diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10866 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1806,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6363636363636366e-07,
+      "loss": 1.307,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.272727272727273e-07,
+      "loss": 1.2963,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.090909090909091e-06,
+      "loss": 1.3499,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4545454545454546e-06,
+      "loss": 1.2905,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 1.2719,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.181818181818182e-06,
+      "loss": 1.3153,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5454545454545456e-06,
+      "loss": 1.2834,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9090909090909093e-06,
+      "loss": 1.3134,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.272727272727273e-06,
+      "loss": 1.3138,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 1.323,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.3056,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.363636363636364e-06,
+      "loss": 1.2911,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.727272727272728e-06,
+      "loss": 1.2967,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.090909090909091e-06,
+      "loss": 1.308,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.358,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8181818181818185e-06,
+      "loss": 1.243,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.181818181818182e-06,
+      "loss": 1.2471,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.545454545454546e-06,
+      "loss": 1.2547,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.90909090909091e-06,
+      "loss": 1.3318,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.219,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.636363636363638e-06,
+      "loss": 1.1466,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.2123,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.363636363636365e-06,
+      "loss": 1.2349,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.727272727272728e-06,
+      "loss": 1.1522,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.2801,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.454545454545456e-06,
+      "loss": 1.1549,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.81818181818182e-06,
+      "loss": 1.1285,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0181818181818182e-05,
+      "loss": 1.1428,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0545454545454546e-05,
+      "loss": 1.1223,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.1249,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1272727272727272e-05,
+      "loss": 1.1341,
+      "step": 31
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1636363636363637e-05,
+      "loss": 1.0909,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 1.0991,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2363636363636364e-05,
+      "loss": 1.1793,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 1.1216,
+      "step": 35
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3090909090909092e-05,
+      "loss": 1.0923,
+      "step": 36
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3454545454545455e-05,
+      "loss": 1.1394,
+      "step": 37
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.381818181818182e-05,
+      "loss": 1.0357,
+      "step": 38
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4181818181818183e-05,
+      "loss": 1.1069,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 1.0954,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4909090909090911e-05,
+      "loss": 1.0539,
+      "step": 41
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5272727272727276e-05,
+      "loss": 1.0363,
+      "step": 42
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.563636363636364e-05,
+      "loss": 0.9925,
+      "step": 43
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.0208,
+      "step": 44
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 1.0407,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.672727272727273e-05,
+      "loss": 0.9305,
+      "step": 46
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7090909090909092e-05,
+      "loss": 1.0081,
+      "step": 47
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7454545454545456e-05,
+      "loss": 1.0083,
+      "step": 48
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.781818181818182e-05,
+      "loss": 1.0231,
+      "step": 49
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.9806,
+      "step": 50
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8545454545454545e-05,
+      "loss": 0.967,
+      "step": 51
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8909090909090912e-05,
+      "loss": 0.9492,
+      "step": 52
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9272727272727275e-05,
+      "loss": 0.9849,
+      "step": 53
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963636363636364e-05,
+      "loss": 0.9775,
+      "step": 54
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.9225,
+      "step": 55
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99999839047643e-05,
+      "loss": 0.9521,
+      "step": 56
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999935619108992e-05,
+      "loss": 0.9116,
+      "step": 57
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999855143189522e-05,
+      "loss": 0.9194,
+      "step": 58
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999974247726494e-05,
+      "loss": 0.9458,
+      "step": 59
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999597621697927e-05,
+      "loss": 0.8837,
+      "step": 60
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999420576954783e-05,
+      "loss": 0.8958,
+      "step": 61
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999921134360541e-05,
+      "loss": 0.9124,
+      "step": 62
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998969922323356e-05,
+      "loss": 0.9324,
+      "step": 63
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998696313885756e-05,
+      "loss": 0.9056,
+      "step": 64
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998390519173376e-05,
+      "loss": 0.8721,
+      "step": 65
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998052539170575e-05,
+      "loss": 0.8764,
+      "step": 66
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997682374965332e-05,
+      "loss": 0.9135,
+      "step": 67
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997280027749227e-05,
+      "loss": 0.902,
+      "step": 68
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996845498817426e-05,
+      "loss": 0.9353,
+      "step": 69
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996378789568708e-05,
+      "loss": 0.9142,
+      "step": 70
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999587990150542e-05,
+      "loss": 0.8422,
+      "step": 71
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995348836233517e-05,
+      "loss": 0.8513,
+      "step": 72
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994785595462518e-05,
+      "loss": 0.8611,
+      "step": 73
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999419018100552e-05,
+      "loss": 0.8376,
+      "step": 74
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993562594779197e-05,
+      "loss": 0.8535,
+      "step": 75
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992902838803773e-05,
+      "loss": 0.8859,
+      "step": 76
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992210915203032e-05,
+      "loss": 0.7863,
+      "step": 77
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991486826204314e-05,
+      "loss": 0.8292,
+      "step": 78
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999073057413849e-05,
+      "loss": 0.8292,
+      "step": 79
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998994216143998e-05,
+      "loss": 0.8445,
+      "step": 80
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989121590646712e-05,
+      "loss": 0.7963,
+      "step": 81
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988268864400147e-05,
+      "loss": 0.8715,
+      "step": 82
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998738398544525e-05,
+      "loss": 0.7858,
+      "step": 83
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986466956630486e-05,
+      "loss": 0.8299,
+      "step": 84
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985517780907816e-05,
+      "loss": 0.7717,
+      "step": 85
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984536461332683e-05,
+      "loss": 0.7436,
+      "step": 86
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983523001064e-05,
+      "loss": 0.8783,
+      "step": 87
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998247740336414e-05,
+      "loss": 0.8235,
+      "step": 88
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998139967159894e-05,
+      "loss": 0.7895,
+      "step": 89
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980289809237662e-05,
+      "loss": 0.8448,
+      "step": 90
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979147819853003e-05,
+      "loss": 0.8832,
+      "step": 91
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977973707121085e-05,
+      "loss": 0.752,
+      "step": 92
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976767474821433e-05,
+      "loss": 0.8359,
+      "step": 93
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997552912683696e-05,
+      "loss": 0.8383,
+      "step": 94
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974258667153976e-05,
+      "loss": 0.8215,
+      "step": 95
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972956099862145e-05,
+      "loss": 0.7951,
+      "step": 96
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997162142915449e-05,
+      "loss": 0.7444,
+      "step": 97
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997025465932738e-05,
+      "loss": 0.7704,
+      "step": 98
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968855794780516e-05,
+      "loss": 0.7566,
+      "step": 99
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996742484001691e-05,
+      "loss": 0.833,
+      "step": 100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965961799642865e-05,
+      "loss": 0.7144,
+      "step": 101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996446667836798e-05,
+      "loss": 0.8733,
+      "step": 102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962939481005123e-05,
+      "loss": 0.7645,
+      "step": 103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961380212470412e-05,
+      "loss": 0.7736,
+      "step": 104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995978887778321e-05,
+      "loss": 0.7169,
+      "step": 105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958165482066094e-05,
+      "loss": 0.7252,
+      "step": 106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956510030544854e-05,
+      "loss": 0.7618,
+      "step": 107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954822528548462e-05,
+      "loss": 0.8237,
+      "step": 108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953102981509073e-05,
+      "loss": 0.8025,
+      "step": 109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951351394961986e-05,
+      "loss": 0.7553,
+      "step": 110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949567774545642e-05,
+      "loss": 0.8051,
+      "step": 111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947752126001597e-05,
+      "loss": 0.6928,
+      "step": 112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994590445517451e-05,
+      "loss": 0.7916,
+      "step": 113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944024768012122e-05,
+      "loss": 0.6856,
+      "step": 114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942113070565233e-05,
+      "loss": 0.7559,
+      "step": 115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940169368987692e-05,
+      "loss": 0.79,
+      "step": 116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993819366953636e-05,
+      "loss": 0.7362,
+      "step": 117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993618597857111e-05,
+      "loss": 0.7431,
+      "step": 118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993414630255479e-05,
+      "loss": 0.7016,
+      "step": 119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932074648053216e-05,
+      "loss": 0.7594,
+      "step": 120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992997102173514e-05,
+      "loss": 0.7028,
+      "step": 121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992783543037224e-05,
+      "loss": 0.7656,
+      "step": 122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992566788083908e-05,
+      "loss": 0.7491,
+      "step": 123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9923468380113105e-05,
+      "loss": 0.7979,
+      "step": 124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992123693527461e-05,
+      "loss": 0.7272,
+      "step": 125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918973553506723e-05,
+      "loss": 0.7282,
+      "step": 126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916678242095377e-05,
+      "loss": 0.7392,
+      "step": 127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991435100842929e-05,
+      "loss": 0.7251,
+      "step": 128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911991859999928e-05,
+      "loss": 0.7074,
+      "step": 129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990960080440151e-05,
+      "loss": 0.7102,
+      "step": 130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907177849330956e-05,
+      "loss": 0.7395,
+      "step": 131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904723002587868e-05,
+      "loss": 0.7332,
+      "step": 132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902236272074514e-05,
+      "loss": 0.7084,
+      "step": 133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98997176657958e-05,
+      "loss": 0.711,
+      "step": 134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897167191859234e-05,
+      "loss": 0.6799,
+      "step": 135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9894584858474916e-05,
+      "loss": 0.6942,
+      "step": 136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98919706739555e-05,
+      "loss": 0.6816,
+      "step": 137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889324646716166e-05,
+      "loss": 0.6934,
+      "step": 138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886646785274595e-05,
+      "loss": 0.7165,
+      "step": 139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883937098250962e-05,
+      "loss": 0.6565,
+      "step": 140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988119559436787e-05,
+      "loss": 0.7163,
+      "step": 141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9878422282450346e-05,
+      "loss": 0.6733,
+      "step": 142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875617171425817e-05,
+      "loss": 0.6946,
+      "step": 143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987278027032407e-05,
+      "loss": 0.6098,
+      "step": 144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869911588277214e-05,
+      "loss": 0.6896,
+      "step": 145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986701113451968e-05,
+      "loss": 0.7067,
+      "step": 146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986407891838816e-05,
+      "loss": 0.6481,
+      "step": 147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98611149493216e-05,
+      "loss": 0.6795,
+      "step": 148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858119236861157e-05,
+      "loss": 0.6542,
+      "step": 149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855091790650166e-05,
+      "loss": 0.6601,
+      "step": 150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852032620434124e-05,
+      "loss": 0.6915,
+      "step": 151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984894173606064e-05,
+      "loss": 0.7162,
+      "step": 152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845819147479418e-05,
+      "loss": 0.6626,
+      "step": 153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9842664864742218e-05,
+      "loss": 0.5869,
+      "step": 154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839478898002825e-05,
+      "loss": 0.6597,
+      "step": 155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836261257517015e-05,
+      "loss": 0.5653,
+      "step": 156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833011953642525e-05,
+      "loss": 0.7077,
+      "step": 157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982973099683902e-05,
+      "loss": 0.6241,
+      "step": 158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982641839766805e-05,
+      "loss": 0.6359,
+      "step": 159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9823074166793034e-05,
+      "loss": 0.7134,
+      "step": 160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98196983149792e-05,
+      "loss": 0.5997,
+      "step": 161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981629085309358e-05,
+      "loss": 0.6794,
+      "step": 162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812851792104955e-05,
+      "loss": 0.6781,
+      "step": 163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809381143083823e-05,
+      "loss": 0.5883,
+      "step": 164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805878917202366e-05,
+      "loss": 0.6178,
+      "step": 165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9802345125734416e-05,
+      "loss": 0.6787,
+      "step": 166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798779780055413e-05,
+      "loss": 0.6661,
+      "step": 167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979518289164237e-05,
+      "loss": 0.5847,
+      "step": 168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979155447207385e-05,
+      "loss": 0.5943,
+      "step": 169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787894533029893e-05,
+      "loss": 0.6429,
+      "step": 170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784203086292028e-05,
+      "loss": 0.6682,
+      "step": 171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.978048014374319e-05,
+      "loss": 0.6624,
+      "step": 172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776725717367705e-05,
+      "loss": 0.674,
+      "step": 173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772939819251247e-05,
+      "loss": 0.6564,
+      "step": 174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769122461580813e-05,
+      "loss": 0.5939,
+      "step": 175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765273656644645e-05,
+      "loss": 0.6076,
+      "step": 176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761393416832226e-05,
+      "loss": 0.5721,
+      "step": 177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9757481754634243e-05,
+      "loss": 0.657,
+      "step": 178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975353868264251e-05,
+      "loss": 0.5935,
+      "step": 179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749564213549965e-05,
+      "loss": 0.6866,
+      "step": 180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745558360150614e-05,
+      "loss": 0.6375,
+      "step": 181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741521135339485e-05,
+      "loss": 0.5605,
+      "step": 182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737452552112596e-05,
+      "loss": 0.5486,
+      "step": 183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9733352623566904e-05,
+      "loss": 0.6137,
+      "step": 184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972922136290028e-05,
+      "loss": 0.5719,
+      "step": 185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725058783411442e-05,
+      "loss": 0.5005,
+      "step": 186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972086489849993e-05,
+      "loss": 0.6094,
+      "step": 187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971663972166606e-05,
+      "loss": 0.5859,
+      "step": 188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971238326651087e-05,
+      "loss": 0.6104,
+      "step": 189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97080955467361e-05,
+      "loss": 0.5672,
+      "step": 190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703776576144106e-05,
+      "loss": 0.6238,
+      "step": 191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699426368637874e-05,
+      "loss": 0.5179,
+      "step": 192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695044938220917e-05,
+      "loss": 0.6333,
+      "step": 193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690632298997272e-05,
+      "loss": 0.5635,
+      "step": 194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686188465171425e-05,
+      "loss": 0.5856,
+      "step": 195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9681713451048298e-05,
+      "loss": 0.5562,
+      "step": 196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967720727103316e-05,
+      "loss": 0.6195,
+      "step": 197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672669939631627e-05,
+      "loss": 0.6174,
+      "step": 198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668101471449575e-05,
+      "loss": 0.5445,
+      "step": 199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966350188119312e-05,
+      "loss": 0.5902,
+      "step": 200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658871183668567e-05,
+      "loss": 0.5857,
+      "step": 201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965420939378234e-05,
+      "loss": 0.5205,
+      "step": 202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649516526540968e-05,
+      "loss": 0.5857,
+      "step": 203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964479259705101e-05,
+      "loss": 0.5321,
+      "step": 204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640037620519013e-05,
+      "loss": 0.5353,
+      "step": 205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635251612251475e-05,
+      "loss": 0.5382,
+      "step": 206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963043458765478e-05,
+      "loss": 0.6368,
+      "step": 207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962558656223516e-05,
+      "loss": 0.5432,
+      "step": 208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620707551598635e-05,
+      "loss": 0.5046,
+      "step": 209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615797571450976e-05,
+      "loss": 0.5129,
+      "step": 210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961085663759763e-05,
+      "loss": 0.5105,
+      "step": 211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605884765943703e-05,
+      "loss": 0.5967,
+      "step": 212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960088197249388e-05,
+      "loss": 0.5923,
+      "step": 213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9595848273352397e-05,
+      "loss": 0.4705,
+      "step": 214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959078368472296e-05,
+      "loss": 0.4703,
+      "step": 215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585688222908726e-05,
+      "loss": 0.5743,
+      "step": 216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958056190431222e-05,
+      "loss": 0.5344,
+      "step": 217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575404745435306e-05,
+      "loss": 0.5581,
+      "step": 218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957021676287912e-05,
+      "loss": 0.4674,
+      "step": 219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9564997973344027e-05,
+      "loss": 0.5176,
+      "step": 220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559748393629553e-05,
+      "loss": 0.4789,
+      "step": 221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554468040634345e-05,
+      "loss": 0.6029,
+      "step": 222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549156931356106e-05,
+      "loss": 0.5209,
+      "step": 223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543815082891547e-05,
+      "loss": 0.4526,
+      "step": 224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953844251243633e-05,
+      "loss": 0.5273,
+      "step": 225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9533039237285012e-05,
+      "loss": 0.5381,
+      "step": 226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527605274830993e-05,
+      "loss": 0.4832,
+      "step": 227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952214064256645e-05,
+      "loss": 0.5944,
+      "step": 228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95166453580823e-05,
+      "loss": 0.5275,
+      "step": 229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511119439068116e-05,
+      "loss": 0.4958,
+      "step": 230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950556290331209e-05,
+      "loss": 0.5333,
+      "step": 231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949997576870098e-05,
+      "loss": 0.5829,
+      "step": 232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949435805322003e-05,
+      "loss": 0.477,
+      "step": 233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488709774952937e-05,
+      "loss": 0.5472,
+      "step": 234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483030952081765e-05,
+      "loss": 0.5055,
+      "step": 235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947732160288692e-05,
+      "loss": 0.4217,
+      "step": 236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471581745747065e-05,
+      "loss": 0.5306,
+      "step": 237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9465811399139068e-05,
+      "loss": 0.4896,
+      "step": 238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460010581637952e-05,
+      "loss": 0.4566,
+      "step": 239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454179311916813e-05,
+      "loss": 0.5399,
+      "step": 240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448317608746795e-05,
+      "loss": 0.4576,
+      "step": 241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442425490996987e-05,
+      "loss": 0.4579,
+      "step": 242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94365029776344e-05,
+      "loss": 0.4215,
+      "step": 243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9430550087723882e-05,
+      "loss": 0.4458,
+      "step": 244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942456684042806e-05,
+      "loss": 0.3872,
+      "step": 245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94185532550073e-05,
+      "loss": 0.45,
+      "step": 246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941250935081961e-05,
+      "loss": 0.4575,
+      "step": 247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406435147320605e-05,
+      "loss": 0.4674,
+      "step": 248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940033066406343e-05,
+      "loss": 0.5271,
+      "step": 249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9394195920698706e-05,
+      "loss": 0.5166,
+      "step": 250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388030936974463e-05,
+      "loss": 0.4582,
+      "step": 251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938183573273607e-05,
+      "loss": 0.5296,
+      "step": 252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375610327926182e-05,
+      "loss": 0.4114,
+      "step": 253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369354742584676e-05,
+      "loss": 0.4864,
+      "step": 254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936306899684857e-05,
+      "loss": 0.4426,
+      "step": 255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935675311095198e-05,
+      "loss": 0.4426,
+      "step": 256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935040710522604e-05,
+      "loss": 0.4625,
+      "step": 257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344031000098836e-05,
+      "loss": 0.4793,
+      "step": 258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933762481609536e-05,
+      "loss": 0.4105,
+      "step": 259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933118857383741e-05,
+      "loss": 0.4372,
+      "step": 260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932472229404356e-05,
+      "loss": 0.418,
+      "step": 261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931822599752907e-05,
+      "loss": 0.5009,
+      "step": 262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931169970520582e-05,
+      "loss": 0.5068,
+      "step": 263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930514343808226e-05,
+      "loss": 0.3825,
+      "step": 264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298557217263315e-05,
+      "loss": 0.448,
+      "step": 265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9291941063950347e-05,
+      "loss": 0.4473,
+      "step": 266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928529499944106e-05,
+      "loss": 0.4189,
+      "step": 267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9278619045129452e-05,
+      "loss": 0.3857,
+      "step": 268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271913222505733e-05,
+      "loss": 0.4155,
+      "step": 269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265177553156262e-05,
+      "loss": 0.3763,
+      "step": 270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258412058763478e-05,
+      "loss": 0.4291,
+      "step": 271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251616761105826e-05,
+      "loss": 0.461,
+      "step": 272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924479168205769e-05,
+      "loss": 0.403,
+      "step": 273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9237936843589317e-05,
+      "loss": 0.3904,
+      "step": 274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231052267766764e-05,
+      "loss": 0.4042,
+      "step": 275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224137976751797e-05,
+      "loss": 0.4866,
+      "step": 276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217193992801843e-05,
+      "loss": 0.466,
+      "step": 277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921022033826992e-05,
+      "loss": 0.4287,
+      "step": 278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9203217035604555e-05,
+      "loss": 0.3841,
+      "step": 279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9196184107349697e-05,
+      "loss": 0.4108,
+      "step": 280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918912157614468e-05,
+      "loss": 0.3974,
+      "step": 281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918202946472412e-05,
+      "loss": 0.3679,
+      "step": 282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917490779591787e-05,
+      "loss": 0.3944,
+      "step": 283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167756592650907e-05,
+      "loss": 0.3984,
+      "step": 284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160575877943296e-05,
+      "loss": 0.4087,
+      "step": 285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9153365674910095e-05,
+      "loss": 0.3373,
+      "step": 286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146126006761282e-05,
+      "loss": 0.3646,
+      "step": 287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91388568968017e-05,
+      "loss": 0.3781,
+      "step": 288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913155836843095e-05,
+      "loss": 0.4065,
+      "step": 289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912423044514334e-05,
+      "loss": 0.4273,
+      "step": 290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116873150527807e-05,
+      "loss": 0.4134,
+      "step": 291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9109486508267818e-05,
+      "loss": 0.3579,
+      "step": 292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910207054214133e-05,
+      "loss": 0.3855,
+      "step": 293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094625276020685e-05,
+      "loss": 0.4026,
+      "step": 294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9087150733872546e-05,
+      "loss": 0.3387,
+      "step": 295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9079646939757815e-05,
+      "loss": 0.4121,
+      "step": 296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907211391783156e-05,
+      "loss": 0.377,
+      "step": 297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.906455169234294e-05,
+      "loss": 0.3995,
+      "step": 298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056960287635107e-05,
+      "loss": 0.4332,
+      "step": 299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049339728145152e-05,
+      "loss": 0.337,
+      "step": 300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904169003840402e-05,
+      "loss": 0.3516,
+      "step": 301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9034011243036414e-05,
+      "loss": 0.3646,
+      "step": 302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902630336676075e-05,
+      "loss": 0.3139,
+      "step": 303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018566434389037e-05,
+      "loss": 0.3375,
+      "step": 304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010800470826826e-05,
+      "loss": 0.3651,
+      "step": 305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900300550107312e-05,
+      "loss": 0.3036,
+      "step": 306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8995181550220296e-05,
+      "loss": 0.3663,
+      "step": 307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898732864345402e-05,
+      "loss": 0.3292,
+      "step": 308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979446806053167e-05,
+      "loss": 0.2403,
+      "step": 309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971536063389745e-05,
+      "loss": 0.3748,
+      "step": 310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8963596440928806e-05,
+      "loss": 0.2925,
+      "step": 311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895562796422837e-05,
+      "loss": 0.3518,
+      "step": 312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894763065893934e-05,
+      "loss": 0.3671,
+      "step": 313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939604550805422e-05,
+      "loss": 0.3642,
+      "step": 314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893154966566303e-05,
+      "loss": 0.3473,
+      "step": 315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892346602944122e-05,
+      "loss": 0.3278,
+      "step": 316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915353668161605e-05,
+      "loss": 0.3582,
+      "step": 317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8907212607938247e-05,
+      "loss": 0.3272,
+      "step": 318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899042874977613e-05,
+      "loss": 0.3371,
+      "step": 319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890844495578448e-05,
+      "loss": 0.2898,
+      "step": 320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882617496131735e-05,
+      "loss": 0.309,
+      "step": 321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874361903120562e-05,
+      "loss": 0.3561,
+      "step": 322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866077743120076e-05,
+      "loss": 0.4506,
+      "step": 323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857765042797382e-05,
+      "loss": 0.3266,
+      "step": 324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8849423828911446e-05,
+      "loss": 0.337,
+      "step": 325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841054128313036e-05,
+      "loss": 0.3064,
+      "step": 326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832655967944607e-05,
+      "loss": 0.2771,
+      "step": 327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824229374840242e-05,
+      "loss": 0.3243,
+      "step": 328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881577437612553e-05,
+      "loss": 0.3039,
+      "step": 329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807290999017524e-05,
+      "loss": 0.2995,
+      "step": 330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798779270824602e-05,
+      "loss": 0.2752,
+      "step": 331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790239218946424e-05,
+      "loss": 0.2605,
+      "step": 332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878167087087382e-05,
+      "loss": 0.337,
+      "step": 333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773074254188706e-05,
+      "loss": 0.357,
+      "step": 334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764449396563994e-05,
+      "loss": 0.2674,
+      "step": 335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755796325763508e-05,
+      "loss": 0.3347,
+      "step": 336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747115069641893e-05,
+      "loss": 0.3191,
+      "step": 337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738405656144522e-05,
+      "loss": 0.3398,
+      "step": 338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729668113307405e-05,
+      "loss": 0.2684,
+      "step": 339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720902469257108e-05,
+      "loss": 0.3881,
+      "step": 340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712108752210646e-05,
+      "loss": 0.3245,
+      "step": 341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703286990475415e-05,
+      "loss": 0.2938,
+      "step": 342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869443721244908e-05,
+      "loss": 0.2759,
+      "step": 343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868555944661949e-05,
+      "loss": 0.2784,
+      "step": 344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8676653721564598e-05,
+      "loss": 0.2511,
+      "step": 345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866772006595235e-05,
+      "loss": 0.3607,
+      "step": 346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86587585085406e-05,
+      "loss": 0.2397,
+      "step": 347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864976907817703e-05,
+      "loss": 0.2595,
+      "step": 348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640751803799036e-05,
+      "loss": 0.3346,
+      "step": 349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631706714433656e-05,
+      "loss": 0.2505,
+      "step": 350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622633839197456e-05,
+      "loss": 0.2566,
+      "step": 351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861353320729644e-05,
+      "loss": 0.3043,
+      "step": 352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8604404848025984e-05,
+      "loss": 0.2872,
+      "step": 353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85952487907707e-05,
+      "loss": 0.2072,
+      "step": 354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858606506500437e-05,
+      "loss": 0.3434,
+      "step": 355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857685370028984e-05,
+      "loss": 0.2177,
+      "step": 356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567614726278927e-05,
+      "loss": 0.2116,
+      "step": 357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558348172712318e-05,
+      "loss": 0.2949,
+      "step": 358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85490540694195e-05,
+      "loss": 0.2891,
+      "step": 359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8539732446318616e-05,
+      "loss": 0.2709,
+      "step": 360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853038333341642e-05,
+      "loss": 0.1872,
+      "step": 361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521006760808135e-05,
+      "loss": 0.2516,
+      "step": 362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511602758677402e-05,
+      "loss": 0.2396,
+      "step": 363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8502171357296144e-05,
+      "loss": 0.3084,
+      "step": 364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8492712587024484e-05,
+      "loss": 0.2571,
+      "step": 365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848322647831065e-05,
+      "loss": 0.306,
+      "step": 366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8473713061690874e-05,
+      "loss": 0.2589,
+      "step": 367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464172367789294e-05,
+      "loss": 0.3728,
+      "step": 368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845460442731785e-05,
+      "loss": 0.2436,
+      "step": 369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84450092710762e-05,
+      "loss": 0.2565,
+      "step": 370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8435386929951596e-05,
+      "loss": 0.3016,
+      "step": 371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8425737434918814e-05,
+      "loss": 0.2441,
+      "step": 372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416060817040028e-05,
+      "loss": 0.236,
+      "step": 373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406357107464733e-05,
+      "loss": 0.2247,
+      "step": 374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839662633742962e-05,
+      "loss": 0.314,
+      "step": 375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8386868538258505e-05,
+      "loss": 0.2581,
+      "step": 376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8377083741362196e-05,
+      "loss": 0.2889,
+      "step": 377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8367271978238422e-05,
+      "loss": 0.2481,
+      "step": 378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835743328047171e-05,
+      "loss": 0.2402,
+      "step": 379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347567679733285e-05,
+      "loss": 0.2891,
+      "step": 380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8337675207780987e-05,
+      "loss": 0.2907,
+      "step": 381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832775589645915e-05,
+      "loss": 0.2655,
+      "step": 382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83178097776985e-05,
+      "loss": 0.2035,
+      "step": 383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830783688351607e-05,
+      "loss": 0.2033,
+      "step": 384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829783724601507e-05,
+      "loss": 0.1965,
+      "step": 385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8287810897384803e-05,
+      "loss": 0.253,
+      "step": 386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277757869900562e-05,
+      "loss": 0.2376,
+      "step": 387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267678195923516e-05,
+      "loss": 0.2201,
+      "step": 388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257571907900616e-05,
+      "loss": 0.2441,
+      "step": 389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8247439038364468e-05,
+      "loss": 0.2334,
+      "step": 390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237279619933265e-05,
+      "loss": 0.2027,
+      "step": 391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822709368531065e-05,
+      "loss": 0.1871,
+      "step": 392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8216881267285624e-05,
+      "loss": 0.2573,
+      "step": 393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8206642398732453e-05,
+      "loss": 0.2155,
+      "step": 394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8196377112610524e-05,
+      "loss": 0.2102,
+      "step": 395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186085441964293e-05,
+      "loss": 0.2304,
+      "step": 396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175767419923115e-05,
+      "loss": 0.2567,
+      "step": 397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81654230797012e-05,
+      "loss": 0.2624,
+      "step": 398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155052454597467e-05,
+      "loss": 0.1704,
+      "step": 399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144655577995442e-05,
+      "loss": 0.249,
+      "step": 400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8134232483363167e-05,
+      "loss": 0.2617,
+      "step": 401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123783204253072e-05,
+      "loss": 0.2077,
+      "step": 402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113307774301874e-05,
+      "loss": 0.155,
+      "step": 403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102806227230485e-05,
+      "loss": 0.2104,
+      "step": 404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8092278596843873e-05,
+      "loss": 0.2316,
+      "step": 405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081724917030983e-05,
+      "loss": 0.1823,
+      "step": 406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80711452217646e-05,
+      "loss": 0.2049,
+      "step": 407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8060539545101268e-05,
+      "loss": 0.2057,
+      "step": 408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804990792118116e-05,
+      "loss": 0.1941,
+      "step": 409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8039250384227967e-05,
+      "loss": 0.1974,
+      "step": 410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802856696854881e-05,
+      "loss": 0.2349,
+      "step": 411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017857708534107e-05,
+      "loss": 0.2497,
+      "step": 412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800712263865747e-05,
+      "loss": 0.1265,
+      "step": 413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996361793475595e-05,
+      "loss": 0.189,
+      "step": 414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798557520762815e-05,
+      "loss": 0.2876,
+      "step": 415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974762915837666e-05,
+      "loss": 0.206,
+      "step": 416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7963924952909414e-05,
+      "loss": 0.1518,
+      "step": 417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7953061353731312e-05,
+      "loss": 0.1771,
+      "step": 418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7942172153273792e-05,
+      "loss": 0.2081,
+      "step": 419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931257386589714e-05,
+      "loss": 0.1266,
+      "step": 420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792031708881422e-05,
+      "loss": 0.1977,
+      "step": 421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790935129516464e-05,
+      "loss": 0.1679,
+      "step": 422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789836004094039e-05,
+      "loss": 0.1254,
+      "step": 423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887343361522832e-05,
+      "loss": 0.1989,
+      "step": 424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7876301292375173e-05,
+      "loss": 0.239,
+      "step": 425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7865233869042356e-05,
+      "loss": 0.1442,
+      "step": 426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7854141127150946e-05,
+      "loss": 0.1739,
+      "step": 427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784302310240899e-05,
+      "loss": 0.1824,
+      "step": 428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783187983060594e-05,
+      "loss": 0.2323,
+      "step": 429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782071134761251e-05,
+      "loss": 0.19,
+      "step": 430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7809517689380572e-05,
+      "loss": 0.2192,
+      "step": 431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798298891943047e-05,
+      "loss": 0.1639,
+      "step": 432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7787054991413774e-05,
+      "loss": 0.1581,
+      "step": 433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7775786023987385e-05,
+      "loss": 0.2352,
+      "step": 434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7764492025939233e-05,
+      "loss": 0.1805,
+      "step": 435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775317303362522e-05,
+      "loss": 0.1865,
+      "step": 436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.774182908348172e-05,
+      "loss": 0.1757,
+      "step": 437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773046021202544e-05,
+      "loss": 0.1945,
+      "step": 438
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719066455853317e-05,
+      "loss": 0.1753,
+      "step": 439
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7707647851642387e-05,
+      "loss": 0.1565,
+      "step": 440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7696204436149678e-05,
+      "loss": 0.1988,
+      "step": 441
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768473624621208e-05,
+      "loss": 0.1419,
+      "step": 442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767324331874624e-05,
+      "loss": 0.1504,
+      "step": 443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766172569074843e-05,
+      "loss": 0.193,
+      "step": 444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650183399294438e-05,
+      "loss": 0.1724,
+      "step": 445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763861648153945e-05,
+      "loss": 0.1638,
+      "step": 446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627024974717914e-05,
+      "loss": 0.2278,
+      "step": 447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761540891614344e-05,
+      "loss": 0.2091,
+      "step": 448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7603768343208665e-05,
+      "loss": 0.1664,
+      "step": 449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7592103293385144e-05,
+      "loss": 0.2408,
+      "step": 450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758041380422322e-05,
+      "loss": 0.1552,
+      "step": 451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7568699913351914e-05,
+      "loss": 0.1037,
+      "step": 452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755696165847879e-05,
+      "loss": 0.1589,
+      "step": 453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7545199077389843e-05,
+      "loss": 0.1488,
+      "step": 454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7533412207949377e-05,
+      "loss": 0.1913,
+      "step": 455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752160108809988e-05,
+      "loss": 0.1761,
+      "step": 456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7509765755861906e-05,
+      "loss": 0.1143,
+      "step": 457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7497906249333943e-05,
+      "loss": 0.2004,
+      "step": 458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7486022606692307e-05,
+      "loss": 0.1966,
+      "step": 459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474114866191004e-05,
+      "loss": 0.1368,
+      "step": 460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7462183066161605e-05,
+      "loss": 0.1773,
+      "step": 461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450227245013138e-05,
+      "loss": 0.1703,
+      "step": 462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743824744123196e-05,
+      "loss": 0.1205,
+      "step": 463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7426243693381625e-05,
+      "loss": 0.139,
+      "step": 464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7414216040102755e-05,
+      "loss": 0.1624,
+      "step": 465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7402164520112943e-05,
+      "loss": 0.2369,
+      "step": 466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7390089172206594e-05,
+      "loss": 0.1661,
+      "step": 467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737799003525482e-05,
+      "loss": 0.1533,
+      "step": 468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365867148205316e-05,
+      "loss": 0.186,
+      "step": 469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735372055008223e-05,
+      "loss": 0.1641,
+      "step": 470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734155027998603e-05,
+      "loss": 0.1588,
+      "step": 471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7329356377093393e-05,
+      "loss": 0.1747,
+      "step": 472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731713888065706e-05,
+      "loss": 0.1963,
+      "step": 473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304897830005734e-05,
+      "loss": 0.2202,
+      "step": 474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7292633264543936e-05,
+      "loss": 0.0796,
+      "step": 475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7280345223751872e-05,
+      "loss": 0.0653,
+      "step": 476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7268033747185332e-05,
+      "loss": 0.1213,
+      "step": 477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7255698874475537e-05,
+      "loss": 0.1445,
+      "step": 478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7243340645329023e-05,
+      "loss": 0.1856,
+      "step": 479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230959099527512e-05,
+      "loss": 0.1491,
+      "step": 480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721855427692779e-05,
+      "loss": 0.1375,
+      "step": 481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206126217461553e-05,
+      "loss": 0.146,
+      "step": 482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7193674961135322e-05,
+      "loss": 0.1077,
+      "step": 483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181200548030274e-05,
+      "loss": 0.1659,
+      "step": 484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7168703018302128e-05,
+      "loss": 0.1624,
+      "step": 485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715618241218103e-05,
+      "loss": 0.0885,
+      "step": 486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7143638769971394e-05,
+      "loss": 0.1332,
+      "step": 487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131072132051802e-05,
+      "loss": 0.1637,
+      "step": 488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711848253887485e-05,
+      "loss": 0.1724,
+      "step": 489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710587003096703e-05,
+      "loss": 0.1332,
+      "step": 490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7093234648928608e-05,
+      "loss": 0.1396,
+      "step": 491
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080576433433462e-05,
+      "loss": 0.1815,
+      "step": 492
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7067895425228994e-05,
+      "loss": 0.1539,
+      "step": 493
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705519166513596e-05,
+      "loss": 0.1148,
+      "step": 494
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042465194048373e-05,
+      "loss": 0.1497,
+      "step": 495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029716052933336e-05,
+      "loss": 0.1266,
+      "step": 496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7016944282830935e-05,
+      "loss": 0.1111,
+      "step": 497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004149924854102e-05,
+      "loss": 0.1485,
+      "step": 498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991333020188473e-05,
+      "loss": 0.1352,
+      "step": 499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697849361009228e-05,
+      "loss": 0.1454,
+      "step": 500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696563173589618e-05,
+      "loss": 0.1079,
+      "step": 501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6952747439003153e-05,
+      "loss": 0.2306,
+      "step": 502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693984076088836e-05,
+      "loss": 0.13,
+      "step": 503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6926911743099012e-05,
+      "loss": 0.1471,
+      "step": 504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6913960427254218e-05,
+      "loss": 0.1614,
+      "step": 505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6900986855044874e-05,
+      "loss": 0.1573,
+      "step": 506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6887991068233525e-05,
+      "loss": 0.0911,
+      "step": 507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687497310865422e-05,
+      "loss": 0.108,
+      "step": 508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6861933018212386e-05,
+      "loss": 0.1965,
+      "step": 509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684887083888469e-05,
+      "loss": 0.1677,
+      "step": 510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6835786612718895e-05,
+      "loss": 0.1011,
+      "step": 511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822680381833755e-05,
+      "loss": 0.1205,
+      "step": 512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6809552188418828e-05,
+      "loss": 0.176,
+      "step": 513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796402074734404e-05,
+      "loss": 0.0933,
+      "step": 514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6783230083111314e-05,
+      "loss": 0.1288,
+      "step": 515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677003625595081e-05,
+      "loss": 0.1063,
+      "step": 516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6756820635724454e-05,
+      "loss": 0.1351,
+      "step": 517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6743583264973953e-05,
+      "loss": 0.1547,
+      "step": 518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6730324186311016e-05,
+      "loss": 0.1117,
+      "step": 519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6717043442417252e-05,
+      "loss": 0.1323,
+      "step": 520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6703741076044e-05,
+      "loss": 0.1162,
+      "step": 521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66904171300122e-05,
+      "loss": 0.1342,
+      "step": 522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667707164721226e-05,
+      "loss": 0.1176,
+      "step": 523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666370467060393e-05,
+      "loss": 0.1667,
+      "step": 524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665031624321613e-05,
+      "loss": 0.0914,
+      "step": 525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663690640814684e-05,
+      "loss": 0.0934,
+      "step": 526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6623475208562953e-05,
+      "loss": 0.12,
+      "step": 527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610022687700133e-05,
+      "loss": 0.1543,
+      "step": 528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6596548888862674e-05,
+      "loss": 0.1208,
+      "step": 529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583053855423373e-05,
+      "loss": 0.1034,
+      "step": 530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6569537630823385e-05,
+      "loss": 0.061,
+      "step": 531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6556000258572066e-05,
+      "loss": 0.0806,
+      "step": 532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6542441782246855e-05,
+      "loss": 0.0517,
+      "step": 533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652886224549313e-05,
+      "loss": 0.1229,
+      "step": 534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515261692024062e-05,
+      "loss": 0.0997,
+      "step": 535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6501640165620468e-05,
+      "loss": 0.1831,
+      "step": 536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6487997710130685e-05,
+      "loss": 0.0731,
+      "step": 537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6474334369470427e-05,
+      "loss": 0.1025,
+      "step": 538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6460650187622626e-05,
+      "loss": 0.1359,
+      "step": 539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6446945208637306e-05,
+      "loss": 0.109,
+      "step": 540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433219476631446e-05,
+      "loss": 0.1164,
+      "step": 541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6419473035788823e-05,
+      "loss": 0.0986,
+      "step": 542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6405705930359873e-05,
+      "loss": 0.0896,
+      "step": 543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639191820466156e-05,
+      "loss": 0.2367,
+      "step": 544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6378109903077234e-05,
+      "loss": 0.1639,
+      "step": 545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636428107005645e-05,
+      "loss": 0.0975,
+      "step": 546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6350431750114885e-05,
+      "loss": 0.0847,
+      "step": 547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6336561987834155e-05,
+      "loss": 0.1108,
+      "step": 548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632267182786167e-05,
+      "loss": 0.0945,
+      "step": 549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6308761314910514e-05,
+      "loss": 0.1173,
+      "step": 550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6294830493759284e-05,
+      "loss": 0.1921,
+      "step": 551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6280879409251945e-05,
+      "loss": 0.1218,
+      "step": 552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266908106297705e-05,
+      "loss": 0.1234,
+      "step": 553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6252916629870842e-05,
+      "loss": 0.1122,
+      "step": 554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623890502501058e-05,
+      "loss": 0.0982,
+      "step": 555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622487333682093e-05,
+      "loss": 0.1936,
+      "step": 556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6210821610470565e-05,
+      "loss": 0.0946,
+      "step": 557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619674989119265e-05,
+      "loss": 0.0843,
+      "step": 558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618265822428471e-05,
+      "loss": 0.112,
+      "step": 559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168546655108494e-05,
+      "loss": 0.1334,
+      "step": 560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154415229089805e-05,
+      "loss": 0.1199,
+      "step": 561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6140263991718363e-05,
+      "loss": 0.0783,
+      "step": 562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6126092988547673e-05,
+      "loss": 0.0681,
+      "step": 563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6111902265194865e-05,
+      "loss": 0.0718,
+      "step": 564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097691867340547e-05,
+      "loss": 0.1929,
+      "step": 565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608346184072865e-05,
+      "loss": 0.1764,
+      "step": 566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6069212231166308e-05,
+      "loss": 0.0755,
+      "step": 567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605494308452369e-05,
+      "loss": 0.1433,
+      "step": 568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6040654446733838e-05,
+      "loss": 0.128,
+      "step": 569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 0.1075,
+      "step": 570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012018881758257e-05,
+      "loss": 0.1227,
+      "step": 571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5997672046751758e-05,
+      "loss": 0.0724,
+      "step": 572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5983305904956198e-05,
+      "loss": 0.0615,
+      "step": 573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596892050261688e-05,
+      "loss": 0.0485,
+      "step": 574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5954515886041077e-05,
+      "loss": 0.1874,
+      "step": 575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5940092101597934e-05,
+      "loss": 0.0855,
+      "step": 576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5925649195718292e-05,
+      "loss": 0.0557,
+      "step": 577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591118721489455e-05,
+      "loss": 0.0839,
+      "step": 578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5896706205680504e-05,
+      "loss": 0.1621,
+      "step": 579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588220621469121e-05,
+      "loss": 0.1425,
+      "step": 580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.586768728860281e-05,
+      "loss": 0.0887,
+      "step": 581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585314947415242e-05,
+      "loss": 0.0988,
+      "step": 582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5838592818137952e-05,
+      "loss": 0.065,
+      "step": 583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5824017367417965e-05,
+      "loss": 0.103,
+      "step": 584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580942316891152e-05,
+      "loss": 0.1096,
+      "step": 585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5794810269598035e-05,
+      "loss": 0.0824,
+      "step": 586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5780178716517115e-05,
+      "loss": 0.0713,
+      "step": 587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5765528556768424e-05,
+      "loss": 0.1284,
+      "step": 588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5750859837511514e-05,
+      "loss": 0.0955,
+      "step": 589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5736172605965687e-05,
+      "loss": 0.1243,
+      "step": 590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572146690940983e-05,
+      "loss": 0.159,
+      "step": 591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570674279518228e-05,
+      "loss": 0.1123,
+      "step": 592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5692000310680645e-05,
+      "loss": 0.0814,
+      "step": 593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567723950336169e-05,
+      "loss": 0.1061,
+      "step": 594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662460420741138e-05,
+      "loss": 0.0884,
+      "step": 595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5647663110393558e-05,
+      "loss": 0.0749,
+      "step": 596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563284761995219e-05,
+      "loss": 0.0887,
+      "step": 597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5618013997108796e-05,
+      "loss": 0.0762,
+      "step": 598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5603162289613503e-05,
+      "loss": 0.0895,
+      "step": 599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558829254527466e-05,
+      "loss": 0.0578,
+      "step": 600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5573404811958678e-05,
+      "loss": 0.0702,
+      "step": 601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555849913758987e-05,
+      "loss": 0.1412,
+      "step": 602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55435755701503e-05,
+      "loss": 0.096,
+      "step": 603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5528634157679647e-05,
+      "loss": 0.0984,
+      "step": 604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5513674948275013e-05,
+      "loss": 0.0391,
+      "step": 605
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5498697990090798e-05,
+      "loss": 0.0614,
+      "step": 606
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5483703331338537e-05,
+      "loss": 0.094,
+      "step": 607
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546869102028675e-05,
+      "loss": 0.135,
+      "step": 608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5453661105260765e-05,
+      "loss": 0.0838,
+      "step": 609
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543861363464259e-05,
+      "loss": 0.0459,
+      "step": 610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5423548656870747e-05,
+      "loss": 0.0734,
+      "step": 611
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408466220440105e-05,
+      "loss": 0.0844,
+      "step": 612
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5393366373901733e-05,
+      "loss": 0.0412,
+      "step": 613
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378249165862758e-05,
+      "loss": 0.0783,
+      "step": 614
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5363114644986184e-05,
+      "loss": 0.0582,
+      "step": 615
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5347962859990744e-05,
+      "loss": 0.0567,
+      "step": 616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5332793859650747e-05,
+      "loss": 0.1037,
+      "step": 617
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5317607692795925e-05,
+      "loss": 0.0977,
+      "step": 618
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530240440831126e-05,
+      "loss": 0.0517,
+      "step": 619
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5287184055136845e-05,
+      "loss": 0.0484,
+      "step": 620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527194668226771e-05,
+      "loss": 0.0638,
+      "step": 621
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256692338753682e-05,
+      "loss": 0.0421,
+      "step": 622
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.524142107369921e-05,
+      "loss": 0.0659,
+      "step": 623
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5226132936263219e-05,
+      "loss": 0.0598,
+      "step": 624
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210827975658935e-05,
+      "loss": 0.138,
+      "step": 625
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5195506241153755e-05,
+      "loss": 0.0648,
+      "step": 626
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5180167782069065e-05,
+      "loss": 0.0479,
+      "step": 627
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164812647780083e-05,
+      "loss": 0.0634,
+      "step": 628
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5149440887715713e-05,
+      "loss": 0.1294,
+      "step": 629
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5134052551358379e-05,
+      "loss": 0.1038,
+      "step": 630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118647688243853e-05,
+      "loss": 0.0868,
+      "step": 631
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103226347961124e-05,
+      "loss": 0.0437,
+      "step": 632
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5087788580152207e-05,
+      "loss": 0.0322,
+      "step": 633
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5072334434512008e-05,
+      "loss": 0.0296,
+      "step": 634
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056863960788148e-05,
+      "loss": 0.0392,
+      "step": 635
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5041377208780814e-05,
+      "loss": 0.056,
+      "step": 636
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5025874228342585e-05,
+      "loss": 0.0668,
+      "step": 637
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5010355069378295e-05,
+      "loss": 0.0538,
+      "step": 638
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499481978184484e-05,
+      "loss": 0.0698,
+      "step": 639
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497926841575105e-05,
+      "loss": 0.1023,
+      "step": 640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4963701021157497e-05,
+      "loss": 0.0467,
+      "step": 641
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948117648176367e-05,
+      "loss": 0.0339,
+      "step": 642
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4932518346971264e-05,
+      "loss": 0.0446,
+      "step": 643
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4916903167757082e-05,
+      "loss": 0.0385,
+      "step": 644
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490127216079981e-05,
+      "loss": 0.0476,
+      "step": 645
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4885625376416406e-05,
+      "loss": 0.0266,
+      "step": 646
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4869962864974599e-05,
+      "loss": 0.0615,
+      "step": 647
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4854284676892757e-05,
+      "loss": 0.0626,
+      "step": 648
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838590862639702e-05,
+      "loss": 0.1254,
+      "step": 649
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4822881472734563e-05,
+      "loss": 0.0438,
+      "step": 650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4807156557746609e-05,
+      "loss": 0.0767,
+      "step": 651
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4791416168295081e-05,
+      "loss": 0.0412,
+      "step": 652
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4775660355049032e-05,
+      "loss": 0.0781,
+      "step": 653
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759889168727175e-05,
+      "loss": 0.0289,
+      "step": 654
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744102660097696e-05,
+      "loss": 0.0625,
+      "step": 655
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472830087997811e-05,
+      "loss": 0.1597,
+      "step": 656
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4712483879235097e-05,
+      "loss": 0.1201,
+      "step": 657
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4696651708784326e-05,
+      "loss": 0.0528,
+      "step": 658
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4680804419590297e-05,
+      "loss": 0.0587,
+      "step": 659
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4664942062666183e-05,
+      "loss": 0.0232,
+      "step": 660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4649064689073659e-05,
+      "loss": 0.0653,
+      "step": 661
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4633172349922742e-05,
+      "loss": 0.0589,
+      "step": 662
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4617265096371616e-05,
+      "loss": 0.099,
+      "step": 663
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4601342979626482e-05,
+      "loss": 0.0401,
+      "step": 664
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4585406050941384e-05,
+      "loss": 0.0226,
+      "step": 665
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456945436161805e-05,
+      "loss": 0.0608,
+      "step": 666
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4553487963005712e-05,
+      "loss": 0.0541,
+      "step": 667
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453750690650097e-05,
+      "loss": 0.1386,
+      "step": 668
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4521511243547588e-05,
+      "loss": 0.0794,
+      "step": 669
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4505501025636367e-05,
+      "loss": 0.0762,
+      "step": 670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448947630430495e-05,
+      "loss": 0.0773,
+      "step": 671
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.447343713113767e-05,
+      "loss": 0.086,
+      "step": 672
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4457383557765385e-05,
+      "loss": 0.0553,
+      "step": 673
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4441315635865299e-05,
+      "loss": 0.0641,
+      "step": 674
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4425233417160813e-05,
+      "loss": 0.0432,
+      "step": 675
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440913695342135e-05,
+      "loss": 0.0427,
+      "step": 676
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439302629646218e-05,
+      "loss": 0.0483,
+      "step": 677
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437690149814427e-05,
+      "loss": 0.0609,
+      "step": 678
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360762610374104e-05,
+      "loss": 0.0684,
+      "step": 679
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4344609685103525e-05,
+      "loss": 0.0439,
+      "step": 680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4328442774329565e-05,
+      "loss": 0.055,
+      "step": 681
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4312261930094262e-05,
+      "loss": 0.0529,
+      "step": 682
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4296067204484523e-05,
+      "loss": 0.054,
+      "step": 683
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427985864963193e-05,
+      "loss": 0.0613,
+      "step": 684
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426363631771259e-05,
+      "loss": 0.0486,
+      "step": 685
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4247400260946948e-05,
+      "loss": 0.0761,
+      "step": 686
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423115053159964e-05,
+      "loss": 0.0584,
+      "step": 687
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421488718197931e-05,
+      "loss": 0.0593,
+      "step": 688
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4198610264438449e-05,
+      "loss": 0.0558,
+      "step": 689
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4182319831373214e-05,
+      "loss": 0.0374,
+      "step": 690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416601593522329e-05,
+      "loss": 0.0429,
+      "step": 691
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4149698628471676e-05,
+      "loss": 0.0636,
+      "step": 692
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4133367963644555e-05,
+      "loss": 0.0836,
+      "step": 693
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4117023993311109e-05,
+      "loss": 0.0446,
+      "step": 694
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410066677008335e-05,
+      "loss": 0.0425,
+      "step": 695
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4084296346615947e-05,
+      "loss": 0.0499,
+      "step": 696
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067912775606063e-05,
+      "loss": 0.0407,
+      "step": 697
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4051516109793193e-05,
+      "loss": 0.0543,
+      "step": 698
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403510640195897e-05,
+      "loss": 0.0369,
+      "step": 699
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018683704927018e-05,
+      "loss": 0.0312,
+      "step": 700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.400224807156278e-05,
+      "loss": 0.066,
+      "step": 701
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985799554773327e-05,
+      "loss": 0.0169,
+      "step": 702
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3969338207507213e-05,
+      "loss": 0.0773,
+      "step": 703
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3952864082754289e-05,
+      "loss": 0.0823,
+      "step": 704
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3936377233545542e-05,
+      "loss": 0.0588,
+      "step": 705
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3919877712952915e-05,
+      "loss": 0.0247,
+      "step": 706
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3903365574089142e-05,
+      "loss": 0.0684,
+      "step": 707
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3886840870107577e-05,
+      "loss": 0.1121,
+      "step": 708
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870303654202022e-05,
+      "loss": 0.0558,
+      "step": 709
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3853753979606552e-05,
+      "loss": 0.1188,
+      "step": 710
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3837191899595356e-05,
+      "loss": 0.0387,
+      "step": 711
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3820617467482541e-05,
+      "loss": 0.0283,
+      "step": 712
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3804030736621993e-05,
+      "loss": 0.0633,
+      "step": 713
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3787431760407177e-05,
+      "loss": 0.0313,
+      "step": 714
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3770820592270982e-05,
+      "loss": 0.0372,
+      "step": 715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754197285685537e-05,
+      "loss": 0.1088,
+      "step": 716
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3737561894162054e-05,
+      "loss": 0.0321,
+      "step": 717
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3720914471250644e-05,
+      "loss": 0.0292,
+      "step": 718
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370425507054014e-05,
+      "loss": 0.0283,
+      "step": 719
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3687583745657941e-05,
+      "loss": 0.059,
+      "step": 720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367090055026983e-05,
+      "loss": 0.0572,
+      "step": 721
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3654205538079796e-05,
+      "loss": 0.0248,
+      "step": 722
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3637498762829872e-05,
+      "loss": 0.0357,
+      "step": 723
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3620780278299956e-05,
+      "loss": 0.0702,
+      "step": 724
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604050138307638e-05,
+      "loss": 0.0918,
+      "step": 725
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3587308396708023e-05,
+      "loss": 0.0371,
+      "step": 726
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.357055510739357e-05,
+      "loss": 0.0976,
+      "step": 727
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355379032429391e-05,
+      "loss": 0.048,
+      "step": 728
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537014101375662e-05,
+      "loss": 0.0383,
+      "step": 729
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3520226492642288e-05,
+      "loss": 0.0501,
+      "step": 730
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3503427552133884e-05,
+      "loss": 0.0882,
+      "step": 731
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3486617333927033e-05,
+      "loss": 0.1091,
+      "step": 732
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3469795892134623e-05,
+      "loss": 0.0463,
+      "step": 733
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3452963280905667e-05,
+      "loss": 0.0505,
+      "step": 734
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3436119554425133e-05,
+      "loss": 0.0243,
+      "step": 735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3419264766913768e-05,
+      "loss": 0.0263,
+      "step": 736
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402398972627935e-05,
+      "loss": 0.0175,
+      "step": 737
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3385522225859415e-05,
+      "loss": 0.0497,
+      "step": 738
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3368634580935252e-05,
+      "loss": 0.0728,
+      "step": 739
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3351736092217573e-05,
+      "loss": 0.0457,
+      "step": 740
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3334826814103407e-05,
+      "loss": 0.035,
+      "step": 741
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3317906801024523e-05,
+      "loss": 0.0841,
+      "step": 742
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3300976107447232e-05,
+      "loss": 0.0446,
+      "step": 743
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3284034787872243e-05,
+      "loss": 0.0543,
+      "step": 744
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3267082896834457e-05,
+      "loss": 0.0539,
+      "step": 745
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250120488902813e-05,
+      "loss": 0.071,
+      "step": 746
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32331476186801e-05,
+      "loss": 0.0376,
+      "step": 747
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.321616434080279e-05,
+      "loss": 0.0631,
+      "step": 748
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3199170709940855e-05,
+      "loss": 0.0479,
+      "step": 749
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3182166780797589e-05,
+      "loss": 0.0222,
+      "step": 750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3165152608109447e-05,
+      "loss": 0.0511,
+      "step": 751
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314812824664585e-05,
+      "loss": 0.0361,
+      "step": 752
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3131093751209023e-05,
+      "loss": 0.0429,
+      "step": 753
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3114049176633806e-05,
+      "loss": 0.0549,
+      "step": 754
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309699457778749e-05,
+      "loss": 0.0461,
+      "step": 755
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3079930009569633e-05,
+      "loss": 0.1184,
+      "step": 756
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3062855526911882e-05,
+      "loss": 0.0152,
+      "step": 757
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3045771184777807e-05,
+      "loss": 0.0318,
+      "step": 758
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3028677038162705e-05,
+      "loss": 0.077,
+      "step": 759
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3011573142093442e-05,
+      "loss": 0.0575,
+      "step": 760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2994459551628264e-05,
+      "loss": 0.0206,
+      "step": 761
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2977336321856633e-05,
+      "loss": 0.081,
+      "step": 762
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2960203507899022e-05,
+      "loss": 0.0708,
+      "step": 763
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943061164906774e-05,
+      "loss": 0.0272,
+      "step": 764
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2925909348061898e-05,
+      "loss": 0.0485,
+      "step": 765
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2908748112576901e-05,
+      "loss": 0.0228,
+      "step": 766
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2891577513694604e-05,
+      "loss": 0.0475,
+      "step": 767
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2874397606687982e-05,
+      "loss": 0.0461,
+      "step": 768
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285720844685996e-05,
+      "loss": 0.079,
+      "step": 769
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2840010089543256e-05,
+      "loss": 0.0895,
+      "step": 770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2822802590100191e-05,
+      "loss": 0.0914,
+      "step": 771
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2805586003922523e-05,
+      "loss": 0.0895,
+      "step": 772
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2788360386431249e-05,
+      "loss": 0.0287,
+      "step": 773
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2771125793076442e-05,
+      "loss": 0.074,
+      "step": 774
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2753882279337076e-05,
+      "loss": 0.0592,
+      "step": 775
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2736629900720832e-05,
+      "loss": 0.0094,
+      "step": 776
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2719368712763927e-05,
+      "loss": 0.0537,
+      "step": 777
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2702098771030944e-05,
+      "loss": 0.034,
+      "step": 778
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2684820131114636e-05,
+      "loss": 0.0418,
+      "step": 779
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.266753284863576e-05,
+      "loss": 0.0854,
+      "step": 780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2650236979242896e-05,
+      "loss": 0.0716,
+      "step": 781
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263293257861226e-05,
+      "loss": 0.0404,
+      "step": 782
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2615619702447535e-05,
+      "loss": 0.0599,
+      "step": 783
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2598298406479684e-05,
+      "loss": 0.0617,
+      "step": 784
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2580968746466775e-05,
+      "loss": 0.03,
+      "step": 785
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2563630778193805e-05,
+      "loss": 0.0755,
+      "step": 786
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2546284557472505e-05,
+      "loss": 0.0644,
+      "step": 787
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252893014014118e-05,
+      "loss": 0.0227,
+      "step": 788
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2511567582064516e-05,
+      "loss": 0.0598,
+      "step": 789
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249419693913341e-05,
+      "loss": 0.0617,
+      "step": 790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2476818267264778e-05,
+      "loss": 0.0489,
+      "step": 791
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2459431622401383e-05,
+      "loss": 0.059,
+      "step": 792
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2442037060511654e-05,
+      "loss": 0.0754,
+      "step": 793
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2424634637589508e-05,
+      "loss": 0.0676,
+      "step": 794
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2407224409654166e-05,
+      "loss": 0.0433,
+      "step": 795
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2389806432749967e-05,
+      "loss": 0.0597,
+      "step": 796
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2372380762946203e-05,
+      "loss": 0.0278,
+      "step": 797
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2354947456336928e-05,
+      "loss": 0.0813,
+      "step": 798
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2337506569040777e-05,
+      "loss": 0.031,
+      "step": 799
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2320058157200785e-05,
+      "loss": 0.009,
+      "step": 800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2302602276984218e-05,
+      "loss": 0.0634,
+      "step": 801
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2285138984582373e-05,
+      "loss": 0.0965,
+      "step": 802
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2267668336210411e-05,
+      "loss": 0.0472,
+      "step": 803
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2250190388107175e-05,
+      "loss": 0.0171,
+      "step": 804
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2232705196535006e-05,
+      "loss": 0.0665,
+      "step": 805
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2215212817779556e-05,
+      "loss": 0.0213,
+      "step": 806
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2197713308149616e-05,
+      "loss": 0.0526,
+      "step": 807
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2180206723976938e-05,
+      "loss": 0.0218,
+      "step": 808
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2162693121616039e-05,
+      "loss": 0.0261,
+      "step": 809
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2145172557444024e-05,
+      "loss": 0.0184,
+      "step": 810
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127645087860427e-05,
+      "loss": 0.0323,
+      "step": 811
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2110110769286988e-05,
+      "loss": 0.0477,
+      "step": 812
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2092569658167514e-05,
+      "loss": 0.031,
+      "step": 813
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207502181096766e-05,
+      "loss": 0.0101,
+      "step": 814
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2057467284174782e-05,
+      "loss": 0.0744,
+      "step": 815
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2039906134297724e-05,
+      "loss": 0.0485,
+      "step": 816
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2022338417866654e-05,
+      "loss": 0.037,
+      "step": 817
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004764191432882e-05,
+      "loss": 0.0317,
+      "step": 818
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987183511568674e-05,
+      "loss": 0.0223,
+      "step": 819
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1969596434867063e-05,
+      "loss": 0.0293,
+      "step": 820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1952003017941677e-05,
+      "loss": 0.0105,
+      "step": 821
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.193440331742656e-05,
+      "loss": 0.0678,
+      "step": 822
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1916797389975974e-05,
+      "loss": 0.0443,
+      "step": 823
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1899185292264226e-05,
+      "loss": 0.0579,
+      "step": 824
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1881567080985499e-05,
+      "loss": 0.045,
+      "step": 825
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1863942812853635e-05,
+      "loss": 0.0329,
+      "step": 826
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1846312544601988e-05,
+      "loss": 0.0785,
+      "step": 827
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1828676332983222e-05,
+      "loss": 0.0527,
+      "step": 828
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181103423476914e-05,
+      "loss": 0.0344,
+      "step": 829
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1793386306750476e-05,
+      "loss": 0.0343,
+      "step": 830
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1775732605736751e-05,
+      "loss": 0.0628,
+      "step": 831
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175807318855606e-05,
+      "loss": 0.0287,
+      "step": 832
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1740408112054895e-05,
+      "loss": 0.0908,
+      "step": 833
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.172273743309797e-05,
+      "loss": 0.0661,
+      "step": 834
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1705061208568042e-05,
+      "loss": 0.0319,
+      "step": 835
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1687379495365703e-05,
+      "loss": 0.0372,
+      "step": 836
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1669692350409223e-05,
+      "loss": 0.0227,
+      "step": 837
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1651999830634356e-05,
+      "loss": 0.0313,
+      "step": 838
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1634301992994157e-05,
+      "loss": 0.0366,
+      "step": 839
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1616598894458799e-05,
+      "loss": 0.0212,
+      "step": 840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1598890592015395e-05,
+      "loss": 0.0787,
+      "step": 841
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.15811771426678e-05,
+      "loss": 0.048,
+      "step": 842
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563458603436443e-05,
+      "loss": 0.0487,
+      "step": 843
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154573503135814e-05,
+      "loss": 0.0559,
+      "step": 844
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1528006483485903e-05,
+      "loss": 0.0464,
+      "step": 845
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1510273016888763e-05,
+      "loss": 0.0236,
+      "step": 846
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1492534688651586e-05,
+      "loss": 0.0311,
+      "step": 847
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1474791555874888e-05,
+      "loss": 0.0302,
+      "step": 848
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1457043675674648e-05,
+      "loss": 0.0497,
+      "step": 849
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1439291105182128e-05,
+      "loss": 0.0253,
+      "step": 850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142153390154369e-05,
+      "loss": 0.0221,
+      "step": 851
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1403772121920616e-05,
+      "loss": 0.0287,
+      "step": 852
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1386005823488902e-05,
+      "loss": 0.0614,
+      "step": 853
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1368235063439103e-05,
+      "loss": 0.0243,
+      "step": 854
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1350459898976137e-05,
+      "loss": 0.0198,
+      "step": 855
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1332680387319094e-05,
+      "loss": 0.0301,
+      "step": 856
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1314896585701058e-05,
+      "loss": 0.042,
+      "step": 857
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1297108551368929e-05,
+      "loss": 0.0462,
+      "step": 858
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1279316341583227e-05,
+      "loss": 0.0239,
+      "step": 859
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1261520013617913e-05,
+      "loss": 0.1448,
+      "step": 860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1243719624760203e-05,
+      "loss": 0.0334,
+      "step": 861
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1225915232310393e-05,
+      "loss": 0.0128,
+      "step": 862
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1208106893581658e-05,
+      "loss": 0.0211,
+      "step": 863
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1190294665899882e-05,
+      "loss": 0.0314,
+      "step": 864
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1172478606603463e-05,
+      "loss": 0.0872,
+      "step": 865
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115465877304314e-05,
+      "loss": 0.1172,
+      "step": 866
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1136835222581792e-05,
+      "loss": 0.0373,
+      "step": 867
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1119008012594275e-05,
+      "loss": 0.0142,
+      "step": 868
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110117720046721e-05,
+      "loss": 0.0307,
+      "step": 869
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.108334284359883e-05,
+      "loss": 0.0598,
+      "step": 870
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1065504999398762e-05,
+      "loss": 0.0215,
+      "step": 871
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1047663725287877e-05,
+      "loss": 0.0719,
+      "step": 872
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1029819078698068e-05,
+      "loss": 0.0253,
+      "step": 873
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.10119711170721e-05,
+      "loss": 0.0283,
+      "step": 874
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.09941198978634e-05,
+      "loss": 0.0229,
+      "step": 875
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0976265478535886e-05,
+      "loss": 0.032,
+      "step": 876
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0958407916563772e-05,
+      "loss": 0.0484,
+      "step": 877
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0940547269431396e-05,
+      "loss": 0.039,
+      "step": 878
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092268359463302e-05,
+      "loss": 0.0338,
+      "step": 879
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090481694967266e-05,
+      "loss": 0.0653,
+      "step": 880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088694739206388e-05,
+      "loss": 0.0591,
+      "step": 881
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0869074979329639e-05,
+      "loss": 0.0171,
+      "step": 882
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0851199769002066e-05,
+      "loss": 0.0403,
+      "step": 883
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0833321818622311e-05,
+      "loss": 0.01,
+      "step": 884
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0815441185740338e-05,
+      "loss": 0.044,
+      "step": 885
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079755792791475e-05,
+      "loss": 0.0243,
+      "step": 886
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.077967210271259e-05,
+      "loss": 0.0222,
+      "step": 887
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0761783767709182e-05,
+      "loss": 0.035,
+      "step": 888
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0743892980487911e-05,
+      "loss": 0.0139,
+      "step": 889
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072599979864007e-05,
+      "loss": 0.0343,
+      "step": 890
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0708104279764648e-05,
+      "loss": 0.0327,
+      "step": 891
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0690206481468174e-05,
+      "loss": 0.0138,
+      "step": 892
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0672306461364497e-05,
+      "loss": 0.0126,
+      "step": 893
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0654404277074626e-05,
+      "loss": 0.0274,
+      "step": 894
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0636499986226538e-05,
+      "loss": 0.0389,
+      "step": 895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0618593646454993e-05,
+      "loss": 0.0444,
+      "step": 896
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0600685315401336e-05,
+      "loss": 0.0392,
+      "step": 897
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0582775050713333e-05,
+      "loss": 0.0248,
+      "step": 898
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0564862910044967e-05,
+      "loss": 0.0163,
+      "step": 899
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0546948951056268e-05,
+      "loss": 0.007,
+      "step": 900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052903323141311e-05,
+      "loss": 0.0577,
+      "step": 901
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511115808787042e-05,
+      "loss": 0.0222,
+      "step": 902
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049319674085509e-05,
+      "loss": 0.0071,
+      "step": 903
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.047527608529958e-05,
+      "loss": 0.0418,
+      "step": 904
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0457353899807947e-05,
+      "loss": 0.0424,
+      "step": 905
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0439430242072547e-05,
+      "loss": 0.0195,
+      "step": 906
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0421505169790487e-05,
+      "loss": 0.0312,
+      "step": 907
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0403578740663412e-05,
+      "loss": 0.0142,
+      "step": 908
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038565101239735e-05,
+      "loss": 0.0701,
+      "step": 909
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0367722042702494e-05,
+      "loss": 0.0187,
+      "step": 910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.034979188929305e-05,
+      "loss": 0.0206,
+      "step": 911
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0331860609887023e-05,
+      "loss": 0.0341,
+      "step": 912
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0313928262206052e-05,
+      "loss": 0.0141,
+      "step": 913
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0295994903975201e-05,
+      "loss": 0.0458,
+      "step": 914
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0278060592922807e-05,
+      "loss": 0.0538,
+      "step": 915
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0260125386780254e-05,
+      "loss": 0.027,
+      "step": 916
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0242189343281816e-05,
+      "loss": 0.0309,
+      "step": 917
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0224252520164466e-05,
+      "loss": 0.0475,
+      "step": 918
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0206314975167684e-05,
+      "loss": 0.0417,
+      "step": 919
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0188376766033268e-05,
+      "loss": 0.0086,
+      "step": 920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0170437950505164e-05,
+      "loss": 0.0688,
+      "step": 921
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015249858632926e-05,
+      "loss": 0.0262,
+      "step": 922
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0134558731253221e-05,
+      "loss": 0.0551,
+      "step": 923
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011661844302628e-05,
+      "loss": 0.01,
+      "step": 924
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0098677779399078e-05,
+      "loss": 0.0313,
+      "step": 925
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0080736798123447e-05,
+      "loss": 0.0687,
+      "step": 926
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006279555695226e-05,
+      "loss": 0.0093,
+      "step": 927
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044854113639213e-05,
+      "loss": 0.0338,
+      "step": 928
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002691252593866e-05,
+      "loss": 0.0657,
+      "step": 929
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0008970851605419e-05,
+      "loss": 0.0504,
+      "step": 930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.991029148394583e-06,
+      "loss": 0.0211,
+      "step": 931
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.973087474061342e-06,
+      "loss": 0.0288,
+      "step": 932
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.955145886360789e-06,
+      "loss": 0.0311,
+      "step": 933
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.937204443047742e-06,
+      "loss": 0.0356,
+      "step": 934
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.919263201876558e-06,
+      "loss": 0.042,
+      "step": 935
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901322220600927e-06,
+      "loss": 0.0592,
+      "step": 936
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88338155697372e-06,
+      "loss": 0.0063,
+      "step": 937
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.865441268746782e-06,
+      "loss": 0.0235,
+      "step": 938
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.847501413670742e-06,
+      "loss": 0.0155,
+      "step": 939
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.829562049494838e-06,
+      "loss": 0.0265,
+      "step": 940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.811623233966737e-06,
+      "loss": 0.0468,
+      "step": 941
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79368502483232e-06,
+      "loss": 0.1058,
+      "step": 942
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.775747479835536e-06,
+      "loss": 0.0788,
+      "step": 943
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.757810656718187e-06,
+      "loss": 0.0165,
+      "step": 944
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.739874613219752e-06,
+      "loss": 0.0536,
+      "step": 945
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.721939407077196e-06,
+      "loss": 0.0114,
+      "step": 946
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7040050960248e-06,
+      "loss": 0.0728,
+      "step": 947
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68607173779395e-06,
+      "loss": 0.0396,
+      "step": 948
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.668139390112977e-06,
+      "loss": 0.0275,
+      "step": 949
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.650208110706954e-06,
+      "loss": 0.0132,
+      "step": 950
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.632277957297509e-06,
+      "loss": 0.0317,
+      "step": 951
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.614348987602656e-06,
+      "loss": 0.0486,
+      "step": 952
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.596421259336592e-06,
+      "loss": 0.0194,
+      "step": 953
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.578494830209518e-06,
+      "loss": 0.0331,
+      "step": 954
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.560569757927456e-06,
+      "loss": 0.0753,
+      "step": 955
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.542646100192056e-06,
+      "loss": 0.0516,
+      "step": 956
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524723914700422e-06,
+      "loss": 0.018,
+      "step": 957
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.506803259144914e-06,
+      "loss": 0.0182,
+      "step": 958
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488884191212961e-06,
+      "loss": 0.0423,
+      "step": 959
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.470966768586892e-06,
+      "loss": 0.0067,
+      "step": 960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.453051048943736e-06,
+      "loss": 0.0446,
+      "step": 961
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.435137089955035e-06,
+      "loss": 0.0607,
+      "step": 962
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.417224949286669e-06,
+      "loss": 0.0256,
+      "step": 963
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.399314684598665e-06,
+      "loss": 0.0322,
+      "step": 964
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.381406353545009e-06,
+      "loss": 0.0193,
+      "step": 965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.363500013773462e-06,
+      "loss": 0.01,
+      "step": 966
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.345595722925377e-06,
+      "loss": 0.0129,
+      "step": 967
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.327693538635507e-06,
+      "loss": 0.0591,
+      "step": 968
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309793518531831e-06,
+      "loss": 0.0708,
+      "step": 969
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.291895720235355e-06,
+      "loss": 0.014,
+      "step": 970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.274000201359935e-06,
+      "loss": 0.0282,
+      "step": 971
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.256107019512094e-06,
+      "loss": 0.0455,
+      "step": 972
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.238216232290821e-06,
+      "loss": 0.0417,
+      "step": 973
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.220327897287411e-06,
+      "loss": 0.0617,
+      "step": 974
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.202442072085255e-06,
+      "loss": 0.0873,
+      "step": 975
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.184558814259664e-06,
+      "loss": 0.0913,
+      "step": 976
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.166678181377689e-06,
+      "loss": 0.0131,
+      "step": 977
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.14880023099794e-06,
+      "loss": 0.0072,
+      "step": 978
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.130925020670365e-06,
+      "loss": 0.0358,
+      "step": 979
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.11305260793612e-06,
+      "loss": 0.0487,
+      "step": 980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.095183050327344e-06,
+      "loss": 0.0503,
+      "step": 981
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07731640536698e-06,
+      "loss": 0.035,
+      "step": 982
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.059452730568604e-06,
+      "loss": 0.0109,
+      "step": 983
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.041592083436233e-06,
+      "loss": 0.0079,
+      "step": 984
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.023734521464117e-06,
+      "loss": 0.0316,
+      "step": 985
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.005880102136601e-06,
+      "loss": 0.0477,
+      "step": 986
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.988028882927904e-06,
+      "loss": 0.0061,
+      "step": 987
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.970180921301934e-06,
+      "loss": 0.0104,
+      "step": 988
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.952336274712128e-06,
+      "loss": 0.0326,
+      "step": 989
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.934495000601241e-06,
+      "loss": 0.0318,
+      "step": 990
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.916657156401172e-06,
+      "loss": 0.0423,
+      "step": 991
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.898822799532794e-06,
+      "loss": 0.0187,
+      "step": 992
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.880991987405728e-06,
+      "loss": 0.0349,
+      "step": 993
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.863164777418208e-06,
+      "loss": 0.0212,
+      "step": 994
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.845341226956864e-06,
+      "loss": 0.0094,
+      "step": 995
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.827521393396538e-06,
+      "loss": 0.0118,
+      "step": 996
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80970533410012e-06,
+      "loss": 0.0598,
+      "step": 997
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.791893106418347e-06,
+      "loss": 0.0288,
+      "step": 998
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77408476768961e-06,
+      "loss": 0.0161,
+      "step": 999
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.756280375239798e-06,
+      "loss": 0.0225,
+      "step": 1000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.738479986382092e-06,
+      "loss": 0.0335,
+      "step": 1001
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.720683658416777e-06,
+      "loss": 0.0565,
+      "step": 1002
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.702891448631073e-06,
+      "loss": 0.023,
+      "step": 1003
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.685103414298945e-06,
+      "loss": 0.1115,
+      "step": 1004
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.667319612680909e-06,
+      "loss": 0.0053,
+      "step": 1005
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.649540101023865e-06,
+      "loss": 0.0219,
+      "step": 1006
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6317649365609e-06,
+      "loss": 0.0502,
+      "step": 1007
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613994176511103e-06,
+      "loss": 0.0198,
+      "step": 1008
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.59622787807939e-06,
+      "loss": 0.0122,
+      "step": 1009
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.578466098456312e-06,
+      "loss": 0.0123,
+      "step": 1010
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.560708894817875e-06,
+      "loss": 0.0047,
+      "step": 1011
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.542956324325357e-06,
+      "loss": 0.0445,
+      "step": 1012
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.525208444125115e-06,
+      "loss": 0.0121,
+      "step": 1013
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.507465311348415e-06,
+      "loss": 0.0076,
+      "step": 1014
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.48972698311124e-06,
+      "loss": 0.0147,
+      "step": 1015
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.471993516514102e-06,
+      "loss": 0.0649,
+      "step": 1016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.454264968641863e-06,
+      "loss": 0.0697,
+      "step": 1017
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.436541396563562e-06,
+      "loss": 0.0433,
+      "step": 1018
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.418822857332205e-06,
+      "loss": 0.0151,
+      "step": 1019
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.401109407984606e-06,
+      "loss": 0.0296,
+      "step": 1020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.383401105541203e-06,
+      "loss": 0.016,
+      "step": 1021
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.365698007005845e-06,
+      "loss": 0.0081,
+      "step": 1022
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.348000169365646e-06,
+      "loss": 0.0361,
+      "step": 1023
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.330307649590782e-06,
+      "loss": 0.0484,
+      "step": 1024
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.312620504634302e-06,
+      "loss": 0.0306,
+      "step": 1025
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.29493879143196e-06,
+      "loss": 0.023,
+      "step": 1026
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.277262566902032e-06,
+      "loss": 0.0217,
+      "step": 1027
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.259591887945108e-06,
+      "loss": 0.018,
+      "step": 1028
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.241926811443946e-06,
+      "loss": 0.0063,
+      "step": 1029
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.224267394263252e-06,
+      "loss": 0.0291,
+      "step": 1030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.206613693249526e-06,
+      "loss": 0.0314,
+      "step": 1031
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.188965765230866e-06,
+      "loss": 0.0184,
+      "step": 1032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.17132366701678e-06,
+      "loss": 0.0077,
+      "step": 1033
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.153687455398013e-06,
+      "loss": 0.0603,
+      "step": 1034
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13605718714637e-06,
+      "loss": 0.0249,
+      "step": 1035
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.118432919014504e-06,
+      "loss": 0.0308,
+      "step": 1036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.100814707735774e-06,
+      "loss": 0.045,
+      "step": 1037
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08320261002403e-06,
+      "loss": 0.0066,
+      "step": 1038
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.065596682573444e-06,
+      "loss": 0.0109,
+      "step": 1039
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.047996982058323e-06,
+      "loss": 0.0111,
+      "step": 1040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.030403565132942e-06,
+      "loss": 0.0261,
+      "step": 1041
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012816488431328e-06,
+      "loss": 0.0165,
+      "step": 1042
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99523580856712e-06,
+      "loss": 0.0228,
+      "step": 1043
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.977661582133349e-06,
+      "loss": 0.0088,
+      "step": 1044
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.960093865702281e-06,
+      "loss": 0.0418,
+      "step": 1045
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.942532715825223e-06,
+      "loss": 0.0264,
+      "step": 1046
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.924978189032342e-06,
+      "loss": 0.0199,
+      "step": 1047
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.907430341832487e-06,
+      "loss": 0.0356,
+      "step": 1048
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.889889230713017e-06,
+      "loss": 0.0407,
+      "step": 1049
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.872354912139578e-06,
+      "loss": 0.0612,
+      "step": 1050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.854827442555977e-06,
+      "loss": 0.0058,
+      "step": 1051
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.837306878383966e-06,
+      "loss": 0.0147,
+      "step": 1052
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.819793276023065e-06,
+      "loss": 0.0339,
+      "step": 1053
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.802286691850382e-06,
+      "loss": 0.0058,
+      "step": 1054
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78478718222045e-06,
+      "loss": 0.0065,
+      "step": 1055
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.767294803464996e-06,
+      "loss": 0.0296,
+      "step": 1056
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.749809611892825e-06,
+      "loss": 0.0431,
+      "step": 1057
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.732331663789592e-06,
+      "loss": 0.0114,
+      "step": 1058
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71486101541763e-06,
+      "loss": 0.0479,
+      "step": 1059
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.697397723015784e-06,
+      "loss": 0.0232,
+      "step": 1060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.679941842799219e-06,
+      "loss": 0.0281,
+      "step": 1061
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.662493430959226e-06,
+      "loss": 0.018,
+      "step": 1062
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.645052543663074e-06,
+      "loss": 0.0485,
+      "step": 1063
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.627619237053799e-06,
+      "loss": 0.0146,
+      "step": 1064
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.610193567250036e-06,
+      "loss": 0.0105,
+      "step": 1065
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.59277559034584e-06,
+      "loss": 0.0603,
+      "step": 1066
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.575365362410495e-06,
+      "loss": 0.0196,
+      "step": 1067
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.557962939488349e-06,
+      "loss": 0.0213,
+      "step": 1068
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.540568377598623e-06,
+      "loss": 0.0395,
+      "step": 1069
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.523181732735225e-06,
+      "loss": 0.0548,
+      "step": 1070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.505803060866591e-06,
+      "loss": 0.0106,
+      "step": 1071
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.488432417935486e-06,
+      "loss": 0.0221,
+      "step": 1072
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471069859858824e-06,
+      "loss": 0.04,
+      "step": 1073
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4537154425274984e-06,
+      "loss": 0.0196,
+      "step": 1074
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.436369221806201e-06,
+      "loss": 0.0201,
+      "step": 1075
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.419031253533228e-06,
+      "loss": 0.0083,
+      "step": 1076
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.401701593520316e-06,
+      "loss": 0.0402,
+      "step": 1077
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.384380297552468e-06,
+      "loss": 0.0057,
+      "step": 1078
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.367067421387742e-06,
+      "loss": 0.0192,
+      "step": 1079
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.349763020757106e-06,
+      "loss": 0.0053,
+      "step": 1080
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3324671513642424e-06,
+      "loss": 0.0263,
+      "step": 1081
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.315179868885367e-06,
+      "loss": 0.0342,
+      "step": 1082
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.297901228969057e-06,
+      "loss": 0.0621,
+      "step": 1083
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.280631287236077e-06,
+      "loss": 0.0321,
+      "step": 1084
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.263370099279173e-06,
+      "loss": 0.0218,
+      "step": 1085
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2461177206629285e-06,
+      "loss": 0.021,
+      "step": 1086
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.228874206923561e-06,
+      "loss": 0.0143,
+      "step": 1087
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.211639613568755e-06,
+      "loss": 0.0404,
+      "step": 1088
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.19441399607748e-06,
+      "loss": 0.0533,
+      "step": 1089
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.17719740989981e-06,
+      "loss": 0.0218,
+      "step": 1090
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.159989910456745e-06,
+      "loss": 0.0081,
+      "step": 1091
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.142791553140045e-06,
+      "loss": 0.0083,
+      "step": 1092
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.125602393312021e-06,
+      "loss": 0.0393,
+      "step": 1093
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.108422486305398e-06,
+      "loss": 0.0101,
+      "step": 1094
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0912518874231055e-06,
+      "loss": 0.0149,
+      "step": 1095
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.074090651938106e-06,
+      "loss": 0.0163,
+      "step": 1096
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.056938835093226e-06,
+      "loss": 0.1022,
+      "step": 1097
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.039796492100981e-06,
+      "loss": 0.0119,
+      "step": 1098
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.02266367814337e-06,
+      "loss": 0.0028,
+      "step": 1099
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.005540448371735e-06,
+      "loss": 0.0319,
+      "step": 1100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.988426857906562e-06,
+      "loss": 0.0528,
+      "step": 1101
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.971322961837299e-06,
+      "loss": 0.0127,
+      "step": 1102
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.954228815222195e-06,
+      "loss": 0.0088,
+      "step": 1103
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.937144473088122e-06,
+      "loss": 0.0231,
+      "step": 1104
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.92006999043037e-06,
+      "loss": 0.0654,
+      "step": 1105
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9030054222125165e-06,
+      "loss": 0.0306,
+      "step": 1106
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8859508233661976e-06,
+      "loss": 0.0024,
+      "step": 1107
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.868906248790981e-06,
+      "loss": 0.0315,
+      "step": 1108
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.851871753354154e-06,
+      "loss": 0.0285,
+      "step": 1109
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.834847391890556e-06,
+      "loss": 0.0468,
+      "step": 1110
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.817833219202411e-06,
+      "loss": 0.0429,
+      "step": 1111
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.80082929005915e-06,
+      "loss": 0.0069,
+      "step": 1112
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.783835659197211e-06,
+      "loss": 0.0537,
+      "step": 1113
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7668523813199e-06,
+      "loss": 0.0332,
+      "step": 1114
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7498795110971905e-06,
+      "loss": 0.0107,
+      "step": 1115
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.732917103165546e-06,
+      "loss": 0.0434,
+      "step": 1116
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.715965212127761e-06,
+      "loss": 0.0354,
+      "step": 1117
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.699023892552772e-06,
+      "loss": 0.0143,
+      "step": 1118
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.682093198975481e-06,
+      "loss": 0.0341,
+      "step": 1119
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.665173185896594e-06,
+      "loss": 0.038,
+      "step": 1120
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.648263907782431e-06,
+      "loss": 0.0299,
+      "step": 1121
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.63136541906475e-06,
+      "loss": 0.0053,
+      "step": 1122
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.61447777414059e-06,
+      "loss": 0.0155,
+      "step": 1123
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597601027372069e-06,
+      "loss": 0.0183,
+      "step": 1124
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.580735233086233e-06,
+      "loss": 0.0165,
+      "step": 1125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.563880445574873e-06,
+      "loss": 0.0363,
+      "step": 1126
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.547036719094335e-06,
+      "loss": 0.0845,
+      "step": 1127
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.530204107865379e-06,
+      "loss": 0.0076,
+      "step": 1128
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.513382666072969e-06,
+      "loss": 0.0202,
+      "step": 1129
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.496572447866121e-06,
+      "loss": 0.0196,
+      "step": 1130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479773507357716e-06,
+      "loss": 0.0134,
+      "step": 1131
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462985898624341e-06,
+      "loss": 0.048,
+      "step": 1132
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.446209675706094e-06,
+      "loss": 0.005,
+      "step": 1133
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.429444892606429e-06,
+      "loss": 0.0078,
+      "step": 1134
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.412691603291979e-06,
+      "loss": 0.0114,
+      "step": 1135
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.395949861692365e-06,
+      "loss": 0.0125,
+      "step": 1136
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.379219721700045e-06,
+      "loss": 0.0369,
+      "step": 1137
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.362501237170129e-06,
+      "loss": 0.0049,
+      "step": 1138
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.345794461920206e-06,
+      "loss": 0.0088,
+      "step": 1139
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3290994497301705e-06,
+      "loss": 0.0231,
+      "step": 1140
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.312416254342062e-06,
+      "loss": 0.0146,
+      "step": 1141
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.295744929459861e-06,
+      "loss": 0.0177,
+      "step": 1142
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2790855287493605e-06,
+      "loss": 0.0289,
+      "step": 1143
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.262438105837947e-06,
+      "loss": 0.0245,
+      "step": 1144
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245802714314464e-06,
+      "loss": 0.0228,
+      "step": 1145
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2291794077290226e-06,
+      "loss": 0.0094,
+      "step": 1146
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.212568239592827e-06,
+      "loss": 0.0307,
+      "step": 1147
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1959692633780076e-06,
+      "loss": 0.0481,
+      "step": 1148
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.179382532517464e-06,
+      "loss": 0.0075,
+      "step": 1149
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.162808100404648e-06,
+      "loss": 0.0117,
+      "step": 1150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1462460203934495e-06,
+      "loss": 0.0213,
+      "step": 1151
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129696345797982e-06,
+      "loss": 0.0182,
+      "step": 1152
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.113159129892427e-06,
+      "loss": 0.0757,
+      "step": 1153
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.096634425910859e-06,
+      "loss": 0.0338,
+      "step": 1154
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.08012228704709e-06,
+      "loss": 0.0341,
+      "step": 1155
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.06362276645446e-06,
+      "loss": 0.0167,
+      "step": 1156
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.047135917245712e-06,
+      "loss": 0.0338,
+      "step": 1157
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.03066179249279e-06,
+      "loss": 0.0194,
+      "step": 1158
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.014200445226674e-06,
+      "loss": 0.0114,
+      "step": 1159
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99775192843722e-06,
+      "loss": 0.0551,
+      "step": 1160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981316295072983e-06,
+      "loss": 0.0282,
+      "step": 1161
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.96489359804103e-06,
+      "loss": 0.0324,
+      "step": 1162
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.948483890206813e-06,
+      "loss": 0.0376,
+      "step": 1163
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.932087224393939e-06,
+      "loss": 0.0213,
+      "step": 1164
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.915703653384057e-06,
+      "loss": 0.0244,
+      "step": 1165
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.899333229916655e-06,
+      "loss": 0.0109,
+      "step": 1166
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.882976006688893e-06,
+      "loss": 0.0107,
+      "step": 1167
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.866632036355444e-06,
+      "loss": 0.0281,
+      "step": 1168
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.850301371528329e-06,
+      "loss": 0.0032,
+      "step": 1169
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.833984064776713e-06,
+      "loss": 0.0053,
+      "step": 1170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.817680168626784e-06,
+      "loss": 0.0326,
+      "step": 1171
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.801389735561558e-06,
+      "loss": 0.0264,
+      "step": 1172
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.785112818020691e-06,
+      "loss": 0.0391,
+      "step": 1173
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7688494684003595e-06,
+      "loss": 0.015,
+      "step": 1174
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.752599739053055e-06,
+      "loss": 0.0182,
+      "step": 1175
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.736363682287414e-06,
+      "loss": 0.0426,
+      "step": 1176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.720141350368072e-06,
+      "loss": 0.0085,
+      "step": 1177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.70393279551548e-06,
+      "loss": 0.0148,
+      "step": 1178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6877380699057405e-06,
+      "loss": 0.0164,
+      "step": 1179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.671557225670439e-06,
+      "loss": 0.0453,
+      "step": 1180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.655390314896474e-06,
+      "loss": 0.0763,
+      "step": 1181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.639237389625896e-06,
+      "loss": 0.0283,
+      "step": 1182
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.623098501855735e-06,
+      "loss": 0.0278,
+      "step": 1183
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.606973703537826e-06,
+      "loss": 0.0166,
+      "step": 1184
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.590863046578655e-06,
+      "loss": 0.0363,
+      "step": 1185
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.574766582839189e-06,
+      "loss": 0.0068,
+      "step": 1186
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.558684364134703e-06,
+      "loss": 0.0049,
+      "step": 1187
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.542616442234618e-06,
+      "loss": 0.0214,
+      "step": 1188
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5265628688623305e-06,
+      "loss": 0.0229,
+      "step": 1189
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.510523695695052e-06,
+      "loss": 0.0062,
+      "step": 1190
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.494498974363633e-06,
+      "loss": 0.0035,
+      "step": 1191
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.478488756452416e-06,
+      "loss": 0.0141,
+      "step": 1192
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.462493093499037e-06,
+      "loss": 0.0776,
+      "step": 1193
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.446512036994287e-06,
+      "loss": 0.0099,
+      "step": 1194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.430545638381956e-06,
+      "loss": 0.0228,
+      "step": 1195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.414593949058619e-06,
+      "loss": 0.0427,
+      "step": 1196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.398657020373521e-06,
+      "loss": 0.0571,
+      "step": 1197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.382734903628386e-06,
+      "loss": 0.0289,
+      "step": 1198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.36682765007726e-06,
+      "loss": 0.0077,
+      "step": 1199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.350935310926344e-06,
+      "loss": 0.0184,
+      "step": 1200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.335057937333821e-06,
+      "loss": 0.0139,
+      "step": 1201
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.319195580409704e-06,
+      "loss": 0.0034,
+      "step": 1202
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.303348291215679e-06,
+      "loss": 0.0616,
+      "step": 1203
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.287516120764906e-06,
+      "loss": 0.0287,
+      "step": 1204
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.271699120021892e-06,
+      "loss": 0.0032,
+      "step": 1205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2558973399023075e-06,
+      "loss": 0.0186,
+      "step": 1206
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.240110831272827e-06,
+      "loss": 0.0087,
+      "step": 1207
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2243396449509685e-06,
+      "loss": 0.0782,
+      "step": 1208
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.208583831704925e-06,
+      "loss": 0.0027,
+      "step": 1209
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.192843442253393e-06,
+      "loss": 0.0282,
+      "step": 1210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.177118527265438e-06,
+      "loss": 0.0677,
+      "step": 1211
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.161409137360304e-06,
+      "loss": 0.0023,
+      "step": 1212
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.145715323107247e-06,
+      "loss": 0.0251,
+      "step": 1213
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1300371350254005e-06,
+      "loss": 0.024,
+      "step": 1214
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.114374623583598e-06,
+      "loss": 0.0032,
+      "step": 1215
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.098727839200192e-06,
+      "loss": 0.0123,
+      "step": 1216
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.083096832242922e-06,
+      "loss": 0.0127,
+      "step": 1217
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.067481653028738e-06,
+      "loss": 0.0086,
+      "step": 1218
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.051882351823636e-06,
+      "loss": 0.0082,
+      "step": 1219
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.036298978842506e-06,
+      "loss": 0.0058,
+      "step": 1220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.020731584248956e-06,
+      "loss": 0.003,
+      "step": 1221
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.005180218155159e-06,
+      "loss": 0.0039,
+      "step": 1222
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.989644930621708e-06,
+      "loss": 0.0139,
+      "step": 1223
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.974125771657416e-06,
+      "loss": 0.009,
+      "step": 1224
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.95862279121919e-06,
+      "loss": 0.0076,
+      "step": 1225
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943136039211854e-06,
+      "loss": 0.0525,
+      "step": 1226
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.927665565487995e-06,
+      "loss": 0.0246,
+      "step": 1227
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.912211419847795e-06,
+      "loss": 0.0224,
+      "step": 1228
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896773652038882e-06,
+      "loss": 0.015,
+      "step": 1229
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.881352311756147e-06,
+      "loss": 0.0073,
+      "step": 1230
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.865947448641624e-06,
+      "loss": 0.0113,
+      "step": 1231
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.85055911228429e-06,
+      "loss": 0.0089,
+      "step": 1232
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.835187352219921e-06,
+      "loss": 0.0045,
+      "step": 1233
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8198322179309395e-06,
+      "loss": 0.0235,
+      "step": 1234
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8044937588462485e-06,
+      "loss": 0.0059,
+      "step": 1235
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.789172024341068e-06,
+      "loss": 0.006,
+      "step": 1236
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7738670637367855e-06,
+      "loss": 0.0207,
+      "step": 1237
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.758578926300791e-06,
+      "loss": 0.0054,
+      "step": 1238
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.743307661246318e-06,
+      "loss": 0.0082,
+      "step": 1239
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.728053317732293e-06,
+      "loss": 0.0026,
+      "step": 1240
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.712815944863161e-06,
+      "loss": 0.0191,
+      "step": 1241
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697595591688744e-06,
+      "loss": 0.0468,
+      "step": 1242
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.682392307204079e-06,
+      "loss": 0.0274,
+      "step": 1243
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.667206140349255e-06,
+      "loss": 0.0156,
+      "step": 1244
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.652037140009259e-06,
+      "loss": 0.0163,
+      "step": 1245
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6368853550138184e-06,
+      "loss": 0.0129,
+      "step": 1246
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621750834137242e-06,
+      "loss": 0.0193,
+      "step": 1247
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.606633626098267e-06,
+      "loss": 0.0209,
+      "step": 1248
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.591533779559902e-06,
+      "loss": 0.0264,
+      "step": 1249
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.576451343129258e-06,
+      "loss": 0.0066,
+      "step": 1250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5613863653574086e-06,
+      "loss": 0.0085,
+      "step": 1251
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5463388947392385e-06,
+      "loss": 0.0028,
+      "step": 1252
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531308979713255e-06,
+      "loss": 0.0022,
+      "step": 1253
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.516296668661464e-06,
+      "loss": 0.0344,
+      "step": 1254
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5013020099092056e-06,
+      "loss": 0.0101,
+      "step": 1255
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.48632505172499e-06,
+      "loss": 0.0208,
+      "step": 1256
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.471365842320354e-06,
+      "loss": 0.0068,
+      "step": 1257
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.456424429849702e-06,
+      "loss": 0.0042,
+      "step": 1258
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.441500862410132e-06,
+      "loss": 0.0374,
+      "step": 1259
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.426595188041327e-06,
+      "loss": 0.0094,
+      "step": 1260
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.411707454725344e-06,
+      "loss": 0.0141,
+      "step": 1261
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.396837710386503e-06,
+      "loss": 0.0038,
+      "step": 1262
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3819860028912095e-06,
+      "loss": 0.0036,
+      "step": 1263
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.367152380047813e-06,
+      "loss": 0.003,
+      "step": 1264
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.352336889606444e-06,
+      "loss": 0.0036,
+      "step": 1265
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.337539579258867e-06,
+      "loss": 0.0129,
+      "step": 1266
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3227604966383115e-06,
+      "loss": 0.0048,
+      "step": 1267
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3079996893193536e-06,
+      "loss": 0.0161,
+      "step": 1268
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.293257204817724e-06,
+      "loss": 0.0027,
+      "step": 1269
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278533090590172e-06,
+      "loss": 0.0061,
+      "step": 1270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.263827394034313e-06,
+      "loss": 0.0085,
+      "step": 1271
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.24914016248849e-06,
+      "loss": 0.0118,
+      "step": 1272
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.23447144323158e-06,
+      "loss": 0.0038,
+      "step": 1273
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.219821283482889e-06,
+      "loss": 0.0252,
+      "step": 1274
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2051897304019694e-06,
+      "loss": 0.0108,
+      "step": 1275
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.190576831088482e-06,
+      "loss": 0.0056,
+      "step": 1276
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.17598263258204e-06,
+      "loss": 0.0031,
+      "step": 1277
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.161407181862052e-06,
+      "loss": 0.0079,
+      "step": 1278
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1468505258475785e-06,
+      "loss": 0.0109,
+      "step": 1279
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.132312711397193e-06,
+      "loss": 0.005,
+      "step": 1280
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.117793785308796e-06,
+      "loss": 0.0099,
+      "step": 1281
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103293794319496e-06,
+      "loss": 0.0082,
+      "step": 1282
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0888127851054495e-06,
+      "loss": 0.0046,
+      "step": 1283
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.074350804281708e-06,
+      "loss": 0.0036,
+      "step": 1284
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.059907898402068e-06,
+      "loss": 0.008,
+      "step": 1285
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.045484113958929e-06,
+      "loss": 0.0055,
+      "step": 1286
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.031079497383124e-06,
+      "loss": 0.0481,
+      "step": 1287
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0166940950437995e-06,
+      "loss": 0.0021,
+      "step": 1288
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0023279532482466e-06,
+      "loss": 0.0307,
+      "step": 1289
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987981118241745e-06,
+      "loss": 0.022,
+      "step": 1290
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.973653636207437e-06,
+      "loss": 0.0113,
+      "step": 1291
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.959345553266163e-06,
+      "loss": 0.0018,
+      "step": 1292
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.945056915476315e-06,
+      "loss": 0.0181,
+      "step": 1293
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930787768833692e-06,
+      "loss": 0.0196,
+      "step": 1294
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.916538159271353e-06,
+      "loss": 0.0117,
+      "step": 1295
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.902308132659457e-06,
+      "loss": 0.0085,
+      "step": 1296
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.888097734805138e-06,
+      "loss": 0.0137,
+      "step": 1297
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8739070114523296e-06,
+      "loss": 0.0052,
+      "step": 1298
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.859736008281642e-06,
+      "loss": 0.0204,
+      "step": 1299
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845584770910201e-06,
+      "loss": 0.0408,
+      "step": 1300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.831453344891508e-06,
+      "loss": 0.0157,
+      "step": 1301
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817341775715291e-06,
+      "loss": 0.0056,
+      "step": 1302
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8032501088073536e-06,
+      "loss": 0.0621,
+      "step": 1303
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7891783895294374e-06,
+      "loss": 0.0056,
+      "step": 1304
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7751266631790707e-06,
+      "loss": 0.0019,
+      "step": 1305
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7610949749894256e-06,
+      "loss": 0.0168,
+      "step": 1306
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7470833701291608e-06,
+      "loss": 0.0037,
+      "step": 1307
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.733091893702294e-06,
+      "loss": 0.0473,
+      "step": 1308
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7191205907480563e-06,
+      "loss": 0.0062,
+      "step": 1309
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.705169506240721e-06,
+      "loss": 0.0285,
+      "step": 1310
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.69123868508949e-06,
+      "loss": 0.0046,
+      "step": 1311
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6773281721383337e-06,
+      "loss": 0.0018,
+      "step": 1312
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6634380121658484e-06,
+      "loss": 0.013,
+      "step": 1313
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.649568249885115e-06,
+      "loss": 0.0069,
+      "step": 1314
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6357189299435538e-06,
+      "loss": 0.012,
+      "step": 1315
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6218900969227688e-06,
+      "loss": 0.0022,
+      "step": 1316
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6080817953384404e-06,
+      "loss": 0.0115,
+      "step": 1317
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5942940696401308e-06,
+      "loss": 0.0326,
+      "step": 1318
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.580526964211183e-06,
+      "loss": 0.0031,
+      "step": 1319
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5667805233685572e-06,
+      "loss": 0.0126,
+      "step": 1320
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.553054791362697e-06,
+      "loss": 0.013,
+      "step": 1321
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5393498123773786e-06,
+      "loss": 0.0279,
+      "step": 1322
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.525665630529579e-06,
+      "loss": 0.0181,
+      "step": 1323
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.512002289869315e-06,
+      "loss": 0.0117,
+      "step": 1324
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4983598343795343e-06,
+      "loss": 0.0073,
+      "step": 1325
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4847383079759446e-06,
+      "loss": 0.0107,
+      "step": 1326
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4711377545068747e-06,
+      "loss": 0.038,
+      "step": 1327
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4575582177531462e-06,
+      "loss": 0.0373,
+      "step": 1328
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4439997414279403e-06,
+      "loss": 0.0112,
+      "step": 1329
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4304623691766193e-06,
+      "loss": 0.0158,
+      "step": 1330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4169461445766274e-06,
+      "loss": 0.0129,
+      "step": 1331
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.403451111137328e-06,
+      "loss": 0.0094,
+      "step": 1332
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3899773122998704e-06,
+      "loss": 0.0147,
+      "step": 1333
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.376524791437047e-06,
+      "loss": 0.0108,
+      "step": 1334
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3630935918531614e-06,
+      "loss": 0.0018,
+      "step": 1335
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.349683756783868e-06,
+      "loss": 0.0106,
+      "step": 1336
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3362953293960722e-06,
+      "loss": 0.0029,
+      "step": 1337
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3229283527877397e-06,
+      "loss": 0.0023,
+      "step": 1338
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3095828699878052e-06,
+      "loss": 0.0047,
+      "step": 1339
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.296258923956005e-06,
+      "loss": 0.0022,
+      "step": 1340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2829565575827506e-06,
+      "loss": 0.0173,
+      "step": 1341
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.269675813688985e-06,
+      "loss": 0.01,
+      "step": 1342
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2564167350260536e-06,
+      "loss": 0.0078,
+      "step": 1343
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2431793642755447e-06,
+      "loss": 0.0206,
+      "step": 1344
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2299637440491904e-06,
+      "loss": 0.0116,
+      "step": 1345
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.216769916888692e-06,
+      "loss": 0.0017,
+      "step": 1346
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.203597925265598e-06,
+      "loss": 0.0092,
+      "step": 1347
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.190447811581172e-06,
+      "loss": 0.0082,
+      "step": 1348
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1773196181662504e-06,
+      "loss": 0.0061,
+      "step": 1349
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1642133872811053e-06,
+      "loss": 0.0042,
+      "step": 1350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.151129161115313e-06,
+      "loss": 0.0018,
+      "step": 1351
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.138066981787614e-06,
+      "loss": 0.0098,
+      "step": 1352
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.125026891345779e-06,
+      "loss": 0.0068,
+      "step": 1353
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.112008931766479e-06,
+      "loss": 0.0262,
+      "step": 1354
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.099013144955131e-06,
+      "loss": 0.0222,
+      "step": 1355
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.086039572745788e-06,
+      "loss": 0.0074,
+      "step": 1356
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0730882569009936e-06,
+      "loss": 0.0054,
+      "step": 1357
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0601592391116406e-06,
+      "loss": 0.0024,
+      "step": 1358
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0472525609968484e-06,
+      "loss": 0.0108,
+      "step": 1359
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.034368264103823e-06,
+      "loss": 0.0145,
+      "step": 1360
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0215063899077224e-06,
+      "loss": 0.0303,
+      "step": 1361
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0086669798115264e-06,
+      "loss": 0.0025,
+      "step": 1362
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9958500751459042e-06,
+      "loss": 0.0042,
+      "step": 1363
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.98305571716907e-06,
+      "loss": 0.0206,
+      "step": 1364
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9702839470666654e-06,
+      "loss": 0.01,
+      "step": 1365
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9575348059516297e-06,
+      "loss": 0.0036,
+      "step": 1366
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.94480833486404e-06,
+      "loss": 0.0045,
+      "step": 1367
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9321045747710098e-06,
+      "loss": 0.0434,
+      "step": 1368
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9194235665665406e-06,
+      "loss": 0.0074,
+      "step": 1369
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.906765351071397e-06,
+      "loss": 0.0216,
+      "step": 1370
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8941299690329705e-06,
+      "loss": 0.0122,
+      "step": 1371
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.881517461125155e-06,
+      "loss": 0.0176,
+      "step": 1372
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.868927867948199e-06,
+      "loss": 0.0027,
+      "step": 1373
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.856361230028608e-06,
+      "loss": 0.003,
+      "step": 1374
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8438175878189743e-06,
+      "loss": 0.0022,
+      "step": 1375
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.831296981697874e-06,
+      "loss": 0.0075,
+      "step": 1376
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8187994519697314e-06,
+      "loss": 0.0021,
+      "step": 1377
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8063250388646814e-06,
+      "loss": 0.022,
+      "step": 1378
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7938737825384487e-06,
+      "loss": 0.0102,
+      "step": 1379
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.781445723072218e-06,
+      "loss": 0.0114,
+      "step": 1380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7690409004724883e-06,
+      "loss": 0.003,
+      "step": 1381
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7566593546709784e-06,
+      "loss": 0.0309,
+      "step": 1382
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.744301125524468e-06,
+      "loss": 0.0319,
+      "step": 1383
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.731966252814672e-06,
+      "loss": 0.016,
+      "step": 1384
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7196547762481275e-06,
+      "loss": 0.0135,
+      "step": 1385
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7073667354560686e-06,
+      "loss": 0.0257,
+      "step": 1386
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.695102169994267e-06,
+      "loss": 0.0022,
+      "step": 1387
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.682861119342941e-06,
+      "loss": 0.0035,
+      "step": 1388
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6706436229066102e-06,
+      "loss": 0.0324,
+      "step": 1389
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6584497200139703e-06,
+      "loss": 0.0018,
+      "step": 1390
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6462794499177702e-06,
+      "loss": 0.0062,
+      "step": 1391
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.634132851794686e-06,
+      "loss": 0.0105,
+      "step": 1392
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6220099647451804e-06,
+      "loss": 0.0017,
+      "step": 1393
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6099108277934105e-06,
+      "loss": 0.0105,
+      "step": 1394
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.59783547988706e-06,
+      "loss": 0.0479,
+      "step": 1395
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5857839598972457e-06,
+      "loss": 0.0047,
+      "step": 1396
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.573756306618378e-06,
+      "loss": 0.0048,
+      "step": 1397
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.56175255876804e-06,
+      "loss": 0.0189,
+      "step": 1398
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.549772754986862e-06,
+      "loss": 0.0038,
+      "step": 1399
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5378169338384017e-06,
+      "loss": 0.0415,
+      "step": 1400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5258851338089995e-06,
+      "loss": 0.0294,
+      "step": 1401
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.513977393307693e-06,
+      "loss": 0.0333,
+      "step": 1402
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.502093750666058e-06,
+      "loss": 0.0023,
+      "step": 1403
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4902342441380965e-06,
+      "loss": 0.0056,
+      "step": 1404
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.478398911900122e-06,
+      "loss": 0.0135,
+      "step": 1405
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4665877920506263e-06,
+      "loss": 0.015,
+      "step": 1406
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.454800922610161e-06,
+      "loss": 0.0027,
+      "step": 1407
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4430383415212135e-06,
+      "loss": 0.0057,
+      "step": 1408
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4313000866480886e-06,
+      "loss": 0.0082,
+      "step": 1409
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.419586195776781e-06,
+      "loss": 0.0213,
+      "step": 1410
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4078967066148582e-06,
+      "loss": 0.0138,
+      "step": 1411
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.396231656791339e-06,
+      "loss": 0.0087,
+      "step": 1412
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3845910838565646e-06,
+      "loss": 0.0139,
+      "step": 1413
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.372975025282088e-06,
+      "loss": 0.0272,
+      "step": 1414
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3613835184605527e-06,
+      "loss": 0.038,
+      "step": 1415
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.349816600705563e-06,
+      "loss": 0.0021,
+      "step": 1416
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3382743092515734e-06,
+      "loss": 0.0019,
+      "step": 1417
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3267566812537645e-06,
+      "loss": 0.0034,
+      "step": 1418
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.315263753787923e-06,
+      "loss": 0.0127,
+      "step": 1419
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3037955638503274e-06,
+      "loss": 0.0144,
+      "step": 1420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.292352148357616e-06,
+      "loss": 0.0017,
+      "step": 1421
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.280933544146684e-06,
+      "loss": 0.0033,
+      "step": 1422
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2695397879745627e-06,
+      "loss": 0.0017,
+      "step": 1423
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.258170916518284e-06,
+      "loss": 0.0141,
+      "step": 1424
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2468269663747833e-06,
+      "loss": 0.0144,
+      "step": 1425
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.23550797406077e-06,
+      "loss": 0.0114,
+      "step": 1426
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2242139760126146e-06,
+      "loss": 0.0229,
+      "step": 1427
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.212945008586229e-06,
+      "loss": 0.0166,
+      "step": 1428
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.201701108056954e-06,
+      "loss": 0.0047,
+      "step": 1429
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1904823106194263e-06,
+      "loss": 0.0279,
+      "step": 1430
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1792886523874955e-06,
+      "loss": 0.0034,
+      "step": 1431
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1681201693940667e-06,
+      "loss": 0.0039,
+      "step": 1432
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1569768975910153e-06,
+      "loss": 0.0306,
+      "step": 1433
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1458588728490594e-06,
+      "loss": 0.0035,
+      "step": 1434
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.134766130957644e-06,
+      "loss": 0.0427,
+      "step": 1435
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1236987076248284e-06,
+      "loss": 0.0079,
+      "step": 1436
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1126566384771728e-06,
+      "loss": 0.0099,
+      "step": 1437
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1016399590596105e-06,
+      "loss": 0.0191,
+      "step": 1438
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0906487048353587e-06,
+      "loss": 0.0133,
+      "step": 1439
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0796829111857854e-06,
+      "loss": 0.0234,
+      "step": 1440
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0687426134102893e-06,
+      "loss": 0.0028,
+      "step": 1441
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.057827846726206e-06,
+      "loss": 0.0068,
+      "step": 1442
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0469386462686925e-06,
+      "loss": 0.0135,
+      "step": 1443
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.03607504709059e-06,
+      "loss": 0.0053,
+      "step": 1444
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.025237084162337e-06,
+      "loss": 0.0033,
+      "step": 1445
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0144247923718507e-06,
+      "loss": 0.0063,
+      "step": 1446
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0036382065244054e-06,
+      "loss": 0.0099,
+      "step": 1447
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.99287736134253e-06,
+      "loss": 0.005,
+      "step": 1448
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.982142291465896e-06,
+      "loss": 0.0216,
+      "step": 1449
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.97143303145119e-06,
+      "loss": 0.0031,
+      "step": 1450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9607496157720364e-06,
+      "loss": 0.0047,
+      "step": 1451
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.950092078818846e-06,
+      "loss": 0.0074,
+      "step": 1452
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.939460454898735e-06,
+      "loss": 0.0196,
+      "step": 1453
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.928854778235402e-06,
+      "loss": 0.0052,
+      "step": 1454
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.91827508296902e-06,
+      "loss": 0.0019,
+      "step": 1455
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9077214031561274e-06,
+      "loss": 0.0333,
+      "step": 1456
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.897193772769519e-06,
+      "loss": 0.0075,
+      "step": 1457
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8866922256981258e-06,
+      "loss": 0.0076,
+      "step": 1458
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8762167957469301e-06,
+      "loss": 0.0263,
+      "step": 1459
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8657675166368361e-06,
+      "loss": 0.0091,
+      "step": 1460
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8553444220045592e-06,
+      "loss": 0.0526,
+      "step": 1461
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8449475454025356e-06,
+      "loss": 0.014,
+      "step": 1462
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8345769202988007e-06,
+      "loss": 0.0089,
+      "step": 1463
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8242325800768867e-06,
+      "loss": 0.023,
+      "step": 1464
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.813914558035711e-06,
+      "loss": 0.0046,
+      "step": 1465
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8036228873894745e-06,
+      "loss": 0.0066,
+      "step": 1466
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.793357601267549e-06,
+      "loss": 0.0019,
+      "step": 1467
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.783118732714375e-06,
+      "loss": 0.0314,
+      "step": 1468
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7729063146893555e-06,
+      "loss": 0.0043,
+      "step": 1469
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7627203800667404e-06,
+      "loss": 0.0139,
+      "step": 1470
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7525609616355355e-06,
+      "loss": 0.0018,
+      "step": 1471
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7424280920993886e-06,
+      "loss": 0.0115,
+      "step": 1472
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7323218040764833e-06,
+      "loss": 0.0027,
+      "step": 1473
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7222421300994384e-06,
+      "loss": 0.0135,
+      "step": 1474
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7121891026151983e-06,
+      "loss": 0.0427,
+      "step": 1475
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.702162753984934e-06,
+      "loss": 0.0045,
+      "step": 1476
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6921631164839348e-06,
+      "loss": 0.0261,
+      "step": 1477
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6821902223015019e-06,
+      "loss": 0.0179,
+      "step": 1478
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6722441035408522e-06,
+      "loss": 0.0024,
+      "step": 1479
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6623247922190155e-06,
+      "loss": 0.0188,
+      "step": 1480
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6524323202667182e-06,
+      "loss": 0.0039,
+      "step": 1481
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6425667195282947e-06,
+      "loss": 0.0143,
+      "step": 1482
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6327280217615793e-06,
+      "loss": 0.0079,
+      "step": 1483
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6229162586378044e-06,
+      "loss": 0.007,
+      "step": 1484
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6131314617414972e-06,
+      "loss": 0.0174,
+      "step": 1485
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6033736625703832e-06,
+      "loss": 0.002,
+      "step": 1486
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5936428925352698e-06,
+      "loss": 0.0121,
+      "step": 1487
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5839391829599725e-06,
+      "loss": 0.0019,
+      "step": 1488
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5742625650811904e-06,
+      "loss": 0.0057,
+      "step": 1489
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5646130700484064e-06,
+      "loss": 0.0051,
+      "step": 1490
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5549907289238031e-06,
+      "loss": 0.0031,
+      "step": 1491
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.54539557268215e-06,
+      "loss": 0.035,
+      "step": 1492
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.535827632210707e-06,
+      "loss": 0.0087,
+      "step": 1493
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5262869383091295e-06,
+      "loss": 0.0022,
+      "step": 1494
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5167735216893508e-06,
+      "loss": 0.0061,
+      "step": 1495
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5072874129755177e-06,
+      "loss": 0.0039,
+      "step": 1496
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4978286427038602e-06,
+      "loss": 0.008,
+      "step": 1497
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4883972413226e-06,
+      "loss": 0.0035,
+      "step": 1498
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.478993239191865e-06,
+      "loss": 0.0063,
+      "step": 1499
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4696166665835853e-06,
+      "loss": 0.0108,
+      "step": 1500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4602675536813859e-06,
+      "loss": 0.0182,
+      "step": 1501
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.450945930580503e-06,
+      "loss": 0.0032,
+      "step": 1502
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4416518272876811e-06,
+      "loss": 0.0061,
+      "step": 1503
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4323852737210765e-06,
+      "loss": 0.0054,
+      "step": 1504
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4231462997101618e-06,
+      "loss": 0.0094,
+      "step": 1505
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4139349349956332e-06,
+      "loss": 0.0071,
+      "step": 1506
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4047512092293003e-06,
+      "loss": 0.0065,
+      "step": 1507
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3955951519740186e-06,
+      "loss": 0.0109,
+      "step": 1508
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3864667927035612e-06,
+      "loss": 0.0089,
+      "step": 1509
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.377366160802549e-06,
+      "loss": 0.02,
+      "step": 1510
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3682932855663455e-06,
+      "loss": 0.035,
+      "step": 1511
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.359248196200963e-06,
+      "loss": 0.002,
+      "step": 1512
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3502309218229715e-06,
+      "loss": 0.0024,
+      "step": 1513
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.341241491459404e-06,
+      "loss": 0.0208,
+      "step": 1514
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3322799340476545e-06,
+      "loss": 0.0051,
+      "step": 1515
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3233462784354046e-06,
+      "loss": 0.0038,
+      "step": 1516
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3144405533805138e-06,
+      "loss": 0.0336,
+      "step": 1517
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3055627875509247e-06,
+      "loss": 0.0022,
+      "step": 1518
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2967130095245883e-06,
+      "loss": 0.0018,
+      "step": 1519
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2878912477893569e-06,
+      "loss": 0.0362,
+      "step": 1520
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2790975307428954e-06,
+      "loss": 0.0054,
+      "step": 1521
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2703318866925962e-06,
+      "loss": 0.0153,
+      "step": 1522
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.261594343855479e-06,
+      "loss": 0.0077,
+      "step": 1523
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.252884930358107e-06,
+      "loss": 0.006,
+      "step": 1524
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2442036742364927e-06,
+      "loss": 0.0095,
+      "step": 1525
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.23555060343601e-06,
+      "loss": 0.0054,
+      "step": 1526
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2269257458112983e-06,
+      "loss": 0.0307,
+      "step": 1527
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2183291291261835e-06,
+      "loss": 0.0093,
+      "step": 1528
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2097607810535784e-06,
+      "loss": 0.0019,
+      "step": 1529
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2012207291754008e-06,
+      "loss": 0.0148,
+      "step": 1530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1927090009824793e-06,
+      "loss": 0.0032,
+      "step": 1531
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1842256238744687e-06,
+      "loss": 0.0177,
+      "step": 1532
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.17577062515976e-06,
+      "loss": 0.0073,
+      "step": 1533
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1673440320553941e-06,
+      "loss": 0.0213,
+      "step": 1534
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1589458716869684e-06,
+      "loss": 0.0193,
+      "step": 1535
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1505761710885555e-06,
+      "loss": 0.004,
+      "step": 1536
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1422349572026237e-06,
+      "loss": 0.0023,
+      "step": 1537
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.133922256879927e-06,
+      "loss": 0.0115,
+      "step": 1538
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.125638096879441e-06,
+      "loss": 0.013,
+      "step": 1539
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1173825038682683e-06,
+      "loss": 0.0031,
+      "step": 1540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1091555044215518e-06,
+      "loss": 0.0171,
+      "step": 1541
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1009571250223904e-06,
+      "loss": 0.0017,
+      "step": 1542
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.092787392061756e-06,
+      "loss": 0.016,
+      "step": 1543
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0846463318383983e-06,
+      "loss": 0.0231,
+      "step": 1544
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0765339705587785e-06,
+      "loss": 0.0051,
+      "step": 1545
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0684503343369722e-06,
+      "loss": 0.0083,
+      "step": 1546
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0603954491945801e-06,
+      "loss": 0.0053,
+      "step": 1547
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0523693410606583e-06,
+      "loss": 0.0058,
+      "step": 1548
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0443720357716303e-06,
+      "loss": 0.0094,
+      "step": 1549
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.036403559071195e-06,
+      "loss": 0.0021,
+      "step": 1550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.02846393661026e-06,
+      "loss": 0.002,
+      "step": 1551
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.020553193946836e-06,
+      "loss": 0.0066,
+      "step": 1552
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0126713565459834e-06,
+      "loss": 0.0145,
+      "step": 1553
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0048184497797087e-06,
+      "loss": 0.0043,
+      "step": 1554
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.969944989268843e-07,
+      "loss": 0.0091,
+      "step": 1555
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.89199529173177e-07,
+      "loss": 0.0214,
+      "step": 1556
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.814335656109675e-07,
+      "loss": 0.0069,
+      "step": 1557
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.736966332392538e-07,
+      "loss": 0.0126,
+      "step": 1558
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.659887569635862e-07,
+      "loss": 0.0309,
+      "step": 1559
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.583099615959834e-07,
+      "loss": 0.0041,
+      "step": 1560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.506602718548496e-07,
+      "loss": 0.0031,
+      "step": 1561
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.430397123648949e-07,
+      "loss": 0.0025,
+      "step": 1562
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.354483076570642e-07,
+      "loss": 0.0136,
+      "step": 1563
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.278860821684388e-07,
+      "loss": 0.0124,
+      "step": 1564
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.203530602421862e-07,
+      "loss": 0.02,
+      "step": 1565
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.12849266127458e-07,
+      "loss": 0.0028,
+      "step": 1566
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.053747239793187e-07,
+      "loss": 0.0014,
+      "step": 1567
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.979294578586739e-07,
+      "loss": 0.0212,
+      "step": 1568
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.905134917321845e-07,
+      "loss": 0.0535,
+      "step": 1569
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.831268494721968e-07,
+      "loss": 0.0245,
+      "step": 1570
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.757695548566603e-07,
+      "loss": 0.0214,
+      "step": 1571
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.684416315690502e-07,
+      "loss": 0.005,
+      "step": 1572
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.611431031983009e-07,
+      "loss": 0.0245,
+      "step": 1573
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.538739932387197e-07,
+      "loss": 0.0536,
+      "step": 1574
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.466343250899101e-07,
+      "loss": 0.005,
+      "step": 1575
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.394241220567067e-07,
+      "loss": 0.0059,
+      "step": 1576
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.322434073490949e-07,
+      "loss": 0.0194,
+      "step": 1577
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.250922040821318e-07,
+      "loss": 0.0078,
+      "step": 1578
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.179705352758794e-07,
+      "loss": 0.0166,
+      "step": 1579
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.108784238553235e-07,
+      "loss": 0.0128,
+      "step": 1580
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.03815892650307e-07,
+      "loss": 0.0191,
+      "step": 1581
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.967829643954495e-07,
+      "loss": 0.0067,
+      "step": 1582
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.897796617300813e-07,
+      "loss": 0.0053,
+      "step": 1583
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.828060071981592e-07,
+      "loss": 0.0232,
+      "step": 1584
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.758620232482083e-07,
+      "loss": 0.0487,
+      "step": 1585
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.689477322332395e-07,
+      "loss": 0.0031,
+      "step": 1586
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.620631564106828e-07,
+      "loss": 0.0025,
+      "step": 1587
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.55208317942312e-07,
+      "loss": 0.0229,
+      "step": 1588
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.483832388941748e-07,
+      "loss": 0.0099,
+      "step": 1589
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.415879412365223e-07,
+      "loss": 0.018,
+      "step": 1590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.348224468437404e-07,
+      "loss": 0.0018,
+      "step": 1591
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.280867774942712e-07,
+      "loss": 0.011,
+      "step": 1592
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.213809548705509e-07,
+      "loss": 0.0048,
+      "step": 1593
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.14705000558944e-07,
+      "loss": 0.0226,
+      "step": 1594
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.08058936049657e-07,
+      "loss": 0.0024,
+      "step": 1595
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.014427827366865e-07,
+      "loss": 0.0125,
+      "step": 1596
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.948565619177427e-07,
+      "loss": 0.0089,
+      "step": 1597
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.883002947941808e-07,
+      "loss": 0.0034,
+      "step": 1598
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.81774002470933e-07,
+      "loss": 0.0106,
+      "step": 1599
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.752777059564431e-07,
+      "loss": 0.0049,
+      "step": 1600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.688114261625933e-07,
+      "loss": 0.0081,
+      "step": 1601
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.623751839046455e-07,
+      "loss": 0.0015,
+      "step": 1602
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.559689999011676e-07,
+      "loss": 0.0105,
+      "step": 1603
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.49592894773965e-07,
+      "loss": 0.0109,
+      "step": 1604
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.432468890480226e-07,
+      "loss": 0.0019,
+      "step": 1605
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.369310031514309e-07,
+      "loss": 0.0016,
+      "step": 1606
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.306452574153255e-07,
+      "loss": 0.0036,
+      "step": 1607
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.243896720738196e-07,
+      "loss": 0.0048,
+      "step": 1608
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.181642672639321e-07,
+      "loss": 0.0081,
+      "step": 1609
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.119690630255392e-07,
+      "loss": 0.0046,
+      "step": 1610
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.058040793012965e-07,
+      "loss": 0.006,
+      "step": 1611
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.996693359365724e-07,
+      "loss": 0.0071,
+      "step": 1612
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.935648526793969e-07,
+      "loss": 0.0026,
+      "step": 1613
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.874906491803922e-07,
+      "loss": 0.0136,
+      "step": 1614
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.81446744992702e-07,
+      "loss": 0.0063,
+      "step": 1615
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.754331595719398e-07,
+      "loss": 0.0029,
+      "step": 1616
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.694499122761222e-07,
+      "loss": 0.0328,
+      "step": 1617
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.634970223656023e-07,
+      "loss": 0.0025,
+      "step": 1618
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.575745090030138e-07,
+      "loss": 0.0023,
+      "step": 1619
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.516823912532088e-07,
+      "loss": 0.0133,
+      "step": 1620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.458206880831863e-07,
+      "loss": 0.0069,
+      "step": 1621
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.399894183620513e-07,
+      "loss": 0.0046,
+      "step": 1622
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.341886008609343e-07,
+      "loss": 0.0081,
+      "step": 1623
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.284182542529382e-07,
+      "loss": 0.0031,
+      "step": 1624
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.226783971130823e-07,
+      "loss": 0.0121,
+      "step": 1625
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.169690479182376e-07,
+      "loss": 0.0045,
+      "step": 1626
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.112902250470663e-07,
+      "loss": 0.0036,
+      "step": 1627
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.056419467799711e-07,
+      "loss": 0.0206,
+      "step": 1628
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.0002423129902e-07,
+      "loss": 0.0145,
+      "step": 1629
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.944370966879087e-07,
+      "loss": 0.0076,
+      "step": 1630
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.88880560931887e-07,
+      "loss": 0.0017,
+      "step": 1631
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.833546419177015e-07,
+      "loss": 0.0165,
+      "step": 1632
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.778593574335488e-07,
+      "loss": 0.0027,
+      "step": 1633
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7239472516900864e-07,
+      "loss": 0.0108,
+      "step": 1634
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.669607627149897e-07,
+      "loss": 0.0066,
+      "step": 1635
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.61557487563673e-07,
+      "loss": 0.0092,
+      "step": 1636
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.561849171084565e-07,
+      "loss": 0.0022,
+      "step": 1637
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.508430686438969e-07,
+      "loss": 0.0122,
+      "step": 1638
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.455319593656571e-07,
+      "loss": 0.0286,
+      "step": 1639
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4025160637044874e-07,
+      "loss": 0.0158,
+      "step": 1640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3500202665597403e-07,
+      "loss": 0.0169,
+      "step": 1641
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.297832371208799e-07,
+      "loss": 0.0094,
+      "step": 1642
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.24595254564697e-07,
+      "loss": 0.0037,
+      "step": 1643
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1943809568778416e-07,
+      "loss": 0.0065,
+      "step": 1644
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1431177709127743e-07,
+      "loss": 0.0218,
+      "step": 1645
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.092163152770412e-07,
+      "loss": 0.0062,
+      "step": 1646
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.041517266476058e-07,
+      "loss": 0.0027,
+      "step": 1647
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.991180275061213e-07,
+      "loss": 0.0258,
+      "step": 1648
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.941152340563004e-07,
+      "loss": 0.0037,
+      "step": 1649
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8914336240237217e-07,
+      "loss": 0.0155,
+      "step": 1650
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.842024285490287e-07,
+      "loss": 0.0028,
+      "step": 1651
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.792924484013649e-07,
+      "loss": 0.0072,
+      "step": 1652
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7441343776484116e-07,
+      "loss": 0.0021,
+      "step": 1653
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.695654123452208e-07,
+      "loss": 0.0025,
+      "step": 1654
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6474838774852694e-07,
+      "loss": 0.0086,
+      "step": 1655
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.599623794809881e-07,
+      "loss": 0.0027,
+      "step": 1656
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.552074029489938e-07,
+      "loss": 0.0053,
+      "step": 1657
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5048347345903235e-07,
+      "loss": 0.0059,
+      "step": 1658
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.457906062176597e-07,
+      "loss": 0.0119,
+      "step": 1659
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4112881633143524e-07,
+      "loss": 0.0018,
+      "step": 1660
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3649811880688055e-07,
+      "loss": 0.0124,
+      "step": 1661
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.318985285504284e-07,
+      "loss": 0.0247,
+      "step": 1662
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2733006036837713e-07,
+      "loss": 0.0051,
+      "step": 1663
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.22792728966842e-07,
+      "loss": 0.0194,
+      "step": 1664
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.182865489517073e-07,
+      "loss": 0.0766,
+      "step": 1665
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.138115348285753e-07,
+      "loss": 0.015,
+      "step": 1666
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.093677010027307e-07,
+      "loss": 0.0138,
+      "step": 1667
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.049550617790853e-07,
+      "loss": 0.0105,
+      "step": 1668
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.005736313621288e-07,
+      "loss": 0.015,
+      "step": 1669
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9622342385589256e-07,
+      "loss": 0.0121,
+      "step": 1670
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9190445326390483e-07,
+      "loss": 0.0016,
+      "step": 1671
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8761673348913e-07,
+      "loss": 0.0063,
+      "step": 1672
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8336027833394175e-07,
+      "loss": 0.0113,
+      "step": 1673
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7913510150007095e-07,
+      "loss": 0.0082,
+      "step": 1674
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.74941216588559e-07,
+      "loss": 0.01,
+      "step": 1675
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7077863709972227e-07,
+      "loss": 0.0139,
+      "step": 1676
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.666473764330979e-07,
+      "loss": 0.0173,
+      "step": 1677
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.625474478874079e-07,
+      "loss": 0.0042,
+      "step": 1678
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5847886466051855e-07,
+      "loss": 0.0426,
+      "step": 1679
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.544416398493899e-07,
+      "loss": 0.0246,
+      "step": 1680
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.504357864500373e-07,
+      "loss": 0.0143,
+      "step": 1681
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4646131735749254e-07,
+      "loss": 0.0087,
+      "step": 1682
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.425182453657604e-07,
+      "loss": 0.0106,
+      "step": 1683
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.386065831677742e-07,
+      "loss": 0.0027,
+      "step": 1684
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3472634335535926e-07,
+      "loss": 0.0079,
+      "step": 1685
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.308775384191897e-07,
+      "loss": 0.0059,
+      "step": 1686
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2706018074875046e-07,
+      "loss": 0.0068,
+      "step": 1687
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2327428263229756e-07,
+      "loss": 0.047,
+      "step": 1688
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1951985625681348e-07,
+      "loss": 0.0058,
+      "step": 1689
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1579691370797408e-07,
+      "loss": 0.0028,
+      "step": 1690
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1210546697010726e-07,
+      "loss": 0.0097,
+      "step": 1691
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0844552792615325e-07,
+      "loss": 0.0068,
+      "step": 1692
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0481710835763e-07,
+      "loss": 0.0152,
+      "step": 1693
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0122021994459007e-07,
+      "loss": 0.0043,
+      "step": 1694
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9765487426558705e-07,
+      "loss": 0.0257,
+      "step": 1695
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9412108279763587e-07,
+      "loss": 0.014,
+      "step": 1696
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061885691618043e-07,
+      "loss": 0.0085,
+      "step": 1697
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8714820789504816e-07,
+      "loss": 0.0022,
+      "step": 1698
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8370914690642117e-07,
+      "loss": 0.0223,
+      "step": 1699
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.803016850208028e-07,
+      "loss": 0.028,
+      "step": 1700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.769258332069712e-07,
+      "loss": 0.0063,
+      "step": 1701
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7358160233195143e-07,
+      "loss": 0.0035,
+      "step": 1702
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7026900316098217e-07,
+      "loss": 0.0076,
+      "step": 1703
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.669880463574758e-07,
+      "loss": 0.0105,
+      "step": 1704
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6373874248298726e-07,
+      "loss": 0.0183,
+      "step": 1705
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.605211019971775e-07,
+      "loss": 0.0014,
+      "step": 1706
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5733513525778343e-07,
+      "loss": 0.0274,
+      "step": 1707
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5418085252058456e-07,
+      "loss": 0.0223,
+      "step": 1708
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5105826393936208e-07,
+      "loss": 0.0068,
+      "step": 1709
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4796737956587758e-07,
+      "loss": 0.0053,
+      "step": 1710
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.449082093498344e-07,
+      "loss": 0.003,
+      "step": 1711
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4188076313884414e-07,
+      "loss": 0.0085,
+      "step": 1712
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3888505067839897e-07,
+      "loss": 0.0017,
+      "step": 1713
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3592108161184058e-07,
+      "loss": 0.0179,
+      "step": 1714
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.329888654803213e-07,
+      "loss": 0.0084,
+      "step": 1715
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.300884117227874e-07,
+      "loss": 0.0063,
+      "step": 1716
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2721972967593366e-07,
+      "loss": 0.0077,
+      "step": 1717
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2438282857418327e-07,
+      "loss": 0.0048,
+      "step": 1718
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2157771754965353e-07,
+      "loss": 0.004,
+      "step": 1719
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1880440563213247e-07,
+      "loss": 0.0421,
+      "step": 1720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.160629017490389e-07,
+      "loss": 0.0316,
+      "step": 1721
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1335321472540462e-07,
+      "loss": 0.0233,
+      "step": 1722
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1067535328383672e-07,
+      "loss": 0.0029,
+      "step": 1723
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0802932604450089e-07,
+      "loss": 0.0077,
+      "step": 1724
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.054151415250837e-07,
+      "loss": 0.022,
+      "step": 1725
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0283280814076701e-07,
+      "loss": 0.0019,
+      "step": 1726
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0028233420420141e-07,
+      "loss": 0.0113,
+      "step": 1727
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.776372792548838e-08,
+      "loss": 0.0045,
+      "step": 1728
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.527699741213481e-08,
+      "loss": 0.018,
+      "step": 1729
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.282215066904632e-08,
+      "loss": 0.0072,
+      "step": 1730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.039919559848953e-08,
+      "loss": 0.0479,
+      "step": 1731
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.800814000007207e-08,
+      "loss": 0.0061,
+      "step": 1732
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.56489915707126e-08,
+      "loss": 0.0115,
+      "step": 1733
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.332175790462305e-08,
+      "loss": 0.0032,
+      "step": 1734
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.10264464932764e-08,
+      "loss": 0.0084,
+      "step": 1735
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.876306472539008e-08,
+      "loss": 0.0113,
+      "step": 1736
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.653161988689705e-08,
+      "loss": 0.005,
+      "step": 1737
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.433211916092143e-08,
+      "loss": 0.0047,
+      "step": 1738
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.216456962776064e-08,
+      "loss": 0.021,
+      "step": 1739
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.002897826485889e-08,
+      "loss": 0.044,
+      "step": 1740
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.792535194678485e-08,
+      "loss": 0.0021,
+      "step": 1741
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.585369744521287e-08,
+      "loss": 0.0676,
+      "step": 1742
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.381402142889404e-08,
+      "loss": 0.0058,
+      "step": 1743
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.180633046364181e-08,
+      "loss": 0.0251,
+      "step": 1744
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.983063101231091e-08,
+      "loss": 0.0122,
+      "step": 1745
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7886929434767256e-08,
+      "loss": 0.0218,
+      "step": 1746
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.597523198787924e-08,
+      "loss": 0.0156,
+      "step": 1747
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.409554482549206e-08,
+      "loss": 0.0187,
+      "step": 1748
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2247873998406697e-08,
+      "loss": 0.041,
+      "step": 1749
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.043222545436099e-08,
+      "loss": 0.0418,
+      "step": 1750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8648605038016385e-08,
+      "loss": 0.0027,
+      "step": 1751
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.689701849092898e-08,
+      "loss": 0.0164,
+      "step": 1752
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5177471451538503e-08,
+      "loss": 0.0099,
+      "step": 1753
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.348996945514827e-08,
+      "loss": 0.0085,
+      "step": 1754
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.183451793390747e-08,
+      "loss": 0.0098,
+      "step": 1755
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.021112221679113e-08,
+      "loss": 0.0152,
+      "step": 1756
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.861978752958795e-08,
+      "loss": 0.0053,
+      "step": 1757
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.706051899487806e-08,
+      "loss": 0.011,
+      "step": 1758
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.553332163202083e-08,
+      "loss": 0.0128,
+      "step": 1759
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4038200357137116e-08,
+      "loss": 0.0208,
+      "step": 1760
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.257515998309258e-08,
+      "loss": 0.0022,
+      "step": 1761
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.114420521948325e-08,
+      "loss": 0.0021,
+      "step": 1762
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9745340672620028e-08,
+      "loss": 0.0017,
+      "step": 1763
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8378570845511988e-08,
+      "loss": 0.0136,
+      "step": 1764
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7043900137858624e-08,
+      "loss": 0.0051,
+      "step": 1765
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5741332846026535e-08,
+      "loss": 0.0098,
+      "step": 1766
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4470873163040532e-08,
+      "loss": 0.0013,
+      "step": 1767
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3232525178570332e-08,
+      "loss": 0.004,
+      "step": 1768
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.202629287891722e-08,
+      "loss": 0.005,
+      "step": 1769
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0852180146999635e-08,
+      "loss": 0.0183,
+      "step": 1770
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.971019076234204e-08,
+      "loss": 0.0223,
+      "step": 1771
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.860032840106163e-08,
+      "loss": 0.0105,
+      "step": 1772
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7522596635857202e-08,
+      "loss": 0.0028,
+      "step": 1773
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.64769989360003e-08,
+      "loss": 0.0389,
+      "step": 1774
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5463538667316313e-08,
+      "loss": 0.0254,
+      "step": 1775
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4482219092183392e-08,
+      "loss": 0.0152,
+      "step": 1776
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3533043369514664e-08,
+      "loss": 0.0049,
+      "step": 1777
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2616014554752698e-08,
+      "loss": 0.0087,
+      "step": 1778
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.173113559985506e-08,
+      "loss": 0.0093,
+      "step": 1779
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0878409353289876e-08,
+      "loss": 0.0032,
+      "step": 1780
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0057838560022514e-08,
+      "loss": 0.0171,
+      "step": 1781
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.269425861510029e-09,
+      "loss": 0.0156,
+      "step": 1782
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.513173795687835e-09,
+      "loss": 0.0018,
+      "step": 1783
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.789084796969715e-09,
+      "loss": 0.0017,
+      "step": 1784
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.0971611962300515e-09,
+      "loss": 0.0194,
+      "step": 1785
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.4374052208049375e-09,
+      "loss": 0.0033,
+      "step": 1786
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.8098189944799655e-09,
+      "loss": 0.0115,
+      "step": 1787
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.214404537483564e-09,
+      "loss": 0.0207,
+      "step": 1788
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.651163766484779e-09,
+      "loss": 0.0039,
+      "step": 1789
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.120098494579949e-09,
+      "loss": 0.0126,
+      "step": 1790
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6212104312949304e-09,
+      "loss": 0.005,
+      "step": 1791
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.154501182573988e-09,
+      "loss": 0.0135,
+      "step": 1792
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7199722507742497e-09,
+      "loss": 0.0062,
+      "step": 1793
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3176250346679253e-09,
+      "loss": 0.0278,
+      "step": 1794
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9474608294267618e-09,
+      "loss": 0.0336,
+      "step": 1795
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6094808266275963e-09,
+      "loss": 0.007,
+      "step": 1796
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3036861142456947e-09,
+      "loss": 0.0055,
+      "step": 1797
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0300776766458687e-09,
+      "loss": 0.0196,
+      "step": 1798
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.886563945891379e-10,
+      "loss": 0.0016,
+      "step": 1799
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.794230452205174e-10,
+      "loss": 0.005,
+      "step": 1800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.023783020723482e-10,
+      "loss": 0.0127,
+      "step": 1801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.575227350598564e-10,
+      "loss": 0.0028,
+      "step": 1802
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4485681048004296e-10,
+      "loss": 0.0048,
+      "step": 1803
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.438089100946343e-11,
+      "loss": 0.003,
+      "step": 1804
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6095235705337884e-11,
+      "loss": 0.0152,
+      "step": 1805
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.0127,
+      "step": 1806
+    },
+    {
+      "epoch": 3.0,
+      "step": 1806,
+      "total_flos": 175698642862080.0,
+      "train_loss": 0.16902999525418164,
+      "train_runtime": 12748.0445,
+      "train_samples_per_second": 4.531,
+      "train_steps_per_second": 0.142
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1806,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 5000,
+  "total_flos": 175698642862080.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}