "loss": 1.1167,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890197442879351e-05,
+      "loss": 1.0855,
+      "step": 247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8892419755415866e-05,
+      "loss": 1.169,
+      "step": 248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.888282612749132e-05,
+      "loss": 0.8536,
+      "step": 249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.887319358704617e-05,
+      "loss": 0.9992,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8863522176277165e-05,
+      "loss": 1.0648,
+      "step": 251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.885381193755134e-05,
+      "loss": 1.0357,
+      "step": 252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.884406291340582e-05,
+      "loss": 0.8187,
+      "step": 253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.883427514654763e-05,
+      "loss": 1.08,
+      "step": 254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.882444867985351e-05,
+      "loss": 1.0366,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8814583556369744e-05,
+      "loss": 1.2414,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8804679819311946e-05,
+      "loss": 1.2844,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 1.1268,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.878475667818231e-05,
+      "loss": 1.001,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8774737361386705e-05,
+      "loss": 1.1177,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8764679605569163e-05,
+      "loss": 1.0555,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8754583454789155e-05,
+      "loss": 1.359,
+      "step": 262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.874444895327435e-05,
+      "loss": 1.0431,
+      "step": 263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8734276145420408e-05,
+      "loss": 1.073,
+      "step": 264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8724065075790814e-05,
+      "loss": 0.8711,
+      "step": 265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.871381578911665e-05,
+      "loss": 1.1563,
+      "step": 266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8703528330296413e-05,
+      "loss": 1.1142,
+      "step": 267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.869320274439583e-05,
+      "loss": 1.1908,
+      "step": 268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8682839076647635e-05,
+      "loss": 0.8458,
+      "step": 269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8672437372451394e-05,
+      "loss": 1.1747,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.86619976773733e-05,
+      "loss": 1.1675,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.865152003714596e-05,
+      "loss": 0.9966,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8641004497668215e-05,
+      "loss": 0.9161,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8630451105004926e-05,
+      "loss": 1.0594,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8619859905386774e-05,
+      "loss": 0.9494,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8609230945210056e-05,
+      "loss": 1.1528,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8598564271036493e-05,
+      "loss": 1.4684,
+      "step": 277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.858785992959301e-05,
+      "loss": 0.7879,
+      "step": 278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8577117967771553e-05,
+      "loss": 1.1083,
+      "step": 279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.856633843262884e-05,
+      "loss": 0.7909,
+      "step": 280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8555521371386217e-05,
+      "loss": 0.8793,
+      "step": 281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8544666831429404e-05,
+      "loss": 1.2699,
+      "step": 282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8533774860308296e-05,
+      "loss": 0.8275,
+      "step": 283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8522845505736774e-05,
+      "loss": 0.8289,
+      "step": 284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.851187881559247e-05,
+      "loss": 1.0701,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.850087483791658e-05,
+      "loss": 0.9635,
+      "step": 286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8489833620913644e-05,
+      "loss": 1.2674,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8478755212951317e-05,
+      "loss": 1.2,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84676396625602e-05,
+      "loss": 0.8902,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8456487018433585e-05,
+      "loss": 1.0719,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8445297329427257e-05,
+      "loss": 1.236,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8434070644559298e-05,
+      "loss": 0.9906,
+      "step": 292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8422807013009835e-05,
+      "loss": 1.2247,
+      "step": 293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8411506484120862e-05,
+      "loss": 1.1634,
+      "step": 294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8400169107396e-05,
+      "loss": 1.1878,
+      "step": 295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8388794932500285e-05,
+      "loss": 1.3076,
+      "step": 296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8377384009259958e-05,
+      "loss": 0.9788,
+      "step": 297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8365936387662238e-05,
+      "loss": 1.056,
+      "step": 298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8354452117855108e-05,
+      "loss": 1.0367,
+      "step": 299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.834293125014709e-05,
+      "loss": 0.8806,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.833137383500704e-05,
+      "loss": 1.0436,
+      "step": 301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.83197799230639e-05,
+      "loss": 1.0296,
+      "step": 302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8308149565106507e-05,
+      "loss": 0.9125,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8296482812083335e-05,
+      "loss": 1.0602,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8284779715102315e-05,
+      "loss": 0.7557,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8273040325430575e-05,
+      "loss": 1.0303,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8261264694494225e-05,
+      "loss": 1.0454,
+      "step": 307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8249452873878146e-05,
+      "loss": 0.9198,
+      "step": 308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.823760491532575e-05,
+      "loss": 1.0419,
+      "step": 309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8225720870738745e-05,
+      "loss": 1.0386,
+      "step": 310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.821380079217694e-05,
+      "loss": 1.058,
+      "step": 311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8201844731857977e-05,
+      "loss": 1.2175,
+      "step": 312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8189852742157125e-05,
+      "loss": 1.1328,
+      "step": 313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8177824875607065e-05,
+      "loss": 1.0908,
+      "step": 314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8165761184897617e-05,
+      "loss": 1.1939,
+      "step": 315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8153661722875548e-05,
+      "loss": 1.0388,
+      "step": 316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.814152654254432e-05,
+      "loss": 0.9037,
+      "step": 317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.812935569706387e-05,
+      "loss": 0.9088,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.811714923975037e-05,
+      "loss": 1.0434,
+      "step": 319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8104907224075992e-05,
+      "loss": 1.0964,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8092629703668677e-05,
+      "loss": 0.9333,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8080316732311907e-05,
+      "loss": 1.2413,
+      "step": 322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.806796836394445e-05,
+      "loss": 1.0349,
+      "step": 323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8055584652660143e-05,
+      "loss": 0.8087,
+      "step": 324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.804316565270765e-05,
+      "loss": 1.1631,
+      "step": 325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8030711418490214e-05,
+      "loss": 1.212,
+      "step": 326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8018222004565436e-05,
+      "loss": 0.8637,
+      "step": 327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8005697465645017e-05,
+      "loss": 1.1309,
+      "step": 328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.799313785659454e-05,
+      "loss": 1.0502,
+      "step": 329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.798054323243321e-05,
+      "loss": 1.0667,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.796791364833362e-05,
+      "loss": 1.1415,
+      "step": 331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7955249159621514e-05,
+      "loss": 1.0799,
+      "step": 332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7942549821775537e-05,
+      "loss": 0.8853,
+      "step": 333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7929815690427004e-05,
+      "loss": 0.9538,
+      "step": 334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7917046821359637e-05,
+      "loss": 1.023,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7904243270509338e-05,
+      "loss": 0.9256,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 1.0729,
+      "step": 337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878532347962947e-05,
+      "loss": 1.2781,
+      "step": 338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7865625088897313e-05,
+      "loss": 0.9976,
+      "step": 339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7852683373309178e-05,
+      "loss": 0.9731,
+      "step": 340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7839707257891622e-05,
+      "loss": 1.0933,
+      "step": 341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7826696799488418e-05,
+      "loss": 1.1952,
+      "step": 342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7813652055093778e-05,
+      "loss": 0.9964,
+      "step": 343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7800573081852124e-05,
+      "loss": 1.0479,
+      "step": 344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7787459937057808e-05,
+      "loss": 1.4034,
+      "step": 345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7774312678154886e-05,
+      "loss": 1.0298,
+      "step": 346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7761131362736845e-05,
+      "loss": 1.1508,
+      "step": 347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7747916048546372e-05,
+      "loss": 1.1461,
+      "step": 348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7734666793475083e-05,
+      "loss": 1.0761,
+      "step": 349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.772138365556328e-05,
+      "loss": 1.1646,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.77080666929997e-05,
+      "loss": 1.2665,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7694715964121235e-05,
+      "loss": 1.066,
+      "step": 352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768133152741272e-05,
+      "loss": 0.9316,
+      "step": 353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7667913441506633e-05,
+      "loss": 0.9874,
+      "step": 354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7654461765182868e-05,
+      "loss": 0.9546,
+      "step": 355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7640976557368462e-05,
+      "loss": 1.1788,
+      "step": 356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.762745787713734e-05,
+      "loss": 1.0041,
+      "step": 357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7613905783710063e-05,
+      "loss": 0.9288,
+      "step": 358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.760032033645356e-05,
+      "loss": 1.084,
+      "step": 359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7586701594880873e-05,
+      "loss": 1.1925,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7573049618650893e-05,
+      "loss": 0.9755,
+      "step": 361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.75593644675681e-05,
+      "loss": 0.8911,
+      "step": 362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7545646201582304e-05,
+      "loss": 1.0825,
+      "step": 363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7531894880788373e-05,
+      "loss": 0.6689,
+      "step": 364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7518110565425993e-05,
+      "loss": 0.7203,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7504293315879364e-05,
+      "loss": 1.1538,
+      "step": 366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7490443192676972e-05,
+      "loss": 0.9287,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.747656025649132e-05,
+      "loss": 1.1147,
+      "step": 368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746264456813863e-05,
+      "loss": 0.727,
+      "step": 369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7448696188578625e-05,
+      "loss": 1.064,
+      "step": 370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7434715178914214e-05,
+      "loss": 0.8907,
+      "step": 371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.742070160039126e-05,
+      "loss": 1.0857,
+      "step": 372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7406655514398302e-05,
+      "loss": 0.9442,
+      "step": 373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7392576982466266e-05,
+      "loss": 0.8933,
+      "step": 374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.737846606626823e-05,
+      "loss": 0.9584,
+      "step": 375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7364322827619128e-05,
+      "loss": 0.97,
+      "step": 376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7350147328475488e-05,
+      "loss": 1.041,
+      "step": 377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7335939630935155e-05,
+      "loss": 0.9326,
+      "step": 378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7321699797237033e-05,
+      "loss": 1.0566,
+      "step": 379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.73074278897608e-05,
+      "loss": 0.9779,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.729312397102663e-05,
+      "loss": 1.1148,
+      "step": 381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7278788103694944e-05,
+      "loss": 0.8958,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7264420350566098e-05,
+      "loss": 1.1199,
+      "step": 383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725002077458014e-05,
+      "loss": 0.9649,
+      "step": 384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7235589438816533e-05,
+      "loss": 0.9317,
+      "step": 385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.722112640649384e-05,
+      "loss": 0.8341,
+      "step": 386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7206631740969505e-05,
+      "loss": 1.1026,
+      "step": 387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.719210550573952e-05,
+      "loss": 1.0872,
+      "step": 388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7177547764438197e-05,
+      "loss": 0.6306,
+      "step": 389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.716295858083785e-05,
+      "loss": 1.0661,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7148338018848533e-05,
+      "loss": 1.0019,
+      "step": 391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7133686142517762e-05,
+      "loss": 0.8352,
+      "step": 392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.711900301603022e-05,
+      "loss": 1.3995,
+      "step": 393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.71042887037075e-05,
+      "loss": 0.8172,
+      "step": 394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7089543270007795e-05,
+      "loss": 1.0348,
+      "step": 395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7074766779525643e-05,
+      "loss": 0.9255,
+      "step": 396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7059959296991622e-05,
+      "loss": 1.0092,
+      "step": 397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7045120887272076e-05,
+      "loss": 0.9224,
+      "step": 398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703025161536883e-05,
+      "loss": 0.9692,
+      "step": 399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015351546418914e-05,
+      "loss": 0.7466,
+      "step": 400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7000420745694256e-05,
+      "loss": 0.9728,
+      "step": 401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6985459278601415e-05,
+      "loss": 0.9605,
+      "step": 402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.697046721068129e-05,
+      "loss": 0.8744,
+      "step": 403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6955444607608833e-05,
+      "loss": 1.0407,
+      "step": 404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6940391535192753e-05,
+      "loss": 0.8633,
+      "step": 405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6925308059375235e-05,
+      "loss": 0.8993,
+      "step": 406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.691019424623166e-05,
+      "loss": 1.089,
+      "step": 407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6895050161970298e-05,
+      "loss": 1.0275,
+      "step": 408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6879875872932026e-05,
+      "loss": 0.9038,
+      "step": 409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.686467144559004e-05,
+      "loss": 0.7568,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6849436946549554e-05,
+      "loss": 0.7134,
+      "step": 411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6834172442547534e-05,
+      "loss": 0.9546,
+      "step": 412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.681887800045237e-05,
+      "loss": 1.0251,
+      "step": 413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68035536872636e-05,
+      "loss": 0.9254,
+      "step": 414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678819957011163e-05,
+      "loss": 0.7845,
+      "step": 415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 1.0221,
+      "step": 416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6757402193092174e-05,
+      "loss": 0.8602,
+      "step": 417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.674195906813711e-05,
+      "loss": 0.813,
+      "step": 418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6726486409043094e-05,
+      "loss": 1.0069,
+      "step": 419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.671098428359037e-05,
+      "loss": 1.0772,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.669545275968827e-05,
+      "loss": 0.8996,
+      "step": 421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6679891905374908e-05,
+      "loss": 0.9241,
+      "step": 422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.666430178881689e-05,
+      "loss": 1.1211,
+      "step": 423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6648682478308998e-05,
+      "loss": 1.0953,
+      "step": 424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.663303404227391e-05,
+      "loss": 0.882,
+      "step": 425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6617356549261897e-05,
+      "loss": 0.8091,
+      "step": 426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6601650067950505e-05,
+      "loss": 0.7909,
+      "step": 427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.658591466714428e-05,
+      "loss": 1.0418,
+      "step": 428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657015041577445e-05,
+      "loss": 0.9729,
+      "step": 429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6554357382898626e-05,
+      "loss": 0.9652,
+      "step": 430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6538535637700498e-05,
+      "loss": 0.7864,
+      "step": 431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6522685249489542e-05,
+      "loss": 1.0191,
+      "step": 432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6506806287700703e-05,
+      "loss": 1.1787,
+      "step": 433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6490898821894096e-05,
+      "loss": 0.8709,
+      "step": 434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6474962921754708e-05,
+      "loss": 0.8921,
+      "step": 435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6458998657092084e-05,
+      "loss": 0.6149,
+      "step": 436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6443006097840018e-05,
+      "loss": 1.04,
+      "step": 437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6426985314056262e-05,
+      "loss": 1.2029,
+      "step": 438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.64109363759222e-05,
+      "loss": 0.9765,
+      "step": 439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6394859353742564e-05,
+      "loss": 1.0275,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6378754317945096e-05,
+      "loss": 0.7519,
+      "step": 441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6362621339080268e-05,
+      "loss": 0.7146,
+      "step": 442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.634646048782096e-05,
+      "loss": 0.9447,
+      "step": 443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330271834962137e-05,
+      "loss": 0.9144,
+      "step": 444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631405545142057e-05,
+      "loss": 1.0455,
+      "step": 445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297811408234504e-05,
+      "loss": 1.051,
+      "step": 446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.628153977656334e-05,
+      "loss": 0.8615,
+      "step": 447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.626524062768735e-05,
+      "loss": 1.0195,
+      "step": 448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6248914033007347e-05,
+      "loss": 0.93,
+      "step": 449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.623256006404436e-05,
+      "loss": 1.0096,
+      "step": 450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.621617879243935e-05,
+      "loss": 0.8477,
+      "step": 451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6199770289952878e-05,
+      "loss": 0.9636,
+      "step": 452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6183334628464783e-05,
+      "loss": 0.7315,
+      "step": 453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6166871879973896e-05,
+      "loss": 1.0451,
+      "step": 454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6150382116597695e-05,
+      "loss": 0.9682,
+      "step": 455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6133865410572e-05,
+      "loss": 0.9976,
+      "step": 456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6117321834250664e-05,
+      "loss": 0.9983,
+      "step": 457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6100751460105244e-05,
+      "loss": 0.9114,
+      "step": 458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084154360724693e-05,
+      "loss": 1.104,
+      "step": 459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067530608815035e-05,
+      "loss": 0.8748,
+      "step": 460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605088027719904e-05,
+      "loss": 0.9229,
+      "step": 461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603420343881594e-05,
+      "loss": 0.7216,
+      "step": 462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6017500166721055e-05,
+      "loss": 1.0407,
+      "step": 463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6000770534085518e-05,
+      "loss": 0.9899,
+      "step": 464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5984014614195936e-05,
+      "loss": 0.7019,
+      "step": 465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5967232480454075e-05,
+      "loss": 0.7566,
+      "step": 466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5950424206376513e-05,
+      "loss": 0.9115,
+      "step": 467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5933589865594373e-05,
+      "loss": 1.031,
+      "step": 468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.591672953185294e-05,
+      "loss": 0.7282,
+      "step": 469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5899843279011376e-05,
+      "loss": 0.9029,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5882931181042388e-05,
+      "loss": 0.8569,
+      "step": 471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5865993312031896e-05,
+      "loss": 0.956,
+      "step": 472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5849029746178716e-05,
+      "loss": 0.7024,
+      "step": 473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583204055779423e-05,
+      "loss": 0.8425,
+      "step": 474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581502582130207e-05,
+      "loss": 0.8771,
+      "step": 475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5797985611237775e-05,
+      "loss": 0.7647,
+      "step": 476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5780920002248484e-05,
+      "loss": 1.0538,
+      "step": 477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57638290690926e-05,
+      "loss": 1.2245,
+      "step": 478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5746712886639448e-05,
+      "loss": 0.9618,
+      "step": 479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.572957152986898e-05,
+      "loss": 0.8931,
+      "step": 480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5712405073871416e-05,
+      "loss": 1.2077,
+      "step": 481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5695213593846933e-05,
+      "loss": 1.0553,
+      "step": 482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5677997165105322e-05,
+      "loss": 1.0305,
+      "step": 483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5660755863065676e-05,
+      "loss": 0.7733,
+      "step": 484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.564348976325604e-05,
+      "loss": 0.774,
+      "step": 485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5626198941313092e-05,
+      "loss": 0.8493,
+      "step": 486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.560888347298181e-05,
+      "loss": 0.7531,
+      "step": 487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.559154343411514e-05,
+      "loss": 1.1483,
+      "step": 488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557417890067366e-05,
+      "loss": 0.8102,
+      "step": 489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556789948725253e-05,
+      "loss": 0.9872,
+      "step": 490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553937665444477e-05,
+      "loss": 0.8057,
+      "step": 491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5521939094113693e-05,
+      "loss": 0.784,
+      "step": 492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5504477344119807e-05,
+      "loss": 0.8196,
+      "step": 493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5486991480956876e-05,
+      "loss": 0.8362,
+      "step": 494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 0.833,
+      "step": 495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5451947721626676e-05,
+      "loss": 0.9491,
+      "step": 496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5434389978973734e-05,
+      "loss": 0.7062,
+      "step": 497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.541680843017971e-05,
+      "loss": 1.0637,
+      "step": 498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.539920315226314e-05,
+      "loss": 0.7493,
+      "step": 499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5381574222346536e-05,
+      "loss": 0.79,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5363921717656004e-05,
+      "loss": 1.0652,
+      "step": 501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5346245715520916e-05,
+      "loss": 0.7523,
+      "step": 502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.53285462933736e-05,
+      "loss": 1.0092,
+      "step": 503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5310823528748962e-05,
+      "loss": 0.9266,
+      "step": 504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5293077499284162e-05,
+      "loss": 0.8713,
+      "step": 505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527530828271829e-05,
+      "loss": 1.1278,
+      "step": 506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257515956891996e-05,
+      "loss": 0.6857,
+      "step": 507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239700599747171e-05,
+      "loss": 0.6272,
+      "step": 508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5221862289326592e-05,
+      "loss": 0.8791,
+      "step": 509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5204001103773594e-05,
+      "loss": 0.9722,
+      "step": 510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5186117121331717e-05,
+      "loss": 0.978,
+      "step": 511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5168210420344358e-05,
+      "loss": 1.0055,
+      "step": 512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5150281079254448e-05,
+      "loss": 1.0012,
+      "step": 513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5132329176604091e-05,
+      "loss": 1.1395,
+      "step": 514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5114354791034225e-05,
+      "loss": 0.6441,
+      "step": 515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5096358001284275e-05,
+      "loss": 1.0738,
+      "step": 516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5078338886191819e-05,
+      "loss": 1.1373,
+      "step": 517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5060297524692227e-05,
+      "loss": 0.7126,
+      "step": 518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042233995818323e-05,
+      "loss": 0.9349,
+      "step": 519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502414837870004e-05,
+      "loss": 0.7913,
+      "step": 520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5006040752564079e-05,
+      "loss": 1.1132,
+      "step": 521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987911196733537e-05,
+      "loss": 0.8354,
+      "step": 522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49697597906276e-05,
+      "loss": 0.796,
+      "step": 523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4951586613761154e-05,
+      "loss": 0.8797,
+      "step": 524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4933391745744462e-05,
+      "loss": 1.2104,
+      "step": 525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4915175266282804e-05,
+      "loss": 0.7973,
+      "step": 526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4896937255176142e-05,
+      "loss": 0.7018,
+      "step": 527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4878677792318751e-05,
+      "loss": 1.0073,
+      "step": 528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4860396957698884e-05,
+      "loss": 0.8833,
+      "step": 529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4842094831398413e-05,
+      "loss": 0.7769,
+      "step": 530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4823771493592483e-05,
+      "loss": 0.7877,
+      "step": 531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4805427024549156e-05,
+      "loss": 0.6533,
+      "step": 532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4787061504629066e-05,
+      "loss": 0.9494,
+      "step": 533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768675014285063e-05,
+      "loss": 0.9719,
+      "step": 534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750267634061859e-05,
+      "loss": 0.8172,
+      "step": 535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4731839444595681e-05,
+      "loss": 1.0048,
+      "step": 536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713390526613908e-05,
+      "loss": 0.9258,
+      "step": 537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694920960934735e-05,
+      "loss": 0.8252,
+      "step": 538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4676430828466787e-05,
+      "loss": 0.8822,
+      "step": 539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4657920210208807e-05,
+      "loss": 0.9563,
+      "step": 540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4639389187249263e-05,
+      "loss": 0.8332,
+      "step": 541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4620837840766015e-05,
+      "loss": 0.868,
+      "step": 542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4602266252025954e-05,
+      "loss": 0.7153,
+      "step": 543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4583674502384648e-05,
+      "loss": 0.9993,
+      "step": 544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4565062673285973e-05,
+      "loss": 0.7638,
+      "step": 545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4546430846261772e-05,
+      "loss": 0.9265,
+      "step": 546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4527779102931496e-05,
+      "loss": 0.8617,
+      "step": 547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4509107525001836e-05,
+      "loss": 0.8822,
+      "step": 548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490416194266371e-05,
+      "loss": 0.8772,
+      "step": 549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471705192605212e-05,
+      "loss": 0.7759,
+      "step": 550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452974601984648e-05,
+      "loss": 1.0502,
+      "step": 551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434224504456766e-05,
+      "loss": 0.9791,
+      "step": 552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415454982159121e-05,
+      "loss": 0.855,
+      "step": 553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396666117314349e-05,
+      "loss": 0.9989,
+      "step": 554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4377857992229825e-05,
+      "loss": 0.8204,
+      "step": 555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4359030689297296e-05,
+      "loss": 0.8762,
+      "step": 556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4340184290992518e-05,
+      "loss": 1.0226,
+      "step": 557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4321318879874899e-05,
+      "loss": 0.9211,
+      "step": 558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4302434538587138e-05,
+      "loss": 0.788,
+      "step": 559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4283531349854853e-05,
+      "loss": 0.8509,
+      "step": 560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4264609396486233e-05,
+      "loss": 0.7472,
+      "step": 561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4245668761371666e-05,
+      "loss": 0.9558,
+      "step": 562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4226709527483373e-05,
+      "loss": 1.1584,
+      "step": 563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207731777875058e-05,
+      "loss": 0.7647,
+      "step": 564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188735595681534e-05,
+      "loss": 1.1635,
+      "step": 565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4169721064118353e-05,
+      "loss": 0.7554,
+      "step": 566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150688266481463e-05,
+      "loss": 0.876,
+      "step": 567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131637286146815e-05,
+      "loss": 0.9712,
+      "step": 568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4112568206570018e-05,
+      "loss": 1.0984,
+      "step": 569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4093481111285973e-05,
+      "loss": 0.8297,
+      "step": 570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.407437608390849e-05,
+      "loss": 0.9969,
+      "step": 571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.405525320812994e-05,
+      "loss": 0.8463,
+      "step": 572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4036112567720882e-05,
+      "loss": 0.8421,
+      "step": 573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 1.0135,
+      "step": 574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.399777832848221e-05,
+      "loss": 0.9378,
+      "step": 575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3978584897581344e-05,
+      "loss": 1.1438,
+      "step": 576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.395937403790673e-05,
+      "loss": 0.7389,
+      "step": 577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3940145833614352e-05,
+      "loss": 0.9196,
+      "step": 578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3920900368936175e-05,
+      "loss": 1.0705,
+      "step": 579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901637728179778e-05,
+      "loss": 0.9146,
+      "step": 580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882357995727975e-05,
+      "loss": 1.0367,
+      "step": 581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386306125603846e-05,
+      "loss": 0.7648,
+      "step": 582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843747593643429e-05,
+      "loss": 0.8911,
+      "step": 583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824417093149205e-05,
+      "loss": 0.8651,
+      "step": 584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805069839235875e-05,
+      "loss": 0.9994,
+      "step": 585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3785705916656919e-05,
+      "loss": 0.7229,
+      "step": 586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3766325410238838e-05,
+      "loss": 0.8816,
+      "step": 587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3746928404880777e-05,
+      "loss": 0.5611,
+      "step": 588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3727514985554158e-05,
+      "loss": 0.9713,
+      "step": 589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3708085237302307e-05,
+      "loss": 0.8487,
+      "step": 590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3688639245240078e-05,
+      "loss": 0.8965,
+      "step": 591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3669177094553497e-05,
+      "loss": 0.8901,
+      "step": 592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3649698870499353e-05,
+      "loss": 0.7168,
+      "step": 593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3630204658404864e-05,
+      "loss": 0.8314,
+      "step": 594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610694543667283e-05,
+      "loss": 0.8771,
+      "step": 595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359116861175352e-05,
+      "loss": 0.9392,
+      "step": 596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571626948199783e-05,
+      "loss": 0.8212,
+      "step": 597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552069638611189e-05,
+      "loss": 0.8027,
+      "step": 598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532496768661397e-05,
+      "loss": 0.8769,
+      "step": 599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512908424092228e-05,
+      "loss": 0.7723,
+      "step": 600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493304690713296e-05,
+      "loss": 0.8419,
+      "step": 601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3473685654401624e-05,
+      "loss": 0.7778,
+      "step": 602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3454051401101278e-05,
+      "loss": 0.8184,
+      "step": 603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3434402016822974e-05,
+      "loss": 0.8222,
+      "step": 604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.341473758764372e-05,
+      "loss": 0.888,
+      "step": 605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3395058199706428e-05,
+      "loss": 0.8362,
+      "step": 606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3375363939219535e-05,
+      "loss": 0.5899,
+      "step": 607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3355654892456635e-05,
+      "loss": 0.7787,
+      "step": 608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335931145756098e-05,
+      "loss": 0.7628,
+      "step": 609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331619278552068e-05,
+      "loss": 0.6979,
+      "step": 610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296439898217158e-05,
+      "loss": 0.8438,
+      "step": 611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276672570375948e-05,
+      "loss": 0.9426,
+      "step": 612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256890888590728e-05,
+      "loss": 0.9836,
+      "step": 613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237094939518043e-05,
+      "loss": 0.7446,
+      "step": 614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217284809876955e-05,
+      "loss": 0.8612,
+      "step": 615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197460586448635e-05,
+      "loss": 0.9367,
+      "step": 616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3177622356076e-05,
+      "loss": 0.9592,
+      "step": 617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3157770205663314e-05,
+      "loss": 0.9729,
+      "step": 618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3137904222175837e-05,
+      "loss": 0.8071,
+      "step": 619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3118024492639418e-05,
+      "loss": 0.8046,
+      "step": 620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.309813110414012e-05,
+      "loss": 0.8051,
+      "step": 621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3078224143823848e-05,
+      "loss": 0.919,
+      "step": 622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3058303698895953e-05,
+      "loss": 0.8787,
+      "step": 623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038369856620863e-05,
+      "loss": 0.8643,
+      "step": 624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018422704321695e-05,
+      "loss": 0.8022,
+      "step": 625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998462329379868e-05,
+      "loss": 0.7079,
+      "step": 626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2978488819234727e-05,
+      "loss": 1.0639,
+      "step": 627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958502261383161e-05,
+      "loss": 0.8724,
+      "step": 628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938502743379212e-05,
+      "loss": 0.8523,
+      "step": 629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2918490352833693e-05,
+      "loss": 0.74,
+      "step": 630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898465177413817e-05,
+      "loss": 0.8312,
+      "step": 631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2878427304842796e-05,
+      "loss": 0.8931,
+      "step": 632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2858376822899466e-05,
+      "loss": 0.7384,
+      "step": 633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2838313819417894e-05,
+      "loss": 0.8051,
+      "step": 634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2818238382287009e-05,
+      "loss": 0.6534,
+      "step": 635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2798150599450203e-05,
+      "loss": 0.7015,
+      "step": 636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2778050558904948e-05,
+      "loss": 0.7233,
+      "step": 637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2757938348702418e-05,
+      "loss": 0.6664,
+      "step": 638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2737814056947096e-05,
+      "loss": 0.9355,
+      "step": 639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717677771796385e-05,
+      "loss": 0.9663,
+      "step": 640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697529581460228e-05,
+      "loss": 0.7847,
+      "step": 641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2677369574200733e-05,
+      "loss": 0.8903,
+      "step": 642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657197838331755e-05,
+      "loss": 0.7245,
+      "step": 643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263701446221854e-05,
+      "loss": 1.0472,
+      "step": 644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616819534277322e-05,
+      "loss": 0.996,
+      "step": 645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596613142974944e-05,
+      "loss": 1.0179,
+      "step": 646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576395376828456e-05,
+      "loss": 0.8861,
+      "step": 647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2556166324404747e-05,
+      "loss": 0.8827,
+      "step": 648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2535926074320144e-05,
+      "loss": 0.6271,
+      "step": 649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2515674715240023e-05,
+      "loss": 0.8094,
+      "step": 650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.249541233587843e-05,
+      "loss": 0.8208,
+      "step": 651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2475139024997684e-05,
+      "loss": 0.9378,
+      "step": 652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.9019,
+      "step": 653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434559963967061e-05,
+      "loss": 0.8439,
+      "step": 654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414254391579706e-05,
+      "loss": 0.9662,
+      "step": 655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393938243197454e-05,
+      "loss": 1.03,
+      "step": 656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373611607818173e-05,
+      "loss": 0.8387,
+      "step": 657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353274574485666e-05,
+      "loss": 0.6743,
+      "step": 658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233292723228928e-05,
+      "loss": 0.6422,
+      "step": 659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312569670363535e-05,
+      "loss": 0.7742,
+      "step": 660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292201977887708e-05,
+      "loss": 0.9026,
+      "step": 661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227182424408546e-05,
+      "loss": 0.8941,
+      "step": 662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251436558224436e-05,
+      "loss": 0.9127,
+      "step": 663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.223103900961588e-05,
+      "loss": 0.7704,
+      "step": 664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2210631687614241e-05,
+      "loss": 0.8596,
+      "step": 665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2190214681616778e-05,
+      "loss": 0.7084,
+      "step": 666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2169788081063181e-05,
+      "loss": 0.7498,
+      "step": 667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2149351975435165e-05,
+      "loss": 0.9467,
+      "step": 668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128906454256078e-05,
+      "loss": 0.8536,
+      "step": 669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2108451607090521e-05,
+      "loss": 1.0086,
+      "step": 670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087987523543951e-05,
+      "loss": 0.897,
+      "step": 671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206751429326228e-05,
+      "loss": 0.748,
+      "step": 672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204703200593149e-05,
+      "loss": 0.782,
+      "step": 673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202654075127724e-05,
+      "loss": 0.7545,
+      "step": 674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006040619064476e-05,
+      "loss": 0.582,
+      "step": 675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985531699097029e-05,
+      "loss": 1.2108,
+      "step": 676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965014081217217e-05,
+      "loss": 0.9376,
+      "step": 677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944487855305478e-05,
+      "loss": 0.7687,
+      "step": 678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.192395311127995e-05,
+      "loss": 0.924,
+      "step": 679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1903409939096082e-05,
+      "loss": 0.6892,
+      "step": 680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1882858428746251e-05,
+      "loss": 0.9455,
+      "step": 681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1862298670259358e-05,
+      "loss": 0.9482,
+      "step": 682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1841730753700434e-05,
+      "loss": 1.078,
+      "step": 683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182115476917025e-05,
+      "loss": 0.6533,
+      "step": 684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800570806804921e-05,
+      "loss": 0.9524,
+      "step": 685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1779978956775507e-05,
+      "loss": 1.16,
+      "step": 686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759379309287617e-05,
+      "loss": 0.7815,
+      "step": 687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738771954581031e-05,
+      "loss": 0.8095,
+      "step": 688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1718156982929282e-05,
+      "loss": 0.6666,
+      "step": 689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169753448463927e-05,
+      "loss": 0.5494,
+      "step": 690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1676904550050874e-05,
+      "loss": 0.7319,
+      "step": 691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1656267269536536e-05,
+      "loss": 0.6384,
+      "step": 692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1635622733500889e-05,
+      "loss": 0.7816,
+      "step": 693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614971032380347e-05,
+      "loss": 0.6948,
+      "step": 694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1594312256642709e-05,
+      "loss": 0.7691,
+      "step": 695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1573646496786766e-05,
+      "loss": 0.8289,
+      "step": 696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.155297384334191e-05,
+      "loss": 0.8023,
+      "step": 697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1532294386867725e-05,
+      "loss": 0.6484,
+      "step": 698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1511608217953595e-05,
+      "loss": 0.9339,
+      "step": 699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490915427218314e-05,
+      "loss": 0.7287,
+      "step": 700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470216105309683e-05,
+      "loss": 1.1426,
+      "step": 701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449510342904111e-05,
+      "loss": 0.8937,
+      "step": 702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428798230706222e-05,
+      "loss": 0.9449,
+      "step": 703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408079859448462e-05,
+      "loss": 0.7313,
+      "step": 704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1387355319890685e-05,
+      "loss": 0.6874,
+      "step": 705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1366624702819768e-05,
+      "loss": 0.946,
+      "step": 706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1345888099049222e-05,
+      "loss": 0.7786,
+      "step": 707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325145599418775e-05,
+      "loss": 0.8062,
+      "step": 708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1304397294793978e-05,
+      "loss": 0.9986,
+      "step": 709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1283643276065826e-05,
+      "loss": 1.0693,
+      "step": 710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1262883634150332e-05,
+      "loss": 0.8225,
+      "step": 711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.124211845998815e-05,
+      "loss": 1.0309,
+      "step": 712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1221347844544158e-05,
+      "loss": 0.8862,
+      "step": 713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200571878807094e-05,
+      "loss": 0.6696,
+      "step": 714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179790653789106e-05,
+      "loss": 0.9876,
+      "step": 715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1159004260525395e-05,
+      "loss": 0.9398,
+      "step": 716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11382127900738e-05,
+      "loss": 0.7528,
+      "step": 717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111741633351441e-05,
+      "loss": 0.7278,
+      "step": 718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109661498194914e-05,
+      "loss": 1.0712,
+      "step": 719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107580882650136e-05,
+      "loss": 1.0084,
+      "step": 720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1054997958315479e-05,
+      "loss": 1.0068,
+      "step": 721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034182468556555e-05,
+      "loss": 0.8344,
+      "step": 722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013362448409884e-05,
+      "loss": 0.7552,
+      "step": 723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0992537989080618e-05,
+      "loss": 0.668,
+      "step": 724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0971709181793349e-05,
+      "loss": 0.648,
+      "step": 725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0950876117791713e-05,
+      "loss": 0.9141,
+      "step": 726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0930038888338008e-05,
+      "loss": 0.6966,
+      "step": 727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0909197584712762e-05,
+      "loss": 0.8998,
+      "step": 728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0888352298214354e-05,
+      "loss": 0.7474,
+      "step": 729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086750312015862e-05,
+      "loss": 0.8072,
+      "step": 730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846650141878436e-05,
+      "loss": 0.8527,
+      "step": 731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.9252,
+      "step": 732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080493315005906e-05,
+      "loss": 0.7015,
+      "step": 733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0784069319267263e-05,
+      "loss": 0.7542,
+      "step": 734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0763202053744999e-05,
+      "loss": 0.9372,
+      "step": 735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742331444904378e-05,
+      "loss": 0.8933,
+      "step": 736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0721457584172159e-05,
+      "loss": 0.9459,
+      "step": 737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700580562989348e-05,
+      "loss": 0.7557,
+      "step": 738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679700472810788e-05,
+      "loss": 1.0306,
+      "step": 739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658817405104776e-05,
+      "loss": 1.019,
+      "step": 740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.063793145135265e-05,
+      "loss": 1.0368,
+      "step": 741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0617042703048381e-05,
+      "loss": 1.036,
+      "step": 742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.05961512516982e-05,
+      "loss": 0.7671,
+      "step": 743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575257188820162e-05,
+      "loss": 0.5994,
+      "step": 744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0554360605943771e-05,
+      "loss": 0.8978,
+      "step": 745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533461594609567e-05,
+      "loss": 0.7816,
+      "step": 746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051256024636873e-05,
+      "loss": 0.7602,
+      "step": 747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0491656652782678e-05,
+      "loss": 0.8232,
+      "step": 748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0470750905422662e-05,
+      "loss": 0.6742,
+      "step": 749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0449843095869371e-05,
+      "loss": 0.7592,
+      "step": 750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0428933315712528e-05,
+      "loss": 0.8451,
+      "step": 751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408021656550482e-05,
+      "loss": 0.9874,
+      "step": 752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387108209989825e-05,
+      "loss": 0.7119,
+      "step": 753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366193067644963e-05,
+      "loss": 1.1831,
+      "step": 754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345276321137748e-05,
+      "loss": 0.7028,
+      "step": 755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0324358062097045e-05,
+      "loss": 1.1761,
+      "step": 756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0303438382158355e-05,
+      "loss": 0.6671,
+      "step": 757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0282517372963401e-05,
+      "loss": 0.7879,
+      "step": 758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026159512615972e-05,
+      "loss": 0.719,
+      "step": 759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0240671733400285e-05,
+      "loss": 0.7732,
+      "step": 760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219747286343076e-05,
+      "loss": 0.697,
+      "step": 761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0198821876650702e-05,
+      "loss": 0.8636,
+      "step": 762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177895595989984e-05,
+      "loss": 0.8438,
+      "step": 763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0156968536031557e-05,
+      "loss": 0.8415,
+      "step": 764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0136040788449472e-05,
+      "loss": 0.8243,
+      "step": 765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115112444920793e-05,
+      "loss": 1.0075,
+      "step": 766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094183597125193e-05,
+      "loss": 0.7546,
+      "step": 767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0073254336744553e-05,
+      "loss": 0.9124,
+      "step": 768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052324755462564e-05,
+      "loss": 0.8877,
+      "step": 769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003139494496432e-05,
+      "loss": 0.7729,
+      "step": 770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0010464996935927e-05,
+      "loss": 1.0638,
+      "step": 771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.989535003064076e-06,
+      "loss": 0.7458,
+      "step": 772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.968605055035685e-06,
+      "loss": 0.7563,
+      "step": 773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94767524453744e-06,
+      "loss": 0.7349,
+      "step": 774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92674566325545e-06,
+      "loss": 0.7939,
+      "step": 775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905816402874812e-06,
+      "loss": 0.8304,
+      "step": 776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884887555079208e-06,
+      "loss": 0.9494,
+      "step": 777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863959211550528e-06,
+      "loss": 0.9037,
+      "step": 778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.843031463968447e-06,
+      "loss": 0.7556,
+      "step": 779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82210440401002e-06,
+      "loss": 0.9066,
+      "step": 780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.801178123349298e-06,
+      "loss": 0.6348,
+      "step": 781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780252713656927e-06,
+      "loss": 0.9619,
+      "step": 782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759328266599719e-06,
+      "loss": 0.9326,
+      "step": 783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738404873840282e-06,
+      "loss": 0.7014,
+      "step": 784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717482627036606e-06,
+      "loss": 0.9081,
+      "step": 785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.696561617841647e-06,
+      "loss": 0.743,
+      "step": 786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.675641937902957e-06,
+      "loss": 0.7922,
+      "step": 787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.654723678862257e-06,
+      "loss": 0.701,
+      "step": 788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63380693235504e-06,
+      "loss": 0.7826,
+      "step": 789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612891790010179e-06,
+      "loss": 0.6355,
+      "step": 790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59197834344952e-06,
+      "loss": 0.8862,
+      "step": 791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571066684287473e-06,
+      "loss": 0.8941,
+      "step": 792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.550156904130627e-06,
+      "loss": 0.67,
+      "step": 793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529249094577342e-06,
+      "loss": 0.7722,
+      "step": 794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.508343347217324e-06,
+      "loss": 0.8484,
+      "step": 795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487439753631273e-06,
+      "loss": 0.5622,
+      "step": 796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466538405390435e-06,
+      "loss": 0.6455,
+      "step": 797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44563939405623e-06,
+      "loss": 0.6225,
+      "step": 798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424742811179843e-06,
+      "loss": 0.7912,
+      "step": 799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.403848748301802e-06,
+      "loss": 0.7194,
+      "step": 800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382957296951619e-06,
+      "loss": 0.8855,
+      "step": 801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362068548647355e-06,
+      "loss": 0.5283,
+      "step": 802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.341182594895226e-06,
+      "loss": 0.6818,
+      "step": 803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320299527189214e-06,
+      "loss": 0.7315,
+      "step": 804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299419437010657e-06,
+      "loss": 0.8214,
+      "step": 805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.278542415827843e-06,
+      "loss": 0.7016,
+      "step": 806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257668555095624e-06,
+      "loss": 0.7175,
+      "step": 807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236797946255005e-06,
+      "loss": 1.0106,
+      "step": 808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.215930680732739e-06,
+      "loss": 0.8016,
+      "step": 809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19506684994094e-06,
+      "loss": 0.9014,
+      "step": 810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.7725,
+      "step": 811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153349858121566e-06,
+      "loss": 0.8251,
+      "step": 812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132496879841381e-06,
+      "loss": 0.9275,
+      "step": 813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11164770178565e-06,
+      "loss": 0.8944,
+      "step": 814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090802415287244e-06,
+      "loss": 0.7406,
+      "step": 815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069961111661993e-06,
+      "loss": 0.7485,
+      "step": 816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049123882208288e-06,
+      "loss": 0.78,
+      "step": 817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.028290818206655e-06,
+      "loss": 0.8187,
+      "step": 818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007462010919387e-06,
+      "loss": 1.0232,
+      "step": 819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986637551590118e-06,
+      "loss": 0.7125,
+      "step": 820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965817531443448e-06,
+      "loss": 0.7246,
+      "step": 821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945002041684525e-06,
+      "loss": 0.9166,
+      "step": 822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924191173498643e-06,
+      "loss": 0.7591,
+      "step": 823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90338501805086e-06,
+      "loss": 0.8935,
+      "step": 824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882583666485593e-06,
+      "loss": 0.6519,
+      "step": 825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861787209926201e-06,
+      "loss": 0.729,
+      "step": 826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.840995739474608e-06,
+      "loss": 0.7879,
+      "step": 827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820209346210899e-06,
+      "loss": 0.7537,
+      "step": 828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79942812119291e-06,
+      "loss": 0.7884,
+      "step": 829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77865215545584e-06,
+      "loss": 0.6975,
+      "step": 830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757881540011857e-06,
+      "loss": 0.8323,
+      "step": 831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737116365849671e-06,
+      "loss": 0.7971,
+      "step": 832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.716356723934176e-06,
+      "loss": 0.4204,
+      "step": 833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.695602705206024e-06,
+      "loss": 0.7577,
+      "step": 834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674854400581227e-06,
+      "loss": 0.6582,
+      "step": 835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654111900950776e-06,
+      "loss": 0.8046,
+      "step": 836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633375297180235e-06,
+      "loss": 0.5877,
+      "step": 837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61264468010932e-06,
+      "loss": 0.6849,
+      "step": 838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59192014055154e-06,
+      "loss": 0.7887,
+      "step": 839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571201769293779e-06,
+      "loss": 0.9509,
+      "step": 840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550489657095892e-06,
+      "loss": 0.9608,
+      "step": 841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529783894690322e-06,
+      "loss": 0.6763,
+      "step": 842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50908457278169e-06,
+      "loss": 0.997,
+      "step": 843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.488391782046408e-06,
+      "loss": 0.9099,
+      "step": 844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46770561313228e-06,
+      "loss": 0.9567,
+      "step": 845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447026156658093e-06,
+      "loss": 0.7009,
+      "step": 846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426353503213235e-06,
+      "loss": 0.8959,
+      "step": 847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405687743357296e-06,
+      "loss": 0.9662,
+      "step": 848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385028967619656e-06,
+      "loss": 0.7942,
+      "step": 849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364377266499111e-06,
+      "loss": 0.622,
+      "step": 850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.343732730463469e-06,
+      "loss": 0.668,
+      "step": 851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32309544994913e-06,
+      "loss": 0.7638,
+      "step": 852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30246551536073e-06,
+      "loss": 0.7187,
+      "step": 853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.281843017070723e-06,
+      "loss": 0.5374,
+      "step": 854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261228045418972e-06,
+      "loss": 0.9376,
+      "step": 855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240620690712385e-06,
+      "loss": 0.783,
+      "step": 856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2200210432245e-06,
+      "loss": 0.6839,
+      "step": 857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199429193195082e-06,
+      "loss": 0.5831,
+      "step": 858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178845230829751e-06,
+      "loss": 0.9138,
+      "step": 859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158269246299571e-06,
+      "loss": 0.7213,
+      "step": 860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.137701329740646e-06,
+      "loss": 0.6962,
+      "step": 861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11714157125375e-06,
+      "loss": 0.7903,
+      "step": 862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096590060903921e-06,
+      "loss": 0.6247,
+      "step": 863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.076046888720053e-06,
+      "loss": 0.6449,
+      "step": 864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055512144694526e-06,
+      "loss": 0.8916,
+      "step": 865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034985918782786e-06,
+      "loss": 0.817,
+      "step": 866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014468300902976e-06,
+      "loss": 0.8125,
+      "step": 867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993959380935527e-06,
+      "loss": 0.7292,
+      "step": 868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973459248722762e-06,
+      "loss": 0.6764,
+      "step": 869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952967994068512e-06,
+      "loss": 0.5519,
+      "step": 870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932485706737725e-06,
+      "loss": 0.7935,
+      "step": 871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912012476456052e-06,
+      "loss": 0.8502,
+      "step": 872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891548392909479e-06,
+      "loss": 0.6013,
+      "step": 873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871093545743927e-06,
+      "loss": 0.7525,
+      "step": 874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85064802456484e-06,
+      "loss": 0.8455,
+      "step": 875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83021191893682e-06,
+      "loss": 0.8244,
+      "step": 876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809785318383224e-06,
+      "loss": 0.6895,
+      "step": 877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789368312385762e-06,
+      "loss": 0.9804,
+      "step": 878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.76896099038412e-06,
+      "loss": 0.8157,
+      "step": 879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748563441775568e-06,
+      "loss": 0.7577,
+      "step": 880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728175755914542e-06,
+      "loss": 0.6727,
+      "step": 881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707798022112292e-06,
+      "loss": 0.8215,
+      "step": 882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.687430329636469e-06,
+      "loss": 1.0002,
+      "step": 883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667072767710722e-06,
+      "loss": 0.7499,
+      "step": 884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646725425514337e-06,
+      "loss": 0.4399,
+      "step": 885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626388392181831e-06,
+      "loss": 0.7406,
+      "step": 886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606061756802548e-06,
+      "loss": 0.6119,
+      "step": 887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585745608420301e-06,
+      "loss": 0.6156,
+      "step": 888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5654400360329415e-06,
+      "loss": 0.7721,
+      "step": 889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.8851,
+      "step": 890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524860975002319e-06,
+      "loss": 0.6232,
+      "step": 891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504587664121572e-06,
+      "loss": 0.7947,
+      "step": 892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484325284759979e-06,
+      "loss": 0.8415,
+      "step": 893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.46407392567986e-06,
+      "loss": 1.027,
+      "step": 894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.443833675595254e-06,
+      "loss": 0.653,
+      "step": 895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423604623171544e-06,
+      "loss": 0.6522,
+      "step": 896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.403386857025061e-06,
+      "loss": 0.6807,
+      "step": 897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383180465722679e-06,
+      "loss": 0.6694,
+      "step": 898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362985537781462e-06,
+      "loss": 0.7308,
+      "step": 899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342802161668249e-06,
+      "loss": 0.7176,
+      "step": 900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322630425799271e-06,
+      "loss": 0.7194,
+      "step": 901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302470418539771e-06,
+      "loss": 0.6474,
+      "step": 902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282322228203621e-06,
+      "loss": 0.7542,
+      "step": 903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.262185943052907e-06,
+      "loss": 0.7356,
+      "step": 904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2420616512975826e-06,
+      "loss": 0.7927,
+      "step": 905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221949441095053e-06,
+      "loss": 0.7103,
+      "step": 906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201849400549799e-06,
+      "loss": 0.643,
+      "step": 907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181761617712991e-06,
+      "loss": 0.5569,
+      "step": 908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1616861805821105e-06,
+      "loss": 0.9826,
+      "step": 909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.141623177100539e-06,
+      "loss": 0.7054,
+      "step": 910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.121572695157209e-06,
+      "loss": 0.6751,
+      "step": 911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.101534822586187e-06,
+      "loss": 0.8685,
+      "step": 912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.081509647166309e-06,
+      "loss": 0.9058,
+      "step": 913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061497256620793e-06,
+      "loss": 0.8665,
+      "step": 914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041497738616842e-06,
+      "loss": 0.7105,
+      "step": 915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021511180765272e-06,
+      "loss": 0.8806,
+      "step": 916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001537670620135e-06,
+      "loss": 0.4642,
+      "step": 917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.981577295678308e-06,
+      "loss": 0.4835,
+      "step": 918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961630143379138e-06,
+      "loss": 0.789,
+      "step": 919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94169630110405e-06,
+      "loss": 0.8083,
+      "step": 920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921775856176154e-06,
+      "loss": 0.7295,
+      "step": 921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90186889585988e-06,
+      "loss": 0.7307,
+      "step": 922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881975507360587e-06,
+      "loss": 0.8289,
+      "step": 923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.862095777824165e-06,
+      "loss": 0.7022,
+      "step": 924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.842229794336688e-06,
+      "loss": 0.5814,
+      "step": 925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8223776439240075e-06,
+      "loss": 0.6448,
+      "step": 926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802539413551368e-06,
+      "loss": 0.7309,
+      "step": 927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782715190123046e-06,
+      "loss": 0.6951,
+      "step": 928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7629050604819594e-06,
+      "loss": 0.7636,
+      "step": 929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7431091114092766e-06,
+      "loss": 0.7437,
+      "step": 930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723327429624052e-06,
+      "loss": 0.6617,
+      "step": 931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703560101782847e-06,
+      "loss": 0.7841,
+      "step": 932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683807214479323e-06,
+      "loss": 0.6457,
+      "step": 933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664068854243905e-06,
+      "loss": 0.679,
+      "step": 934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644345107543366e-06,
+      "loss": 0.6486,
+      "step": 935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624636060780467e-06,
+      "loss": 0.8788,
+      "step": 936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6049418002935775e-06,
+      "loss": 0.7345,
+      "step": 937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585262412356284e-06,
+      "loss": 0.6826,
+      "step": 938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.565597983177028e-06,
+      "loss": 0.7681,
+      "step": 939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.545948598898727e-06,
+      "loss": 0.7052,
+      "step": 940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.526314345598377e-06,
+      "loss": 0.612,
+      "step": 941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506695309286706e-06,
+      "loss": 0.75,
+      "step": 942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487091575907776e-06,
+      "loss": 0.6772,
+      "step": 943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4675032313386055e-06,
+      "loss": 0.9731,
+      "step": 944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447930361388812e-06,
+      "loss": 0.6734,
+      "step": 945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428373051800221e-06,
+      "loss": 0.6216,
+      "step": 946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408831388246482e-06,
+      "loss": 0.5586,
+      "step": 947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.38930545633272e-06,
+      "loss": 0.717,
+      "step": 948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36979534159514e-06,
+      "loss": 0.6395,
+      "step": 949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350301129500651e-06,
+      "loss": 0.6771,
+      "step": 950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3308229054465075e-06,
+      "loss": 0.5819,
+      "step": 951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311360754759923e-06,
+      "loss": 0.7821,
+      "step": 952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291914762697695e-06,
+      "loss": 0.6041,
+      "step": 953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.272485014445844e-06,
+      "loss": 0.6796,
+      "step": 954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2530715951192265e-06,
+      "loss": 0.6541,
+      "step": 955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.233674589761163e-06,
+      "loss": 0.5839,
+      "step": 956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.214294083343083e-06,
+      "loss": 0.7272,
+      "step": 957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194930160764128e-06,
+      "loss": 0.808,
+      "step": 958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175582906850797e-06,
+      "loss": 0.8211,
+      "step": 959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156252406356576e-06,
+      "loss": 0.9395,
+      "step": 960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136938743961543e-06,
+      "loss": 0.4457,
+      "step": 961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117642004272026e-06,
+      "loss": 0.7951,
+      "step": 962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0983622718202286e-06,
+      "loss": 0.9041,
+      "step": 963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07909963106383e-06,
+      "loss": 0.7472,
+      "step": 964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059854166385653e-06,
+      "loss": 0.6402,
+      "step": 965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040625962093275e-06,
+      "loss": 0.7034,
+      "step": 966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021415102418659e-06,
+      "loss": 0.5582,
+      "step": 967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0022216715177915e-06,
+      "loss": 0.7399,
+      "step": 968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.623,
+      "step": 969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.963887432279119e-06,
+      "loss": 0.9114,
+      "step": 970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.944746791870062e-06,
+      "loss": 0.8572,
+      "step": 971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.925623916091514e-06,
+      "loss": 0.6522,
+      "step": 972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.906518888714029e-06,
+      "loss": 0.7311,
+      "step": 973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88743179342998e-06,
+      "loss": 0.7049,
+      "step": 974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8683627138531885e-06,
+      "loss": 0.6232,
+      "step": 975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849311733518539e-06,
+      "loss": 0.6607,
+      "step": 976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830278935881644e-06,
+      "loss": 0.7205,
+      "step": 977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811264404318468e-06,
+      "loss": 0.9883,
+      "step": 978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792268222124943e-06,
+      "loss": 0.8189,
+      "step": 979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773290472516631e-06,
+      "loss": 0.5899,
+      "step": 980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.754331238628339e-06,
+      "loss": 0.7482,
+      "step": 981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.735390603513769e-06,
+      "loss": 0.7866,
+      "step": 982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716468650145152e-06,
+      "loss": 0.7033,
+      "step": 983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.697565461412866e-06,
+      "loss": 0.5855,
+      "step": 984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.678681120125102e-06,
+      "loss": 0.706,
+      "step": 985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6598157090074875e-06,
+      "loss": 0.8853,
+      "step": 986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.640969310702708e-06,
+      "loss": 0.7026,
+      "step": 987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6221420077701795e-06,
+      "loss": 0.6372,
+      "step": 988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603333882685655e-06,
+      "loss": 0.7702,
+      "step": 989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584545017840886e-06,
+      "loss": 0.638,
+      "step": 990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565775495543238e-06,
+      "loss": 0.7515,
+      "step": 991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5470253980153555e-06,
+      "loss": 0.5526,
+      "step": 992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528294807394788e-06,
+      "loss": 0.6354,
+      "step": 993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509583805733631e-06,
+      "loss": 0.6034,
+      "step": 994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.490892474998169e-06,
+      "loss": 0.7049,
+      "step": 995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.472220897068505e-06,
+      "loss": 0.6691,
+      "step": 996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453569153738227e-06,
+      "loss": 0.7224,
+      "step": 997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434937326714031e-06,
+      "loss": 0.7885,
+      "step": 998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.416325497615356e-06,
+      "loss": 0.7007,
+      "step": 999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.397733747974045e-06,
+      "loss": 0.6495,
+      "step": 1000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.379162159233989e-06,
+      "loss": 1.0051,
+      "step": 1001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.360610812750742e-06,
+      "loss": 0.7239,
+      "step": 1002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3420797897912e-06,
+      "loss": 0.6127,
+      "step": 1003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3235691715332185e-06,
+      "loss": 0.8505,
+      "step": 1004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305079039065271e-06,
+      "loss": 0.4961,
+      "step": 1005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286609473386093e-06,
+      "loss": 0.666,
+      "step": 1006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26816055540432e-06,
+      "loss": 0.7696,
+      "step": 1007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249732365938142e-06,
+      "loss": 0.7836,
+      "step": 1008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231324985714942e-06,
+      "loss": 0.5329,
+      "step": 1009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212938495370939e-06,
+      "loss": 0.7618,
+      "step": 1010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194572975450847e-06,
+      "loss": 0.6802,
+      "step": 1011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176228506407523e-06,
+      "loss": 0.7101,
+      "step": 1012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15790516860159e-06,
+      "loss": 0.7519,
+      "step": 1013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.139603042301118e-06,
+      "loss": 0.9543,
+      "step": 1014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.121322207681252e-06,
+      "loss": 0.5607,
+      "step": 1015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.103062744823861e-06,
+      "loss": 0.6489,
+      "step": 1016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.084824733717197e-06,
+      "loss": 0.6329,
+      "step": 1017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.066608254255545e-06,
+      "loss": 0.7712,
+      "step": 1018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0484133862388505e-06,
+      "loss": 0.7605,
+      "step": 1019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0302402093724025e-06,
+      "loss": 0.7083,
+      "step": 1020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012088803266461e-06,
+      "loss": 0.8589,
+      "step": 1021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993959247435923e-06,
+      "loss": 0.3896,
+      "step": 1022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975851621299959e-06,
+      "loss": 0.9145,
+      "step": 1023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957766004181681e-06,
+      "loss": 0.7796,
+      "step": 1024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939702475307776e-06,
+      "loss": 0.7618,
+      "step": 1025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921661113808186e-06,
+      "loss": 0.6221,
+      "step": 1026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903641998715728e-06,
+      "loss": 0.7982,
+      "step": 1027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885645208965779e-06,
+      "loss": 0.7585,
+      "step": 1028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.867670823395915e-06,
+      "loss": 0.8069,
+      "step": 1029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849718920745556e-06,
+      "loss": 0.6446,
+      "step": 1030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8317895796556455e-06,
+      "loss": 0.7561,
+      "step": 1031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.813882878668291e-06,
+      "loss": 0.7561,
+      "step": 1032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.795998896226411e-06,
+      "loss": 0.6164,
+      "step": 1033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7781377106734105e-06,
+      "loss": 0.645,
+      "step": 1034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760299400252832e-06,
+      "loss": 0.7348,
+      "step": 1035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742484043108006e-06,
+      "loss": 0.8045,
+      "step": 1036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724691717281711e-06,
+      "loss": 0.7948,
+      "step": 1037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706922500715841e-06,
+      "loss": 1.0262,
+      "step": 1038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689176471251043e-06,
+      "loss": 0.9289,
+      "step": 1039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671453706626402e-06,
+      "loss": 0.8146,
+      "step": 1040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653754284479087e-06,
+      "loss": 0.6265,
+      "step": 1041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6360782823440006e-06,
+      "loss": 0.6143,
+      "step": 1042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618425777653465e-06,
+      "loss": 0.6355,
+      "step": 1043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.600796847736863e-06,
+      "loss": 0.5679,
+      "step": 1044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.583191569820295e-06,
+      "loss": 0.8092,
+      "step": 1045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.565610021026266e-06,
+      "loss": 0.6558,
+      "step": 1046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.548052278373327e-06,
+      "loss": 0.7752,
+      "step": 1047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 0.6883,
+      "step": 1048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.513008519043128e-06,
+      "loss": 0.6957,
+      "step": 1049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.495522655880193e-06,
+      "loss": 0.64,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478060905886309e-06,
+      "loss": 0.5229,
+      "step": 1051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460623345555235e-06,
+      "loss": 0.7186,
+      "step": 1052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443210051274749e-06,
+      "loss": 0.798,
+      "step": 1053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.42582109932634e-06,
+      "loss": 0.8893,
+      "step": 1054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408456565884863e-06,
+      "loss": 0.631,
+      "step": 1055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391116527018192e-06,
+      "loss": 0.6812,
+      "step": 1056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3738010586869105e-06,
+      "loss": 0.8687,
+      "step": 1057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356510236743966e-06,
+      "loss": 0.5581,
+      "step": 1058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.339244136934327e-06,
+      "loss": 0.833,
+      "step": 1059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.32200283489468e-06,
+      "loss": 0.5412,
+      "step": 1060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.304786406153073e-06,
+      "loss": 0.8145,
+      "step": 1061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.287594926128589e-06,
+      "loss": 0.8303,
+      "step": 1062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.270428470131024e-06,
+      "loss": 0.8466,
+      "step": 1063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2532871133605546e-06,
+      "loss": 0.7351,
+      "step": 1064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.236170930907404e-06,
+      "loss": 0.7297,
+      "step": 1065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219079997751515e-06,
+      "loss": 0.7313,
+      "step": 1066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.202014388762227e-06,
+      "loss": 0.8383,
+      "step": 1067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.184974178697932e-06,
+      "loss": 0.7202,
+      "step": 1068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167959442205769e-06,
+      "loss": 0.733,
+      "step": 1069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150970253821287e-06,
+      "loss": 0.5812,
+      "step": 1070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134006687968105e-06,
+      "loss": 0.4754,
+      "step": 1071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117068818957617e-06,
+      "loss": 0.7126,
+      "step": 1072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100156720988627e-06,
+      "loss": 0.5688,
+      "step": 1073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.083270468147065e-06,
+      "loss": 0.7572,
+      "step": 1074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.066410134405635e-06,
+      "loss": 0.6395,
+      "step": 1075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.0495757936234915e-06,
+      "loss": 0.6818,
+      "step": 1076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.032767519545932e-06,
+      "loss": 0.6393,
+      "step": 1077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.015985385804065e-06,
+      "loss": 0.4327,
+      "step": 1078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9992294659144816e-06,
+      "loss": 0.6692,
+      "step": 1079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.982499833278946e-06,
+      "loss": 0.8291,
+      "step": 1080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9657965611840636e-06,
+      "loss": 0.7373,
+      "step": 1081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94911972280096e-06,
+      "loss": 0.6257,
+      "step": 1082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932469391184968e-06,
+      "loss": 0.6047,
+      "step": 1083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915845639275309e-06,
+      "loss": 0.5854,
+      "step": 1084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.899248539894756e-06,
+      "loss": 0.6829,
+      "step": 1085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8826781657493374e-06,
+      "loss": 0.6714,
+      "step": 1086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866134589428004e-06,
+      "loss": 0.6594,
+      "step": 1087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84961788340231e-06,
+      "loss": 0.3894,
+      "step": 1088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.833128120026107e-06,
+      "loss": 0.8394,
+      "step": 1089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.816665371535223e-06,
+      "loss": 0.7187,
+      "step": 1090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.80022971004713e-06,
+      "loss": 0.8104,
+      "step": 1091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783821207560653e-06,
+      "loss": 0.5395,
+      "step": 1092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7674399359556423e-06,
+      "loss": 0.6831,
+      "step": 1093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7510859669926545e-06,
+      "loss": 0.5605,
+      "step": 1094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7347593723126507e-06,
+      "loss": 0.5516,
+      "step": 1095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.718460223436662e-06,
+      "loss": 0.7403,
+      "step": 1096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7021885917655e-06,
+      "loss": 0.7927,
+      "step": 1097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6859445485794345e-06,
+      "loss": 0.9325,
+      "step": 1098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.669728165037866e-06,
+      "loss": 0.7251,
+      "step": 1099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.653539512179044e-06,
+      "loss": 0.6149,
+      "step": 1100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6373786609197336e-06,
+      "loss": 0.6681,
+      "step": 1101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6212456820549047e-06,
+      "loss": 0.8337,
+      "step": 1102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6051406462574377e-06,
+      "loss": 0.5706,
+      "step": 1103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.589063624077802e-06,
+      "loss": 0.9666,
+      "step": 1104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.573014685943742e-06,
+      "loss": 0.7274,
+      "step": 1105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5569939021599843e-06,
+      "loss": 0.7435,
+      "step": 1106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.541001342907917e-06,
+      "loss": 0.8181,
+      "step": 1107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5250370782452903e-06,
+      "loss": 0.7684,
+      "step": 1108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5091011781059014e-06,
+      "loss": 0.5358,
+      "step": 1109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4931937122992986e-06,
+      "loss": 0.959,
+      "step": 1110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4773147505104586e-06,
+      "loss": 0.6861,
+      "step": 1111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4614643622995014e-06,
+      "loss": 0.6678,
+      "step": 1112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.445642617101378e-06,
+      "loss": 0.5765,
+      "step": 1113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.429849584225552e-06,
+      "loss": 0.8282,
+      "step": 1114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.414085332855721e-06,
+      "loss": 0.6932,
+      "step": 1115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3983499320494993e-06,
+      "loss": 0.628,
+      "step": 1116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3826434507381078e-06,
+      "loss": 0.7046,
+      "step": 1117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.366965957726095e-06,
+      "loss": 0.8682,
+      "step": 1118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3513175216910077e-06,
+      "loss": 0.7117,
+      "step": 1119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.335698211183115e-06,
+      "loss": 0.515,
+      "step": 1120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.320108094625094e-06,
+      "loss": 0.6268,
+      "step": 1121
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.304547240311732e-06,
+      "loss": 0.5603,
+      "step": 1122
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2890157164096315e-06,
+      "loss": 0.6765,
+      "step": 1123
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2735135909569093e-06,
+      "loss": 0.6429,
+      "step": 1124
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.25804093186289e-06,
+      "loss": 0.5753,
+      "step": 1125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2425978069078266e-06,
+      "loss": 0.5302,
+      "step": 1126
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 0.6865,
+      "step": 1127
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2118004298883733e-06,
+      "loss": 0.6185,
+      "step": 1128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196446312736401e-06,
+      "loss": 0.6255,
+      "step": 1129
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.181121999547635e-06,
+      "loss": 0.6627,
+      "step": 1130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.165827557452468e-06,
+      "loss": 0.7987,
+      "step": 1131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1505630534504463e-06,
+      "loss": 0.5806,
+      "step": 1132
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1353285544099664e-06,
+      "loss": 0.7365,
+      "step": 1133
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.120124127067978e-06,
+      "loss": 0.677,
+      "step": 1134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.104949838029704e-06,
+      "loss": 0.8424,
+      "step": 1135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0898057537683403e-06,
+      "loss": 0.7804,
+      "step": 1136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0746919406247634e-06,
+      "loss": 0.5798,
+      "step": 1137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.059608464807248e-06,
+      "loss": 0.7545,
+      "step": 1138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0445553923911687e-06,
+      "loss": 0.6401,
+      "step": 1139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0295327893187086e-06,
+      "loss": 0.725,
+      "step": 1140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0145407213985878e-06,
+      "loss": 0.7314,
+      "step": 1141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.999579254305748e-06,
+      "loss": 0.7078,
+      "step": 1142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9846484535810884e-06,
+      "loss": 0.6414,
+      "step": 1143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9697483846311724e-06,
+      "loss": 0.5637,
+      "step": 1144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9548791127279286e-06,
+      "loss": 0.6947,
+      "step": 1145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.940040703008381e-06,
+      "loss": 0.6049,
+      "step": 1146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9252332204743615e-06,
+      "loss": 0.5955,
+      "step": 1147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.910456729992207e-06,
+      "loss": 0.8021,
+      "step": 1148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8957112962925037e-06,
+      "loss": 0.7591,
+      "step": 1149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8809969839697814e-06,
+      "loss": 0.5431,
+      "step": 1150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.86631385748224e-06,
+      "loss": 0.8251,
+      "step": 1151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.851661981151466e-06,
+      "loss": 0.6248,
+      "step": 1152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.837041419162151e-06,
+      "loss": 0.6087,
+      "step": 1153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8224522355618024e-06,
+      "loss": 0.5434,
+      "step": 1154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.807894494260478e-06,
+      "loss": 0.674,
+      "step": 1155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7933682590305e-06,
+      "loss": 0.73,
+      "step": 1156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7788735935061606e-06,
+      "loss": 0.8597,
+      "step": 1157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7644105611834693e-06,
+      "loss": 0.6182,
+      "step": 1158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7499792254198597e-06,
+      "loss": 0.8006,
+      "step": 1159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735579649433905e-06,
+      "loss": 0.6252,
+      "step": 1160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.721211896305059e-06,
+      "loss": 0.658,
+      "step": 1161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.706876028973372e-06,
+      "loss": 0.611,
+      "step": 1162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6925721102392043e-06,
+      "loss": 0.4574,
+      "step": 1163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678300202762969e-06,
+      "loss": 0.6949,
+      "step": 1164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.664060369064847e-06,
+      "loss": 0.5682,
+      "step": 1165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6498526715245143e-06,
+      "loss": 0.5362,
+      "step": 1166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.635677172380875e-06,
+      "loss": 0.6507,
+      "step": 1167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6215339337317712e-06,
+      "loss": 0.7918,
+      "step": 1168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6074230175337347e-06,
+      "loss": 0.7415,
+      "step": 1169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.593344485601702e-06,
+      "loss": 0.6548,
+      "step": 1170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.579298399608741e-06,
+      "loss": 0.7254,
+      "step": 1171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.565284821085787e-06,
+      "loss": 0.6199,
+      "step": 1172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5513038114213794e-06,
+      "loss": 0.5958,
+      "step": 1173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.53735543186137e-06,
+      "loss": 0.575,
+      "step": 1174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5234397435086832e-06,
+      "loss": 0.9039,
+      "step": 1175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.50955680732303e-06,
+      "loss": 0.564,
+      "step": 1176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4957066841206414e-06,
+      "loss": 0.9338,
+      "step": 1177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4818894345740117e-06,
+      "loss": 0.804,
+      "step": 1178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4681051192116267e-06,
+      "loss": 0.5413,
+      "step": 1179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.454353798417698e-06,
+      "loss": 0.4573,
+      "step": 1180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.440635532431901e-06,
+      "loss": 0.429,
+      "step": 1181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.426950381349109e-06,
+      "loss": 0.7027,
+      "step": 1182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.413298405119128e-06,
+      "loss": 0.5972,
+      "step": 1183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3996796635464393e-06,
+      "loss": 0.4859,
+      "step": 1184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.386094216289939e-06,
+      "loss": 0.7188,
+      "step": 1185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.372542122862661e-06,
+      "loss": 0.6162,
+      "step": 1186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3590234426315396e-06,
+      "loss": 0.6116,
+      "step": 1187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.345538234817135e-06,
+      "loss": 0.5575,
+      "step": 1188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3320865584933694e-06,
+      "loss": 0.6728,
+      "step": 1189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3186684725872855e-06,
+      "loss": 0.8614,
+      "step": 1190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3052840358787687e-06,
+      "loss": 0.6153,
+      "step": 1191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2919333070003056e-06,
+      "loss": 0.602,
+      "step": 1192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2786163444367214e-06,
+      "loss": 0.528,
+      "step": 1193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265333206524919e-06,
+      "loss": 0.6123,
+      "step": 1194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2520839514536297e-06,
+      "loss": 0.7723,
+      "step": 1195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2388686372631586e-06,
+      "loss": 0.5088,
+      "step": 1196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2256873218451167e-06,
+      "loss": 0.6385,
+      "step": 1197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2125400629421923e-06,
+      "loss": 0.7624,
+      "step": 1198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.19942691814788e-06,
+      "loss": 0.6488,
+      "step": 1199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.186347944906224e-06,
+      "loss": 0.6695,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.173303200511586e-06,
+      "loss": 0.5402,
+      "step": 1201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1602927421083817e-06,
+      "loss": 0.5804,
+      "step": 1202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.147316626690824e-06,
+      "loss": 0.5325,
+      "step": 1203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.134374911102688e-06,
+      "loss": 0.7602,
+      "step": 1204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1214676520370567e-06,
+      "loss": 0.7783,
+      "step": 1205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 0.6411,
+      "step": 1206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0957567294906633e-06,
+      "loss": 0.7189,
+      "step": 1207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0829531786403666e-06,
+      "loss": 0.5476,
+      "step": 1208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070184309572999e-06,
+      "loss": 0.6509,
+      "step": 1209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.057450178224464e-06,
+      "loss": 0.7513,
+      "step": 1210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.044750840378489e-06,
+      "loss": 0.7379,
+      "step": 1211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0320863516663826e-06,
+      "loss": 0.7514,
+      "step": 1212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.019456767566794e-06,
+      "loss": 0.6873,
+      "step": 1213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0068621434054613e-06,
+      "loss": 0.6967,
+      "step": 1214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9943025343549837e-06,
+      "loss": 0.7971,
+      "step": 1215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9817779954345683e-06,
+      "loss": 0.7452,
+      "step": 1216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9692885815097884e-06,
+      "loss": 0.6814,
+      "step": 1217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9568343472923524e-06,
+      "loss": 0.7421,
+      "step": 1218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9444153473398595e-06,
+      "loss": 0.6543,
+      "step": 1219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9320316360555525e-06,
+      "loss": 0.5772,
+      "step": 1220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919683267688095e-06,
+      "loss": 0.5449,
+      "step": 1221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9073702963313245e-06,
+      "loss": 0.7731,
+      "step": 1222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8950927759240111e-06,
+      "loss": 0.6612,
+      "step": 1223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8828507602496326e-06,
+      "loss": 0.6438,
+      "step": 1224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.870644302936132e-06,
+      "loss": 0.6427,
+      "step": 1225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8584734574556828e-06,
+      "loss": 0.8236,
+      "step": 1226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8463382771244554e-06,
+      "loss": 0.4808,
+      "step": 1227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.834238815102387e-06,
+      "loss": 0.5901,
+      "step": 1228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8221751243929376e-06,
+      "loss": 0.6244,
+      "step": 1229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8101472578428736e-06,
+      "loss": 0.6371,
+      "step": 1230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.798155268142029e-06,
+      "loss": 0.7068,
+      "step": 1231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.786199207823064e-06,
+      "loss": 0.6294,
+      "step": 1232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.774279129261255e-06,
+      "loss": 0.7077,
+      "step": 1233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7623950846742543e-06,
+      "loss": 0.5818,
+      "step": 1234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7505471261218542e-06,
+      "loss": 0.6221,
+      "step": 1235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7387353055057787e-06,
+      "loss": 0.5998,
+      "step": 1236
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7269596745694295e-06,
+      "loss": 0.6977,
+      "step": 1237
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7152202848976874e-06,
+      "loss": 0.5594,
+      "step": 1238
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7035171879166667e-06,
+      "loss": 0.7077,
+      "step": 1239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6918504348934972e-06,
+      "loss": 0.7839,
+      "step": 1240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.680220076936101e-06,
+      "loss": 0.7338,
+      "step": 1241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6686261649929636e-06,
+      "loss": 0.6401,
+      "step": 1242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6570687498529114e-06,
+      "loss": 0.7917,
+      "step": 1243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6455478821448956e-06,
+      "loss": 0.6874,
+      "step": 1244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6340636123377661e-06,
+      "loss": 0.5414,
+      "step": 1245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6226159907400452e-06,
+      "loss": 0.7897,
+      "step": 1246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.611205067499716e-06,
+      "loss": 0.743,
+      "step": 1247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5998308926040028e-06,
+      "loss": 0.4696,
+      "step": 1248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5884935158791393e-06,
+      "loss": 0.73,
+      "step": 1249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.577192986990166e-06,
+      "loss": 0.4349,
+      "step": 1250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5659293554407061e-06,
+      "loss": 0.5869,
+      "step": 1251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5547026705727453e-06,
+      "loss": 0.499,
+      "step": 1252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5435129815664195e-06,
+      "loss": 0.5725,
+      "step": 1253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5323603374398021e-06,
+      "loss": 0.7754,
+      "step": 1254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5212447870486835e-06,
+      "loss": 0.7479,
+      "step": 1255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5101663790863597e-06,
+      "loss": 0.6868,
+      "step": 1256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4991251620834213e-06,
+      "loss": 0.6088,
+      "step": 1257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4881211844075317e-06,
+      "loss": 0.7008,
+      "step": 1258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.477154494263231e-06,
+      "loss": 0.6697,
+      "step": 1259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.466225139691706e-06,
+      "loss": 0.4531,
+      "step": 1260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4553331685705986e-06,
+      "loss": 0.5632,
+      "step": 1261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4444786286137847e-06,
+      "loss": 0.5391,
+      "step": 1262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4336615673711617e-06,
+      "loss": 0.6972,
+      "step": 1263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.422882032228452e-06,
+      "loss": 0.9929,
+      "step": 1264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.412140070406991e-06,
+      "loss": 0.9555,
+      "step": 1265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4014357289635093e-06,
+      "loss": 0.7442,
+      "step": 1266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3907690547899466e-06,
+      "loss": 0.4465,
+      "step": 1267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3801400946132304e-06,
+      "loss": 0.5789,
+      "step": 1268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3695488949950764e-06,
+      "loss": 0.5879,
+      "step": 1269
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3589955023317858e-06,
+      "loss": 0.7894,
+      "step": 1270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3484799628540424e-06,
+      "loss": 0.8426,
+      "step": 1271
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3380023226267037e-06,
+      "loss": 0.6864,
+      "step": 1272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3275626275486076e-06,
+      "loss": 0.6158,
+      "step": 1273
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3171609233523696e-06,
+      "loss": 0.8419,
+      "step": 1274
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3067972556041753e-06,
+      "loss": 0.5988,
+      "step": 1275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2964716697035894e-06,
+      "loss": 0.7875,
+      "step": 1276
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2861842108833545e-06,
+      "loss": 0.68,
+      "step": 1277
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.275934924209188e-06,
+      "loss": 0.7033,
+      "step": 1278
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2657238545795924e-06,
+      "loss": 0.7607,
+      "step": 1279
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.255551046725656e-06,
+      "loss": 0.7912,
+      "step": 1280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.245416545210848e-06,
+      "loss": 0.6783,
+      "step": 1281
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.235320394430839e-06,
+      "loss": 0.7052,
+      "step": 1282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.225262638613296e-06,
+      "loss": 0.4525,
+      "step": 1283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2152433218176906e-06,
+      "loss": 0.6644,
+      "step": 1284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 0.7664,
+      "step": 1285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1953201806880543e-06,
+      "loss": 0.646,
+      "step": 1286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1854164436302585e-06,
+      "loss": 0.6547,
+      "step": 1287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1755513201464942e-06,
+      "loss": 0.6318,
+      "step": 1288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1657248534523768e-06,
+      "loss": 0.7454,
+      "step": 1289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1559370865941833e-06,
+      "loss": 0.6378,
+      "step": 1290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1461880624486622e-06,
+      "loss": 0.49,
+      "step": 1291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.136477823722837e-06,
+      "loss": 0.7684,
+      "step": 1292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1268064129538325e-06,
+      "loss": 0.6478,
+      "step": 1293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1171738725086833e-06,
+      "loss": 0.6242,
+      "step": 1294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1075802445841367e-06,
+      "loss": 0.5972,
+      "step": 1295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.098025571206489e-06,
+      "loss": 0.7083,
+      "step": 1296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0885098942313843e-06,
+      "loss": 0.6538,
+      "step": 1297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.07903325534364e-06,
+      "loss": 0.6693,
+      "step": 1298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0695956960570607e-06,
+      "loss": 0.6744,
+      "step": 1299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0601972577142583e-06,
+      "loss": 0.6549,
+      "step": 1300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0508379814864633e-06,
+      "loss": 0.7489,
+      "step": 1301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0415179083733595e-06,
+      "loss": 0.7953,
+      "step": 1302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0322370792028936e-06,
+      "loss": 0.6414,
+      "step": 1303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0229955346310928e-06,
+      "loss": 0.6207,
+      "step": 1304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.013793315141899e-06,
+      "loss": 0.4866,
+      "step": 1305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0046304610469794e-06,
+      "loss": 0.8297,
+      "step": 1306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.955070124855592e-07,
+      "loss": 0.6215,
+      "step": 1307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.864230094242444e-07,
+      "loss": 0.5916,
+      "step": 1308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.773784916568363e-07,
+      "loss": 0.5763,
+      "step": 1309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.683734988041705e-07,
+      "loss": 0.8333,
+      "step": 1310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.594080703139364e-07,
+      "loss": 0.6872,
+      "step": 1311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.504822454605078e-07,
+      "loss": 0.6878,
+      "step": 1312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.415960633447674e-07,
+      "loss": 0.6065,
+      "step": 1313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.327495628939376e-07,
+      "loss": 0.5778,
+      "step": 1314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.239427828614067e-07,
+      "loss": 0.6641,
+      "step": 1315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.151757618265656e-07,
+      "loss": 0.6984,
+      "step": 1316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.064485381946364e-07,
+      "loss": 0.6404,
+      "step": 1317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.977611501964945e-07,
+      "loss": 0.4855,
+      "step": 1318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.891136358885188e-07,
+      "loss": 0.6243,
+      "step": 1319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.80506033152414e-07,
+      "loss": 0.5621,
+      "step": 1320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.719383796950398e-07,
+      "loss": 0.7138,
+      "step": 1321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.634107130482583e-07,
+      "loss": 0.7704,
+      "step": 1322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.54923070568765e-07,
+      "loss": 0.555,
+      "step": 1323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.464754894379157e-07,
+      "loss": 0.6067,
+      "step": 1324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.380680066615787e-07,
+      "loss": 0.6778,
+      "step": 1325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.297006590699653e-07,
+      "loss": 0.7357,
+      "step": 1326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.213734833174635e-07,
+      "loss": 0.7122,
+      "step": 1327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.130865158824908e-07,
+      "loss": 0.6462,
+      "step": 1328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.048397930673168e-07,
+      "loss": 0.5978,
+      "step": 1329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.966333509979207e-07,
+      "loss": 0.5989,
+      "step": 1330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.884672256238246e-07,
+      "loss": 0.7428,
+      "step": 1331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.803414527179343e-07,
+      "loss": 0.4109,
+      "step": 1332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.722560678763879e-07,
+      "loss": 0.5326,
+      "step": 1333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.642111065183999e-07,
+      "loss": 0.8095,
+      "step": 1334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.562066038860994e-07,
+      "loss": 0.3998,
+      "step": 1335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.482425950443828e-07,
+      "loss": 0.4552,
+      "step": 1336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.403191148807598e-07,
+      "loss": 0.5983,
+      "step": 1337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.324361981051909e-07,
+      "loss": 0.6676,
+      "step": 1338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.245938792499496e-07,
+      "loss": 0.6618,
+      "step": 1339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.167921926694621e-07,
+      "loss": 0.8047,
+      "step": 1340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.090311725401588e-07,
+      "loss": 0.6471,
+      "step": 1341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.013108528603252e-07,
+      "loss": 0.7749,
+      "step": 1342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.936312674499535e-07,
+      "loss": 0.6222,
+      "step": 1343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.859924499505899e-07,
+      "loss": 0.565,
+      "step": 1344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.783944338251925e-07,
+      "loss": 0.5916,
+      "step": 1345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.708372523579864e-07,
+      "loss": 0.6422,
+      "step": 1346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.633209386543083e-07,
+      "loss": 0.7611,
+      "step": 1347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.558455256404739e-07,
+      "loss": 0.843,
+      "step": 1348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.484110460636239e-07,
+      "loss": 0.5538,
+      "step": 1349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.410175324915824e-07,
+      "loss": 0.9871,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.336650173127224e-07,
+      "loss": 0.7375,
+      "step": 1351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.263535327358094e-07,
+      "loss": 0.8378,
+      "step": 1352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.190831107898743e-07,
+      "loss": 0.8079,
+      "step": 1353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.118537833240645e-07,
+      "loss": 0.6197,
+      "step": 1354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.046655820075076e-07,
+      "loss": 0.5733,
+      "step": 1355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.975185383291715e-07,
+      "loss": 0.5503,
+      "step": 1356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.904126835977309e-07,
+      "loss": 0.5361,
+      "step": 1357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.833480489414178e-07,
+      "loss": 0.6707,
+      "step": 1358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.763246653079024e-07,
+      "loss": 0.6195,
+      "step": 1359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.693425634641447e-07,
+      "loss": 0.5342,
+      "step": 1360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.624017739962639e-07,
+      "loss": 0.6316,
+      "step": 1361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.555023273094029e-07,
+      "loss": 0.6579,
+      "step": 1362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.486442536276026e-07,
+      "loss": 0.8218,
+      "step": 1363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 0.8092,
+      "step": 1364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.350523452689826e-07,
+      "loss": 0.709,
+      "step": 1365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.283185701335125e-07,
+      "loss": 0.7974,
+      "step": 1366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.216262870855282e-07,
+      "loss": 0.5458,
+      "step": 1367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.149755254415556e-07,
+      "loss": 0.6991,
+      "step": 1368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.083663143362316e-07,
+      "loss": 0.6421,
+      "step": 1369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.017986827221733e-07,
+      "loss": 0.6066,
+      "step": 1370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.952726593698509e-07,
+      "loss": 0.7421,
+      "step": 1371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.887882728674687e-07,
+      "loss": 0.716,
+      "step": 1372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.823455516208275e-07,
+      "loss": 0.8488,
+      "step": 1373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7594452385321455e-07,
+      "loss": 0.5161,
+      "step": 1374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695852176052662e-07,
+      "loss": 0.6216,
+      "step": 1375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6326766073485627e-07,
+      "loss": 0.6651,
+      "step": 1376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5699188091696823e-07,
+      "loss": 0.7079,
+      "step": 1377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.507579056435729e-07,
+      "loss": 0.9289,
+      "step": 1378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.4456576222351224e-07,
+      "loss": 0.565,
+      "step": 1379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.384154777823801e-07,
+      "loss": 0.5149,
+      "step": 1380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.32307079262394e-07,
+      "loss": 0.4998,
+      "step": 1381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.262405934222902e-07,
+      "loss": 0.7036,
+      "step": 1382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.202160468371985e-07,
+      "loss": 0.6881,
+      "step": 1383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.142334658985259e-07,
+      "loss": 0.5309,
+      "step": 1384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.082928768138461e-07,
+      "loss": 0.8124,
+      "step": 1385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.023943056067792e-07,
+      "loss": 0.7166,
+      "step": 1386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9653777811687886e-07,
+      "loss": 0.7861,
+      "step": 1387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.907233199995231e-07,
+      "loss": 0.6898,
+      "step": 1388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8495095672579584e-07,
+      "loss": 0.8657,
+      "step": 1389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7922071358238e-07,
+      "loss": 0.7548,
+      "step": 1390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.735326156714436e-07,
+      "loss": 0.4765,
+      "step": 1391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6788668791053585e-07,
+      "loss": 0.6524,
+      "step": 1392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.622829550324658e-07,
+      "loss": 0.5831,
+      "step": 1393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5672144158520825e-07,
+      "loss": 0.7088,
+      "step": 1394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.512021719317882e-07,
+      "loss": 0.5414,
+      "step": 1395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4572517025017294e-07,
+      "loss": 0.543,
+      "step": 1396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.402904605331703e-07,
+      "loss": 0.4932,
+      "step": 1397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.348980665883217e-07,
+      "loss": 0.8195,
+      "step": 1398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.295480120378003e-07,
+      "loss": 0.6956,
+      "step": 1399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.242403203183031e-07,
+      "loss": 0.5288,
+      "step": 1400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1897501468094893e-07,
+      "loss": 0.6775,
+      "step": 1401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1375211819118176e-07,
+      "loss": 0.517,
+      "step": 1402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0857165372866536e-07,
+      "loss": 0.6431,
+      "step": 1403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.03433643987181e-07,
+      "loss": 0.7087,
+      "step": 1404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.983381114745332e-07,
+      "loss": 0.6239,
+      "step": 1405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9328507851244857e-07,
+      "loss": 0.7307,
+      "step": 1406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.882745672364784e-07,
+      "loss": 0.8221,
+      "step": 1407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8330659959589944e-07,
+      "loss": 0.8859,
+      "step": 1408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7838119735362433e-07,
+      "loss": 0.5674,
+      "step": 1409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7349838208609594e-07,
+      "loss": 0.968,
+      "step": 1410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.686581751832029e-07,
+      "loss": 0.5863,
+      "step": 1411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.6386059784818007e-07,
+      "loss": 0.586,
+      "step": 1412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5910567109751707e-07,
+      "loss": 0.4816,
+      "step": 1413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.543934157608685e-07,
+      "loss": 0.6908,
+      "step": 1414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.497238524809575e-07,
+      "loss": 0.4419,
+      "step": 1415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4509700171348906e-07,
+      "loss": 0.7137,
+      "step": 1416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4051288372706206e-07,
+      "loss": 0.5822,
+      "step": 1417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3597151860307644e-07,
+      "loss": 0.5835,
+      "step": 1418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3147292623564633e-07,
+      "loss": 0.5606,
+      "step": 1419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2701712633151573e-07,
+      "loss": 0.531,
+      "step": 1420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2260413840996532e-07,
+      "loss": 0.6814,
+      "step": 1421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1823398180273903e-07,
+      "loss": 0.6338,
+      "step": 1422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.139066756539454e-07,
+      "loss": 0.4642,
+      "step": 1423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0962223891998312e-07,
+      "loss": 0.46,
+      "step": 1424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.053806903694544e-07,
+      "loss": 0.7683,
+      "step": 1425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.011820485830851e-07,
+      "loss": 0.5843,
+      "step": 1426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9702633195363918e-07,
+      "loss": 0.7642,
+      "step": 1427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9291355868584217e-07,
+      "loss": 0.5366,
+      "step": 1428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.888437467962989e-07,
+      "loss": 0.7209,
+      "step": 1429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8481691411341817e-07,
+      "loss": 0.7166,
+      "step": 1430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.808330782773282e-07,
+      "loss": 0.6696,
+      "step": 1431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.768922567398068e-07,
+      "loss": 0.4763,
+      "step": 1432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7299446676419918e-07,
+      "loss": 0.8473,
+      "step": 1433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6913972542534463e-07,
+      "loss": 0.537,
+      "step": 1434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6532804960950334e-07,
+      "loss": 0.6276,
+      "step": 1435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6155945601427747e-07,
+      "loss": 0.6124,
+      "step": 1436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5783396114854464e-07,
+      "loss": 0.7245,
+      "step": 1437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5415158133238018e-07,
+      "loss": 0.666,
+      "step": 1438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5051233269698819e-07,
+      "loss": 0.6598,
+      "step": 1439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.469162311846295e-07,
+      "loss": 0.6151,
+      "step": 1440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4336329254855507e-07,
+      "loss": 0.624,
+      "step": 1441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.398535323529304e-07,
+      "loss": 0.6366,
+      "step": 1442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 0.7696,
+      "step": 1443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3296360859389345e-07,
+      "loss": 0.6682,
+      "step": 1444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2958347521280001e-07,
+      "loss": 0.8734,
+      "step": 1445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.262465806366664e-07,
+      "loss": 0.7482,
+      "step": 1446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2295293948324627e-07,
+      "loss": 0.5415,
+      "step": 1447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.197025661808171e-07,
+      "loss": 0.596,
+      "step": 1448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1649547496811686e-07,
+      "loss": 0.6838,
+      "step": 1449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1333167989427629e-07,
+      "loss": 0.548,
+      "step": 1450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1021119481876342e-07,
+      "loss": 0.7138,
+      "step": 1451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0713403341132244e-07,
+      "loss": 0.5097,
+      "step": 1452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.04100209151905e-07,
+      "loss": 0.6496,
+      "step": 1453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0110973533062674e-07,
+      "loss": 0.6377,
+      "step": 1454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.816262504769414e-08,
+      "loss": 0.7717,
+      "step": 1455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.525889121335563e-08,
+      "loss": 0.6304,
+      "step": 1456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.239854654784164e-08,
+      "loss": 0.4941,
+      "step": 1457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.95816035813124e-08,
+      "loss": 0.4916,
+      "step": 1458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.680807465379803e-08,
+      "loss": 0.5904,
+      "step": 1459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.407797191514855e-08,
+      "loss": 0.8097,
+      "step": 1460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.139130732497947e-08,
+      "loss": 0.6241,
+      "step": 1461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.87480926526163e-08,
+      "loss": 0.6928,
+      "step": 1462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.614833947705013e-08,
+      "loss": 0.6305,
+      "step": 1463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.359205918688105e-08,
+      "loss": 0.7322,
+      "step": 1464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.10792629802659e-08,
+      "loss": 0.6147,
+      "step": 1465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.860996186487945e-08,
+      "loss": 0.5974,
+      "step": 1466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.618416665785554e-08,
+      "loss": 0.5111,
+      "step": 1467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.380188798574493e-08,
+      "loss": 0.7318,
+      "step": 1468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.146313628446865e-08,
+      "loss": 0.7382,
+      "step": 1469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9167921799271336e-08,
+      "loss": 0.8323,
+      "step": 1470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.6916254584675755e-08,
+      "loss": 0.7254,
+      "step": 1471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.470814450444062e-08,
+      "loss": 0.5302,
+      "step": 1472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.2543601231513916e-08,
+      "loss": 0.4469,
+      "step": 1473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.0422634247996315e-08,
+      "loss": 0.6713,
+      "step": 1474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.8345252845093395e-08,
+      "loss": 0.5212,
+      "step": 1475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.631146612307791e-08,
+      "loss": 0.5234,
+      "step": 1476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4321282991250935e-08,
+      "loss": 0.7289,
+      "step": 1477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2374712167901896e-08,
+      "loss": 0.7531,
+      "step": 1478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0471762180268606e-08,
+      "loss": 0.6841,
+      "step": 1479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.861244136450171e-08,
+      "loss": 0.8871,
+      "step": 1480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.67967578656292e-08,
+      "loss": 0.7217,
+      "step": 1481
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5024719637517525e-08,
+      "loss": 0.6196,
+      "step": 1482
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3296334442839416e-08,
+      "loss": 0.58,
+      "step": 1483
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.161160985304168e-08,
+      "loss": 0.7772,
+      "step": 1484
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9970553248305224e-08,
+      "loss": 0.3969,
+      "step": 1485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.837317181751842e-08,
+      "loss": 0.5792,
+      "step": 1486
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.681947255824602e-08,
+      "loss": 0.4368,
+      "step": 1487
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.530946227669362e-08,
+      "loss": 0.6145,
+      "step": 1488
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3843147587683245e-08,
+      "loss": 0.5608,
+      "step": 1489
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.242053491462115e-08,
+      "loss": 0.632,
+      "step": 1490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1041630489470054e-08,
+      "loss": 0.562,
+      "step": 1491
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9706440352723623e-08,
+      "loss": 0.514,
+      "step": 1492
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.841497035337647e-08,
+      "loss": 0.6054,
+      "step": 1493
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7167226148901983e-08,
+      "loss": 0.6792,
+      "step": 1494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5963213205227868e-08,
+      "loss": 0.5515,
+      "step": 1495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4802936796708412e-08,
+      "loss": 0.6416,
+      "step": 1496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3686402006103384e-08,
+      "loss": 0.722,
+      "step": 1497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.261361372455916e-08,
+      "loss": 0.6085,
+      "step": 1498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.158457665157986e-08,
+      "loss": 0.4539,
+      "step": 1499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0599295295012913e-08,
+      "loss": 0.6393,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.657773971029071e-09,
+      "loss": 0.4972,
+      "step": 1501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.760016804099103e-09,
+      "loss": 0.7402,
+      "step": 1502
+    },
+    {
+      "epoch": 1.0,
+      "step": 1502,
+      "total_flos": 2.5089528249607782e+17,
+      "train_loss": 0.9972007662930914,
+      "train_runtime": 16518.4987,
+      "train_samples_per_second": 5.82,
+      "train_steps_per_second": 0.091
+    }
+  ],
+  "max_steps": 1502,
+  "num_train_epochs": 1,
+  "total_flos": 2.5089528249607782e+17,
+  "trial_name": null,
+  "trial_params": null
+}