diff --git "a/llava-v1.6-13b-unk-vqa-v1.0/checkpoint-1000/trainer_state.json" "b/llava-v1.6-13b-unk-vqa-v1.0/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/llava-v1.6-13b-unk-vqa-v1.0/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9230769230769231,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25e-07,
+      "loss": 3.1205,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 3.2041,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 3.0942,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 3.1416,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.125e-06,
+      "loss": 3.0353,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 2.4541,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 1.2905,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.9696,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.625e-06,
+      "loss": 0.6378,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4545,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.875e-06,
+      "loss": 0.4719,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.4088,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 0.3443,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.3182,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.2585,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.3624,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0625e-05,
+      "loss": 0.3333,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.125e-05,
+      "loss": 0.3265,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1875e-05,
+      "loss": 0.38,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.25e-05,
+      "loss": 0.253,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3125e-05,
+      "loss": 0.3657,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.375e-05,
+      "loss": 0.309,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4375e-05,
+      "loss": 0.2855,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2678,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.3087,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.2798,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6875e-05,
+      "loss": 0.3017,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.2998,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8125e-05,
+      "loss": 0.2728,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.3714,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9375e-05,
+      "loss": 0.354,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.2786,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999951432210905e-05,
+      "loss": 0.275,
+      "step": 33
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999805729315383e-05,
+      "loss": 0.3081,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999956289272873e-05,
+      "loss": 0.3132,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999922292480975e-05,
+      "loss": 0.2784,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998785828860744e-05,
+      "loss": 0.2323,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998251609127465e-05,
+      "loss": 0.276,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999762027079909e-05,
+      "loss": 0.3328,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996891820008165e-05,
+      "loss": 0.3719,
+      "step": 40
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996066263830533e-05,
+      "loss": 0.3319,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995143610285275e-05,
+      "loss": 0.2987,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994123868334655e-05,
+      "loss": 0.3052,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993007047883988e-05,
+      "loss": 0.2432,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999179315978157e-05,
+      "loss": 0.3362,
+      "step": 45
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999048221581858e-05,
+      "loss": 0.283,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989074228728942e-05,
+      "loss": 0.2764,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987569212189224e-05,
+      "loss": 0.2523,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985967180818493e-05,
+      "loss": 0.2617,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998426815017817e-05,
+      "loss": 0.3292,
+      "step": 50
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998247213677188e-05,
+      "loss": 0.3109,
+      "step": 51
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980579158045322e-05,
+      "loss": 0.262,
+      "step": 52
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978589232386036e-05,
+      "loss": 0.3014,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997650237912329e-05,
+      "loss": 0.2481,
+      "step": 54
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997431861852785e-05,
+      "loss": 0.2872,
+      "step": 55
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972037971811802e-05,
+      "loss": 0.3319,
+      "step": 56
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996966046112834e-05,
+      "loss": 0.2787,
+      "step": 57
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996718610957155e-05,
+      "loss": 0.3093,
+      "step": 58
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964614941176194e-05,
+      "loss": 0.2548,
+      "step": 59
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961946980917457e-05,
+      "loss": 0.3733,
+      "step": 60
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995918225471073e-05,
+      "loss": 0.294,
+      "step": 61
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956320789411338e-05,
+      "loss": 0.2942,
+      "step": 62
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953362612814294e-05,
+      "loss": 0.2796,
+      "step": 63
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950307753654016e-05,
+      "loss": 0.2941,
+      "step": 64
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994715624160405e-05,
+      "loss": 0.3374,
+      "step": 65
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99439081072768e-05,
+      "loss": 0.2752,
+      "step": 66
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940563382223196e-05,
+      "loss": 0.3242,
+      "step": 67
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 0.306,
+      "step": 68
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9933584290831593e-05,
+      "loss": 0.3409,
+      "step": 69
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9929949992285397e-05,
+      "loss": 0.26,
+      "step": 70
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992621923859581e-05,
+      "loss": 1.007,
+      "step": 71
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9922392066001724e-05,
+      "loss": 0.2694,
+      "step": 72
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99184685116786e-05,
+      "loss": 0.2936,
+      "step": 73
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914448613738107e-05,
+      "loss": 0.3006,
+      "step": 74
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991033241122776e-05,
+      "loss": 0.3055,
+      "step": 75
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9906119944130527e-05,
+      "loss": 0.3066,
+      "step": 76
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9901811253364458e-05,
+      "loss": 0.3078,
+      "step": 77
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897406380782262e-05,
+      "loss": 0.2577,
+      "step": 78
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989290536917093e-05,
+      "loss": 0.3208,
+      "step": 79
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 0.2781,
+      "step": 80
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988361510467761e-05,
+      "loss": 0.2654,
+      "step": 81
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878825942037147e-05,
+      "loss": 0.2726,
+      "step": 82
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9873940820849714e-05,
+      "loss": 0.3371,
+      "step": 83
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9868959788567213e-05,
+      "loss": 0.3278,
+      "step": 84
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9863882893573188e-05,
+      "loss": 0.3134,
+      "step": 85
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985871018518236e-05,
+      "loss": 0.3313,
+      "step": 86
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9853441713640123e-05,
+      "loss": 0.3215,
+      "step": 87
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.3173,
+      "step": 88
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9842617686733546e-05,
+      "loss": 0.2663,
+      "step": 89
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9837062236509013e-05,
+      "loss": 0.2583,
+      "step": 90
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983141123341168e-05,
+      "loss": 0.3422,
+      "step": 91
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9825664732332886e-05,
+      "loss": 0.2444,
+      "step": 92
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9819822789091597e-05,
+      "loss": 0.2466,
+      "step": 93
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981388546043388e-05,
+      "loss": 0.306,
+      "step": 94
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 0.3069,
+      "step": 95
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9801724878485438e-05,
+      "loss": 0.3436,
+      "step": 96
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979550174331724e-05,
+      "loss": 0.251,
+      "step": 97
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9789183458976485e-05,
+      "loss": 0.249,
+      "step": 98
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97827700868362e-05,
+      "loss": 0.3166,
+      "step": 99
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977626168919305e-05,
+      "loss": 0.2768,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9769658329266718e-05,
+      "loss": 0.3419,
+      "step": 101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9762960071199334e-05,
+      "loss": 0.2657,
+      "step": 102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9756166980054812e-05,
+      "loss": 0.2507,
+      "step": 103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.3192,
+      "step": 104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9742296563395218e-05,
+      "loss": 0.2493,
+      "step": 105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9735219372611232e-05,
+      "loss": 0.3267,
+      "step": 106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9728047618210995e-05,
+      "loss": 0.3166,
+      "step": 107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9720781369857747e-05,
+      "loss": 0.3536,
+      "step": 108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9713420698132614e-05,
+      "loss": 0.2781,
+      "step": 109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970596567453391e-05,
+      "loss": 0.2651,
+      "step": 110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9698416371476434e-05,
+      "loss": 0.3015,
+      "step": 111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.969077286229078e-05,
+      "loss": 0.2732,
+      "step": 112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9683035221222617e-05,
+      "loss": 0.3363,
+      "step": 113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9675203523431964e-05,
+      "loss": 0.3117,
+      "step": 114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9667277844992476e-05,
+      "loss": 0.2747,
+      "step": 115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.2725,
+      "step": 116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9651144855025265e-05,
+      "loss": 0.2411,
+      "step": 117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964293770020628e-05,
+      "loss": 0.3084,
+      "step": 118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9634636878154393e-05,
+      "loss": 0.3256,
+      "step": 119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962624246950012e-05,
+      "loss": 0.2898,
+      "step": 120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9617754555783045e-05,
+      "loss": 0.2852,
+      "step": 121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9609173219450998e-05,
+      "loss": 0.2666,
+      "step": 122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960049854385929e-05,
+      "loss": 0.3299,
+      "step": 123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9591730613269878e-05,
+      "loss": 0.3056,
+      "step": 124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9582869512850576e-05,
+      "loss": 0.2957,
+      "step": 125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957391532867418e-05,
+      "loss": 0.3051,
+      "step": 126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956486814771769e-05,
+      "loss": 0.28,
+      "step": 127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 0.2352,
+      "step": 128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9546495147888134e-05,
+      "loss": 0.3698,
+      "step": 129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953716950748227e-05,
+      "loss": 0.2826,
+      "step": 130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9527751227228964e-05,
+      "loss": 0.2779,
+      "step": 131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9518240398613226e-05,
+      "loss": 0.2996,
+      "step": 132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9508637114019037e-05,
+      "loss": 0.3028,
+      "step": 133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9498941466728462e-05,
+      "loss": 0.2569,
+      "step": 134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9489153550920726e-05,
+      "loss": 0.3437,
+      "step": 135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947927346167132e-05,
+      "loss": 0.2831,
+      "step": 136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946930129495106e-05,
+      "loss": 0.2383,
+      "step": 137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945923714762516e-05,
+      "loss": 0.3219,
+      "step": 138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9449081117452304e-05,
+      "loss": 0.2474,
+      "step": 139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 0.2878,
+      "step": 140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9428493804062013e-05,
+      "loss": 0.3193,
+      "step": 141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9418062720820636e-05,
+      "loss": 0.2517,
+      "step": 142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9407540154682473e-05,
+      "loss": 0.2633,
+      "step": 143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.3241,
+      "step": 144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9386220983449652e-05,
+      "loss": 0.2682,
+      "step": 145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9375424585439994e-05,
+      "loss": 0.2289,
+      "step": 146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9364537118701542e-05,
+      "loss": 0.2688,
+      "step": 147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935355868899034e-05,
+      "loss": 0.3555,
+      "step": 148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9342489402945997e-05,
+      "loss": 0.2492,
+      "step": 149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9331329368090664e-05,
+      "loss": 0.2575,
+      "step": 150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932007869282799e-05,
+      "loss": 0.3059,
+      "step": 151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9308737486442045e-05,
+      "loss": 0.2851,
+      "step": 152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9297305859096305e-05,
+      "loss": 0.2762,
+      "step": 153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9285783921832537e-05,
+      "loss": 0.2746,
+      "step": 154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927417178656975e-05,
+      "loss": 0.2788,
+      "step": 155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926246956610309e-05,
+      "loss": 0.3524,
+      "step": 156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9250677374102752e-05,
+      "loss": 0.2743,
+      "step": 157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.3199,
+      "step": 158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9226823534550418e-05,
+      "loss": 0.3208,
+      "step": 159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921476211870408e-05,
+      "loss": 0.3132,
+      "step": 160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9202611194733107e-05,
+      "loss": 0.3034,
+      "step": 161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9190370880666206e-05,
+      "loss": 0.2826,
+      "step": 162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9178041295400383e-05,
+      "loss": 0.308,
+      "step": 163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9165622558699763e-05,
+      "loss": 0.2221,
+      "step": 164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9153114791194475e-05,
+      "loss": 0.2341,
+      "step": 165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9140518114379433e-05,
+      "loss": 0.2754,
+      "step": 166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912783265061319e-05,
+      "loss": 0.3542,
+      "step": 167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9115058523116734e-05,
+      "loss": 0.2733,
+      "step": 168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9102195855972287e-05,
+      "loss": 0.2774,
+      "step": 169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908924477412211e-05,
+      "loss": 0.2869,
+      "step": 170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9076205403367287e-05,
+      "loss": 0.2277,
+      "step": 171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9063077870366504e-05,
+      "loss": 0.2445,
+      "step": 172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.90498623026348e-05,
+      "loss": 0.3144,
+      "step": 173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903655882854237e-05,
+      "loss": 0.2575,
+      "step": 174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9023167577313267e-05,
+      "loss": 0.3029,
+      "step": 175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.2996,
+      "step": 176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8996122264603202e-05,
+      "loss": 0.2593,
+      "step": 177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898246846582844e-05,
+      "loss": 0.2958,
+      "step": 178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8968727415326885e-05,
+      "loss": 0.3741,
+      "step": 179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.895489924657301e-05,
+      "loss": 0.3453,
+      "step": 180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894098409388754e-05,
+      "loss": 0.3498,
+      "step": 181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8926982092436117e-05,
+      "loss": 0.2838,
+      "step": 182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8912893378227984e-05,
+      "loss": 0.2185,
+      "step": 183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8898718088114688e-05,
+      "loss": 0.2758,
+      "step": 184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8884456359788725e-05,
+      "loss": 0.2896,
+      "step": 185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.887010833178222e-05,
+      "loss": 0.2704,
+      "step": 186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8855674143465567e-05,
+      "loss": 0.3143,
+      "step": 187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8841153935046098e-05,
+      "loss": 0.2777,
+      "step": 188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8826547847566692e-05,
+      "loss": 0.347,
+      "step": 189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8811856022904423e-05,
+      "loss": 0.3351,
+      "step": 190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8797078603769184e-05,
+      "loss": 0.2882,
+      "step": 191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8782215733702286e-05,
+      "loss": 0.3131,
+      "step": 192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876726755707508e-05,
+      "loss": 0.271,
+      "step": 193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8752234219087538e-05,
+      "loss": 0.3061,
+      "step": 194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8737115865766865e-05,
+      "loss": 0.2788,
+      "step": 195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8721912643966055e-05,
+      "loss": 0.3215,
+      "step": 196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8706624701362485e-05,
+      "loss": 0.2239,
+      "step": 197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8691252186456465e-05,
+      "loss": 0.2239,
+      "step": 198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8675795248569816e-05,
+      "loss": 0.2771,
+      "step": 199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.2776,
+      "step": 200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8644628705240636e-05,
+      "loss": 0.3079,
+      "step": 201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862891940253613e-05,
+      "loss": 0.2759,
+      "step": 202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8613126282324092e-05,
+      "loss": 0.271,
+      "step": 203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8597249498011906e-05,
+      "loss": 0.298,
+      "step": 204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.858128920381963e-05,
+      "loss": 0.3204,
+      "step": 205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8565245554778516e-05,
+      "loss": 0.2648,
+      "step": 206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854911870672947e-05,
+      "loss": 0.2512,
+      "step": 207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8532908816321557e-05,
+      "loss": 0.2963,
+      "step": 208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8516616041010495e-05,
+      "loss": 0.249,
+      "step": 209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8500240539057093e-05,
+      "loss": 0.2876,
+      "step": 210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.848378246952574e-05,
+      "loss": 0.2451,
+      "step": 211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 0.2507,
+      "step": 212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8450619267995283e-05,
+      "loss": 0.3187,
+      "step": 213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.843391445812886e-05,
+      "loss": 0.2954,
+      "step": 214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84171277249467e-05,
+      "loss": 0.304,
+      "step": 215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8400259231507716e-05,
+      "loss": 0.3065,
+      "step": 216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8383309141664992e-05,
+      "loss": 0.2334,
+      "step": 217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83662776200642e-05,
+      "loss": 0.3215,
+      "step": 218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8349164832142015e-05,
+      "loss": 0.2169,
+      "step": 219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833197094412449e-05,
+      "loss": 0.2751,
+      "step": 220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 0.2806,
+      "step": 221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8297340536644877e-05,
+      "loss": 0.2745,
+      "step": 222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.827990435356725e-05,
+      "loss": 0.2909,
+      "step": 223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.3213,
+      "step": 224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8244790875571582e-05,
+      "loss": 0.3003,
+      "step": 225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8227113921730336e-05,
+      "loss": 0.262,
+      "step": 226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8209357053342325e-05,
+      "loss": 0.2862,
+      "step": 227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819152044288992e-05,
+      "loss": 0.3163,
+      "step": 228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8173604263630066e-05,
+      "loss": 0.2482,
+      "step": 229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8155608689592604e-05,
+      "loss": 0.2723,
+      "step": 230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8137533895578585e-05,
+      "loss": 0.2783,
+      "step": 231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811938005715857e-05,
+      "loss": 0.2566,
+      "step": 232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8101147350670905e-05,
+      "loss": 0.2433,
+      "step": 233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8082835953220055e-05,
+      "loss": 0.26,
+      "step": 234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.806444604267483e-05,
+      "loss": 0.2293,
+      "step": 235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8045977797666685e-05,
+      "loss": 0.2662,
+      "step": 236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8027431397587993e-05,
+      "loss": 0.2926,
+      "step": 237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8008807022590283e-05,
+      "loss": 0.2753,
+      "step": 238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7990104853582494e-05,
+      "loss": 0.2503,
+      "step": 239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7971325072229227e-05,
+      "loss": 0.257,
+      "step": 240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7952467860948975e-05,
+      "loss": 0.235,
+      "step": 241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7933533402912354e-05,
+      "loss": 0.2411,
+      "step": 242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.791452188204031e-05,
+      "loss": 0.3642,
+      "step": 243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7895433483002356e-05,
+      "loss": 0.312,
+      "step": 244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7876268391214756e-05,
+      "loss": 0.2922,
+      "step": 245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785702679283874e-05,
+      "loss": 0.281,
+      "step": 246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7837708874778683e-05,
+      "loss": 0.2728,
+      "step": 247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.3026,
+      "step": 248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7798844830928818e-05,
+      "loss": 0.2507,
+      "step": 249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777929908264715e-05,
+      "loss": 0.2845,
+      "step": 250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.775967776969405e-05,
+      "loss": 0.2366,
+      "step": 251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7739981082662275e-05,
+      "loss": 0.2627,
+      "step": 252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.772020921287674e-05,
+      "loss": 0.2913,
+      "step": 253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7700362352392632e-05,
+      "loss": 0.2583,
+      "step": 254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7680440693993586e-05,
+      "loss": 0.2789,
+      "step": 255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.2587,
+      "step": 256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7640373758216075e-05,
+      "loss": 0.3167,
+      "step": 257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.762022887003011e-05,
+      "loss": 0.3075,
+      "step": 258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7600009962310417e-05,
+      "loss": 0.2624,
+      "step": 259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757971723145453e-05,
+      "loss": 0.3145,
+      "step": 260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7559350874577066e-05,
+      "loss": 0.2992,
+      "step": 261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.75389110895078e-05,
+      "loss": 0.2608,
+      "step": 262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7518398074789776e-05,
+      "loss": 0.3234,
+      "step": 263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7497812029677344e-05,
+      "loss": 0.3116,
+      "step": 264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7477153154134244e-05,
+      "loss": 0.2946,
+      "step": 265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7456421648831658e-05,
+      "loss": 0.2553,
+      "step": 266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.743561771514626e-05,
+      "loss": 0.3001,
+      "step": 267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.741474155515827e-05,
+      "loss": 0.3131,
+      "step": 268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.739379337164946e-05,
+      "loss": 0.2931,
+      "step": 269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.737277336810124e-05,
+      "loss": 0.3154,
+      "step": 270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7351681748692622e-05,
+      "loss": 0.3235,
+      "step": 271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.2721,
+      "step": 272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7309284482486494e-05,
+      "loss": 0.253,
+      "step": 273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7287979247517285e-05,
+      "loss": 0.2777,
+      "step": 274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7266603220340273e-05,
+      "loss": 0.3038,
+      "step": 275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7245156608592727e-05,
+      "loss": 0.3049,
+      "step": 276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7223639620597556e-05,
+      "loss": 0.3453,
+      "step": 277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7202052465361268e-05,
+      "loss": 0.2584,
+      "step": 278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.718039535257194e-05,
+      "loss": 0.3027,
+      "step": 279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7158668492597186e-05,
+      "loss": 0.2321,
+      "step": 280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7136872096482123e-05,
+      "loss": 0.238,
+      "step": 281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7115006375947304e-05,
+      "loss": 0.3371,
+      "step": 282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7093071543386667e-05,
+      "loss": 0.3646,
+      "step": 283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.309,
+      "step": 284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7048995395118253e-05,
+      "loss": 0.3419,
+      "step": 285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7026854507546694e-05,
+      "loss": 0.3216,
+      "step": 286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7004645364217584e-05,
+      "loss": 0.2727,
+      "step": 287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698236818086073e-05,
+      "loss": 0.2941,
+      "step": 288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6960023173866834e-05,
+      "loss": 0.2508,
+      "step": 289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693761056028542e-05,
+      "loss": 0.2225,
+      "step": 290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6915130557822698e-05,
+      "loss": 0.3133,
+      "step": 291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.689258338483947e-05,
+      "loss": 0.2674,
+      "step": 292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686996926034902e-05,
+      "loss": 0.2953,
+      "step": 293
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6847288404014937e-05,
+      "loss": 0.3301,
+      "step": 294
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.682454103614904e-05,
+      "loss": 0.2626,
+      "step": 295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6801727377709195e-05,
+      "loss": 0.3068,
+      "step": 296
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.67788476502972e-05,
+      "loss": 0.2985,
+      "step": 297
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6755902076156606e-05,
+      "loss": 0.2519,
+      "step": 298
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6732890878170573e-05,
+      "loss": 0.2772,
+      "step": 299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.67098142798597e-05,
+      "loss": 0.2787,
+      "step": 300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.668667250537987e-05,
+      "loss": 0.3226,
+      "step": 301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6663465779520042e-05,
+      "loss": 0.2774,
+      "step": 302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6640194327700087e-05,
+      "loss": 0.3848,
+      "step": 303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6616858375968596e-05,
+      "loss": 0.2963,
+      "step": 304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659345815100069e-05,
+      "loss": 0.2928,
+      "step": 305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6569993880095807e-05,
+      "loss": 0.2956,
+      "step": 306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6546465791175498e-05,
+      "loss": 0.259,
+      "step": 307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6522874112781213e-05,
+      "loss": 0.2371,
+      "step": 308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6499219074072087e-05,
+      "loss": 0.2935,
+      "step": 309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6475500904822707e-05,
+      "loss": 0.2653,
+      "step": 310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.645171983542088e-05,
+      "loss": 0.3192,
+      "step": 311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.3067,
+      "step": 312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.640396992076379e-05,
+      "loss": 0.2214,
+      "step": 313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6380001539330088e-05,
+      "loss": 0.2741,
+      "step": 314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6355971185382547e-05,
+      "loss": 0.2275,
+      "step": 315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6331879092341402e-05,
+      "loss": 0.3036,
+      "step": 316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6307725494226586e-05,
+      "loss": 0.2903,
+      "step": 317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6283510625655474e-05,
+      "loss": 0.2693,
+      "step": 318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6259234721840595e-05,
+      "loss": 0.2829,
+      "step": 319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.2574,
+      "step": 320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6210500752291682e-05,
+      "loss": 0.3058,
+      "step": 321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6186043159937884e-05,
+      "loss": 0.3361,
+      "step": 322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.616152547909618e-05,
+      "loss": 0.3129,
+      "step": 323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6136947947920477e-05,
+      "loss": 0.283,
+      "step": 324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.611231080514605e-05,
+      "loss": 0.2761,
+      "step": 325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608761429008721e-05,
+      "loss": 0.2744,
+      "step": 326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.606285864263498e-05,
+      "loss": 0.2499,
+      "step": 327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6038044103254775e-05,
+      "loss": 0.3227,
+      "step": 328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.601317091298406e-05,
+      "loss": 0.3156,
+      "step": 329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5988239313430004e-05,
+      "loss": 0.368,
+      "step": 330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5963249546767144e-05,
+      "loss": 0.2803,
+      "step": 331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5938201855735017e-05,
+      "loss": 0.2881,
+      "step": 332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5913096483635827e-05,
+      "loss": 0.2658,
+      "step": 333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5887933674332048e-05,
+      "loss": 0.303,
+      "step": 334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5862713672244092e-05,
+      "loss": 0.2942,
+      "step": 335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5837436722347902e-05,
+      "loss": 0.3167,
+      "step": 336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5812103070172592e-05,
+      "loss": 0.2853,
+      "step": 337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.578671296179806e-05,
+      "loss": 0.3242,
+      "step": 338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5761266643852587e-05,
+      "loss": 0.2558,
+      "step": 339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.573576436351046e-05,
+      "loss": 0.2507,
+      "step": 340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5710206368489555e-05,
+      "loss": 0.272,
+      "step": 341
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5684592907048925e-05,
+      "loss": 0.305,
+      "step": 342
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5658924227986415e-05,
+      "loss": 0.2581,
+      "step": 343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.563320058063622e-05,
+      "loss": 0.2313,
+      "step": 344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.560742221486648e-05,
+      "loss": 0.2633,
+      "step": 345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5581589381076843e-05,
+      "loss": 0.2827,
+      "step": 346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.2869,
+      "step": 347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5529761313679396e-05,
+      "loss": 0.2416,
+      "step": 348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5503766583506522e-05,
+      "loss": 0.3046,
+      "step": 349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5477718392178716e-05,
+      "loss": 0.2946,
+      "step": 350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.545161699271659e-05,
+      "loss": 0.299,
+      "step": 351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5425462638657597e-05,
+      "loss": 0.2842,
+      "step": 352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5399255584053568e-05,
+      "loss": 0.2632,
+      "step": 353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5372996083468242e-05,
+      "loss": 0.2635,
+      "step": 354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5346684391974792e-05,
+      "loss": 0.2951,
+      "step": 355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.3626,
+      "step": 356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.529390545908857e-05,
+      "loss": 0.2715,
+      "step": 357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.526743873036701e-05,
+      "loss": 0.306,
+      "step": 358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5240920836074777e-05,
+      "loss": 0.2744,
+      "step": 359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5214352033794981e-05,
+      "loss": 0.2504,
+      "step": 360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5187732581605217e-05,
+      "loss": 0.2561,
+      "step": 361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5161062738075068e-05,
+      "loss": 0.29,
+      "step": 362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5134342762263606e-05,
+      "loss": 0.3179,
+      "step": 363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5107572913716859e-05,
+      "loss": 0.2706,
+      "step": 364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5080753452465296e-05,
+      "loss": 0.2969,
+      "step": 365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.505388463902131e-05,
+      "loss": 0.3527,
+      "step": 366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.502696673437667e-05,
+      "loss": 0.2439,
+      "step": 367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.3277,
+      "step": 368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4972984697834238e-05,
+      "loss": 0.2584,
+      "step": 369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4945921090294076e-05,
+      "loss": 0.2403,
+      "step": 370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4918809440263435e-05,
+      "loss": 0.2998,
+      "step": 371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4891650011092896e-05,
+      "loss": 0.3289,
+      "step": 372
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.486444306659714e-05,
+      "loss": 0.2605,
+      "step": 373
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4837188871052399e-05,
+      "loss": 0.2751,
+      "step": 374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4809887689193878e-05,
+      "loss": 0.2747,
+      "step": 375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4782539786213184e-05,
+      "loss": 0.2412,
+      "step": 376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4755145427755755e-05,
+      "loss": 0.248,
+      "step": 377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4727704879918272e-05,
+      "loss": 0.2754,
+      "step": 378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4700218409246087e-05,
+      "loss": 0.3088,
+      "step": 379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4672686282730622e-05,
+      "loss": 0.2608,
+      "step": 380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4645108767806778e-05,
+      "loss": 0.3526,
+      "step": 381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4617486132350343e-05,
+      "loss": 0.3136,
+      "step": 382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4589818644675378e-05,
+      "loss": 0.2658,
+      "step": 383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4562106573531632e-05,
+      "loss": 0.2535,
+      "step": 384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4534350188101905e-05,
+      "loss": 0.309,
+      "step": 385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4506549757999456e-05,
+      "loss": 0.2513,
+      "step": 386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4478705553265363e-05,
+      "loss": 0.2805,
+      "step": 387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4450817844365924e-05,
+      "loss": 0.2679,
+      "step": 388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4422886902190014e-05,
+      "loss": 0.2682,
+      "step": 389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4394912998046451e-05,
+      "loss": 0.2175,
+      "step": 390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.436689640366137e-05,
+      "loss": 0.2748,
+      "step": 391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.289,
+      "step": 392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4310736233141926e-05,
+      "loss": 0.3036,
+      "step": 393
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4282593202522627e-05,
+      "loss": 0.2636,
+      "step": 394
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4254408572686642e-05,
+      "loss": 0.2123,
+      "step": 395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4226182617406996e-05,
+      "loss": 0.2737,
+      "step": 396
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4197915610858143e-05,
+      "loss": 0.2761,
+      "step": 397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4169607827613284e-05,
+      "loss": 0.3161,
+      "step": 398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4141259542641706e-05,
+      "loss": 0.3019,
+      "step": 399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4112871031306118e-05,
+      "loss": 0.3382,
+      "step": 400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4084442569359964e-05,
+      "loss": 0.3034,
+      "step": 401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4055974432944753e-05,
+      "loss": 0.3254,
+      "step": 402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4027466898587375e-05,
+      "loss": 0.2363,
+      "step": 403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3998920243197408e-05,
+      "loss": 0.2609,
+      "step": 404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3970334744064451e-05,
+      "loss": 0.282,
+      "step": 405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3941710678855396e-05,
+      "loss": 0.2528,
+      "step": 406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.391304832561175e-05,
+      "loss": 0.2296,
+      "step": 407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3884347962746949e-05,
+      "loss": 0.4148,
+      "step": 408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3855609869043618e-05,
+      "loss": 0.2966,
+      "step": 409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.3317,
+      "step": 410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3798021606081713e-05,
+      "loss": 0.2712,
+      "step": 411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3769171996210053e-05,
+      "loss": 0.3334,
+      "step": 412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3740285774268282e-05,
+      "loss": 0.2987,
+      "step": 413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.371136322084438e-05,
+      "loss": 0.2348,
+      "step": 414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3682404616879246e-05,
+      "loss": 0.2654,
+      "step": 415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.3012,
+      "step": 416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3624380382837017e-05,
+      "loss": 0.3319,
+      "step": 417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3595315316381676e-05,
+      "loss": 0.2922,
+      "step": 418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3566215326623131e-05,
+      "loss": 0.3088,
+      "step": 419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3537080696225815e-05,
+      "loss": 0.2753,
+      "step": 420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3507911708190646e-05,
+      "loss": 0.2246,
+      "step": 421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3478708645852272e-05,
+      "loss": 0.2534,
+      "step": 422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3449471792876333e-05,
+      "loss": 0.3303,
+      "step": 423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.2282,
+      "step": 424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3390897851312667e-05,
+      "loss": 0.2453,
+      "step": 425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.336156133168631e-05,
+      "loss": 0.35,
+      "step": 426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3332192159339595e-05,
+      "loss": 0.3439,
+      "step": 427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 0.3345,
+      "step": 428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3273356997916106e-05,
+      "loss": 0.2997,
+      "step": 429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3243891580338074e-05,
+      "loss": 0.2903,
+      "step": 430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3214394653031616e-05,
+      "loss": 0.2787,
+      "step": 431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3184866502516846e-05,
+      "loss": 0.2297,
+      "step": 432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3155307415617156e-05,
+      "loss": 0.2647,
+      "step": 433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3125717679456447e-05,
+      "loss": 0.2671,
+      "step": 434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309609758145633e-05,
+      "loss": 0.2983,
+      "step": 435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3066447409333345e-05,
+      "loss": 0.2799,
+      "step": 436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3036767451096148e-05,
+      "loss": 0.2767,
+      "step": 437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.300705799504273e-05,
+      "loss": 0.2439,
+      "step": 438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2977319329757616e-05,
+      "loss": 0.322,
+      "step": 439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2947551744109044e-05,
+      "loss": 0.2731,
+      "step": 440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2917755527246179e-05,
+      "loss": 0.2215,
+      "step": 441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.28879309685963e-05,
+      "loss": 0.2944,
+      "step": 442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2858078357861979e-05,
+      "loss": 0.2609,
+      "step": 443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2828197985018276e-05,
+      "loss": 0.2675,
+      "step": 444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2798290140309924e-05,
+      "loss": 0.3018,
+      "step": 445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2768355114248493e-05,
+      "loss": 0.2931,
+      "step": 446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2738393197609602e-05,
+      "loss": 0.3087,
+      "step": 447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2708404681430054e-05,
+      "loss": 0.2885,
+      "step": 448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2678389857005033e-05,
+      "loss": 0.2766,
+      "step": 449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2648349015885272e-05,
+      "loss": 0.2884,
+      "step": 450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2618282449874221e-05,
+      "loss": 0.3036,
+      "step": 451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.2656,
+      "step": 452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2558073311638604e-05,
+      "loss": 0.2637,
+      "step": 453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2527931324258975e-05,
+      "loss": 0.2751,
+      "step": 454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.249776478167227e-05,
+      "loss": 0.2935,
+      "step": 455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2467573976902936e-05,
+      "loss": 0.3568,
+      "step": 456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2437359203211109e-05,
+      "loss": 0.2683,
+      "step": 457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2407120754089733e-05,
+      "loss": 0.3217,
+      "step": 458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2376858923261732e-05,
+      "loss": 0.2843,
+      "step": 459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2346574004677154e-05,
+      "loss": 0.2699,
+      "step": 460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2316266292510305e-05,
+      "loss": 0.2738,
+      "step": 461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2285936081156897e-05,
+      "loss": 0.3254,
+      "step": 462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2255583665231196e-05,
+      "loss": 0.2704,
+      "step": 463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.3453,
+      "step": 464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2194813399195518e-05,
+      "loss": 0.2691,
+      "step": 465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2164396139381029e-05,
+      "loss": 0.31,
+      "step": 466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2133957855579501e-05,
+      "loss": 0.2453,
+      "step": 467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.210349884345496e-05,
+      "loss": 0.3133,
+      "step": 468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2073019398872778e-05,
+      "loss": 0.3025,
+      "step": 469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2042519817896805e-05,
+      "loss": 0.2544,
+      "step": 470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2012000396786485e-05,
+      "loss": 0.2853,
+      "step": 471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1981461431993978e-05,
+      "loss": 0.2783,
+      "step": 472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.2279,
+      "step": 473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1920326058117364e-05,
+      "loss": 0.2778,
+      "step": 474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1889730242875243e-05,
+      "loss": 0.2748,
+      "step": 475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1859116071629148e-05,
+      "loss": 0.2916,
+      "step": 476
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1828483841751597e-05,
+      "loss": 0.3167,
+      "step": 477
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1797833850790527e-05,
+      "loss": 0.306,
+      "step": 478
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1767166396466404e-05,
+      "loss": 0.2927,
+      "step": 479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.2514,
+      "step": 480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1705780289456069e-05,
+      "loss": 0.2871,
+      "step": 481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1675062233047365e-05,
+      "loss": 0.2858,
+      "step": 482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1644327905824808e-05,
+      "loss": 0.2593,
+      "step": 483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1613577606328068e-05,
+      "loss": 0.3254,
+      "step": 484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1582811633251949e-05,
+      "loss": 0.2778,
+      "step": 485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1552030285443516e-05,
+      "loss": 0.2236,
+      "step": 486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1521233861899168e-05,
+      "loss": 0.2434,
+      "step": 487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1490422661761744e-05,
+      "loss": 0.2188,
+      "step": 488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1459596984317622e-05,
+      "loss": 0.2963,
+      "step": 489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1428757128993801e-05,
+      "loss": 0.2587,
+      "step": 490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1397903395354996e-05,
+      "loss": 0.2658,
+      "step": 491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1367036083100735e-05,
+      "loss": 0.2833,
+      "step": 492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1336155492062439e-05,
+      "loss": 0.2941,
+      "step": 493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.130526192220052e-05,
+      "loss": 0.2516,
+      "step": 494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1274355673601446e-05,
+      "loss": 0.2299,
+      "step": 495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1243437046474854e-05,
+      "loss": 0.277,
+      "step": 496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1212506341150615e-05,
+      "loss": 0.3144,
+      "step": 497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.118156385807593e-05,
+      "loss": 0.3107,
+      "step": 498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1150609897812387e-05,
+      "loss": 0.2422,
+      "step": 499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 0.221,
+      "step": 500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1088668748519646e-05,
+      "loss": 0.291,
+      "step": 501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.105768216115938e-05,
+      "loss": 0.2208,
+      "step": 502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1026685299942286e-05,
+      "loss": 0.2631,
+      "step": 503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0995678465958168e-05,
+      "loss": 0.2549,
+      "step": 504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0964661960393703e-05,
+      "loss": 0.2731,
+      "step": 505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0933636084529507e-05,
+      "loss": 0.2862,
+      "step": 506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0902601139737225e-05,
+      "loss": 0.2651,
+      "step": 507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0871557427476585e-05,
+      "loss": 0.2657,
+      "step": 508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0840505249292477e-05,
+      "loss": 0.2755,
+      "step": 509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0809444906812034e-05,
+      "loss": 0.2464,
+      "step": 510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0778376701741688e-05,
+      "loss": 0.2556,
+      "step": 511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 0.2953,
+      "step": 512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0716217911035952e-05,
+      "loss": 0.2458,
+      "step": 513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0685127929183567e-05,
+      "loss": 0.3125,
+      "step": 514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0654031292301432e-05,
+      "loss": 0.306,
+      "step": 515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0622928302448523e-05,
+      "loss": 0.2939,
+      "step": 516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0591819261745528e-05,
+      "loss": 0.3073,
+      "step": 517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0560704472371919e-05,
+      "loss": 0.2334,
+      "step": 518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0529584236562995e-05,
+      "loss": 0.2722,
+      "step": 519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0498458856606972e-05,
+      "loss": 0.2811,
+      "step": 520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0467328634842024e-05,
+      "loss": 0.2257,
+      "step": 521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0436193873653362e-05,
+      "loss": 0.212,
+      "step": 522
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0405054875470287e-05,
+      "loss": 0.2155,
+      "step": 523
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.037391194276326e-05,
+      "loss": 0.26,
+      "step": 524
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0342765378040953e-05,
+      "loss": 0.2031,
+      "step": 525
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0311615483847333e-05,
+      "loss": 0.1839,
+      "step": 526
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.028046256275869e-05,
+      "loss": 0.2833,
+      "step": 527
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0249306917380731e-05,
+      "loss": 0.2025,
+      "step": 528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0218148850345613e-05,
+      "loss": 0.227,
+      "step": 529
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0186988664309023e-05,
+      "loss": 0.1985,
+      "step": 530
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0155826661947232e-05,
+      "loss": 0.1746,
+      "step": 531
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0124663145954152e-05,
+      "loss": 0.2172,
+      "step": 532
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0093498419038394e-05,
+      "loss": 0.2291,
+      "step": 533
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0062332783920337e-05,
+      "loss": 0.2097,
+      "step": 534
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0031166543329179e-05,
+      "loss": 0.2216,
+      "step": 535
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.2027,
+      "step": 536
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.968833456670824e-06,
+      "loss": 0.2203,
+      "step": 537
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.937667216079665e-06,
+      "loss": 0.223,
+      "step": 538
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.90650158096161e-06,
+      "loss": 0.191,
+      "step": 539
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.87533685404585e-06,
+      "loss": 0.177,
+      "step": 540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.844173338052771e-06,
+      "loss": 0.2027,
+      "step": 541
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.81301133569098e-06,
+      "loss": 0.2097,
+      "step": 542
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.78185114965439e-06,
+      "loss": 0.2292,
+      "step": 543
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.750693082619274e-06,
+      "loss": 0.1777,
+      "step": 544
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.719537437241311e-06,
+      "loss": 0.1725,
+      "step": 545
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.68838451615267e-06,
+      "loss": 0.1936,
+      "step": 546
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.65723462195905e-06,
+      "loss": 0.2074,
+      "step": 547
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.626088057236745e-06,
+      "loss": 0.207,
+      "step": 548
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.594945124529718e-06,
+      "loss": 0.2384,
+      "step": 549
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.563806126346643e-06,
+      "loss": 0.2318,
+      "step": 550
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.532671365157979e-06,
+      "loss": 0.2157,
+      "step": 551
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.501541143393028e-06,
+      "loss": 0.1848,
+      "step": 552
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.470415763437003e-06,
+      "loss": 0.258,
+      "step": 553
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.439295527628083e-06,
+      "loss": 0.2399,
+      "step": 554
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.408180738254472e-06,
+      "loss": 0.1745,
+      "step": 555
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.377071697551479e-06,
+      "loss": 0.1893,
+      "step": 556
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.34596870769857e-06,
+      "loss": 0.1848,
+      "step": 557
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.314872070816435e-06,
+      "loss": 0.1992,
+      "step": 558
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.28378208896405e-06,
+      "loss": 0.1764,
+      "step": 559
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 0.2098,
+      "step": 560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.221623298258315e-06,
+      "loss": 0.2656,
+      "step": 561
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.190555093187968e-06,
+      "loss": 0.2056,
+      "step": 562
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.159494750707527e-06,
+      "loss": 0.2249,
+      "step": 563
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.128442572523418e-06,
+      "loss": 0.1823,
+      "step": 564
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.097398860262777e-06,
+      "loss": 0.2077,
+      "step": 565
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.066363915470494e-06,
+      "loss": 0.2163,
+      "step": 566
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.0353380396063e-06,
+      "loss": 0.2313,
+      "step": 567
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.004321534041836e-06,
+      "loss": 0.1737,
+      "step": 568
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.973314700057717e-06,
+      "loss": 0.1861,
+      "step": 569
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.942317838840625e-06,
+      "loss": 0.1689,
+      "step": 570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.911331251480357e-06,
+      "loss": 0.2246,
+      "step": 571
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 0.1814,
+      "step": 572
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.849390102187615e-06,
+      "loss": 0.2076,
+      "step": 573
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.818436141924072e-06,
+      "loss": 0.1892,
+      "step": 574
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.787493658849387e-06,
+      "loss": 0.2049,
+      "step": 575
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.756562953525151e-06,
+      "loss": 0.186,
+      "step": 576
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.72564432639856e-06,
+      "loss": 0.2115,
+      "step": 577
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.694738077799487e-06,
+      "loss": 0.2053,
+      "step": 578
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.663844507937563e-06,
+      "loss": 0.2018,
+      "step": 579
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.632963916899268e-06,
+      "loss": 0.1945,
+      "step": 580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.602096604645009e-06,
+      "loss": 0.2089,
+      "step": 581
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.571242871006202e-06,
+      "loss": 0.2161,
+      "step": 582
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.540403015682382e-06,
+      "loss": 0.2438,
+      "step": 583
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.509577338238255e-06,
+      "loss": 0.1983,
+      "step": 584
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.478766138100834e-06,
+      "loss": 0.2422,
+      "step": 585
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.447969714556484e-06,
+      "loss": 0.1839,
+      "step": 586
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.417188366748051e-06,
+      "loss": 0.1894,
+      "step": 587
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.386422393671934e-06,
+      "loss": 0.1977,
+      "step": 588
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.355672094175192e-06,
+      "loss": 0.2107,
+      "step": 589
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.324937766952638e-06,
+      "loss": 0.2266,
+      "step": 590
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.294219710543931e-06,
+      "loss": 0.1574,
+      "step": 591
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.2412,
+      "step": 592
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.232833603533601e-06,
+      "loss": 0.1782,
+      "step": 593
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.202166149209475e-06,
+      "loss": 0.1725,
+      "step": 594
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.171516158248406e-06,
+      "loss": 0.2064,
+      "step": 595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.140883928370855e-06,
+      "loss": 0.2102,
+      "step": 596
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.11026975712476e-06,
+      "loss": 0.1976,
+      "step": 597
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.079673941882639e-06,
+      "loss": 0.2059,
+      "step": 598
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.1822,
+      "step": 599
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.018538568006027e-06,
+      "loss": 0.1861,
+      "step": 600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.987999603213518e-06,
+      "loss": 0.1705,
+      "step": 601
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.957480182103198e-06,
+      "loss": 0.2187,
+      "step": 602
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.926980601127225e-06,
+      "loss": 0.1835,
+      "step": 603
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.896501156545044e-06,
+      "loss": 0.1959,
+      "step": 604
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.866042144420502e-06,
+      "loss": 0.1378,
+      "step": 605
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.835603860618973e-06,
+      "loss": 0.2076,
+      "step": 606
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.805186600804489e-06,
+      "loss": 0.2367,
+      "step": 607
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.1519,
+      "step": 608
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.744416334768809e-06,
+      "loss": 0.216,
+      "step": 609
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.714063918843106e-06,
+      "loss": 0.1993,
+      "step": 610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.6837337074897e-06,
+      "loss": 0.174,
+      "step": 611
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.653425995322852e-06,
+      "loss": 0.2305,
+      "step": 612
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.623141076738271e-06,
+      "loss": 0.1902,
+      "step": 613
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.592879245910273e-06,
+      "loss": 0.1773,
+      "step": 614
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.562640796788893e-06,
+      "loss": 0.1512,
+      "step": 615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.532426023097063e-06,
+      "loss": 0.2326,
+      "step": 616
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.50223521832773e-06,
+      "loss": 0.206,
+      "step": 617
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.472068675741024e-06,
+      "loss": 0.2196,
+      "step": 618
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4419266883614e-06,
+      "loss": 0.1776,
+      "step": 619
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.2184,
+      "step": 620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.38171755012578e-06,
+      "loss": 0.1662,
+      "step": 621
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3516509841147276e-06,
+      "loss": 0.1931,
+      "step": 622
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.321610142994971e-06,
+      "loss": 0.2129,
+      "step": 623
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.291595318569951e-06,
+      "loss": 0.2201,
+      "step": 624
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2616068023904e-06,
+      "loss": 0.267,
+      "step": 625
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2316448857515076e-06,
+      "loss": 0.1923,
+      "step": 626
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.201709859690081e-06,
+      "loss": 0.1925,
+      "step": 627
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.171802014981726e-06,
+      "loss": 0.217,
+      "step": 628
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.141921642138025e-06,
+      "loss": 0.203,
+      "step": 629
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.112069031403704e-06,
+      "loss": 0.206,
+      "step": 630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.082244472753823e-06,
+      "loss": 0.2189,
+      "step": 631
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.052448255890958e-06,
+      "loss": 0.2004,
+      "step": 632
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.022680670242387e-06,
+      "loss": 0.187,
+      "step": 633
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.992942004957271e-06,
+      "loss": 0.1921,
+      "step": 634
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.963232548903853e-06,
+      "loss": 0.1626,
+      "step": 635
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.933552590666659e-06,
+      "loss": 0.1767,
+      "step": 636
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.903902418543671e-06,
+      "loss": 0.1935,
+      "step": 637
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.874282320543557e-06,
+      "loss": 0.2,
+      "step": 638
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.844692584382848e-06,
+      "loss": 0.2252,
+      "step": 639
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.815133497483157e-06,
+      "loss": 0.1718,
+      "step": 640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.785605346968387e-06,
+      "loss": 0.1835,
+      "step": 641
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7561084196619306e-06,
+      "loss": 0.2316,
+      "step": 642
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7266430020839e-06,
+      "loss": 0.2072,
+      "step": 643
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 0.2091,
+      "step": 644
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.66780784066041e-06,
+      "loss": 0.2072,
+      "step": 645
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.638438668313695e-06,
+      "loss": 0.1658,
+      "step": 646
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.609102148687333e-06,
+      "loss": 0.2096,
+      "step": 647
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.1684,
+      "step": 648
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.550528207123667e-06,
+      "loss": 0.2183,
+      "step": 649
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.521291354147727e-06,
+      "loss": 0.2208,
+      "step": 650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.492088291809355e-06,
+      "loss": 0.235,
+      "step": 651
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.462919303774186e-06,
+      "loss": 0.2127,
+      "step": 652
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.43378467337687e-06,
+      "loss": 0.1986,
+      "step": 653
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.404684683618325e-06,
+      "loss": 0.1875,
+      "step": 654
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.375619617162985e-06,
+      "loss": 0.2033,
+      "step": 655
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 0.1907,
+      "step": 656
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.317595383120756e-06,
+      "loss": 0.234,
+      "step": 657
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.288636779155621e-06,
+      "loss": 0.2268,
+      "step": 658
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2597142257317185e-06,
+      "loss": 0.1908,
+      "step": 659
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.230828003789949e-06,
+      "loss": 0.2285,
+      "step": 660
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.201978393918291e-06,
+      "loss": 0.1939,
+      "step": 661
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.1946,
+      "step": 662
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.144390130956384e-06,
+      "loss": 0.2038,
+      "step": 663
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.115652037253054e-06,
+      "loss": 0.2027,
+      "step": 664
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.086951674388252e-06,
+      "loss": 0.2253,
+      "step": 665
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.058289321144608e-06,
+      "loss": 0.194,
+      "step": 666
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.02966525593555e-06,
+      "loss": 0.1677,
+      "step": 667
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.001079756802592e-06,
+      "loss": 0.2007,
+      "step": 668
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.97253310141263e-06,
+      "loss": 0.2115,
+      "step": 669
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.944025567055251e-06,
+      "loss": 0.2347,
+      "step": 670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.91555743064004e-06,
+      "loss": 0.1728,
+      "step": 671
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.887128968693887e-06,
+      "loss": 0.2301,
+      "step": 672
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.858740457358298e-06,
+      "loss": 0.1849,
+      "step": 673
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.830392172386723e-06,
+      "loss": 0.2344,
+      "step": 674
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.802084389141862e-06,
+      "loss": 0.2254,
+      "step": 675
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.773817382593008e-06,
+      "loss": 0.1937,
+      "step": 676
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.745591427313365e-06,
+      "loss": 0.1956,
+      "step": 677
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.717406797477371e-06,
+      "loss": 0.1914,
+      "step": 678
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.689263766858072e-06,
+      "loss": 0.1453,
+      "step": 679
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.2375,
+      "step": 680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.633103596338631e-06,
+      "loss": 0.186,
+      "step": 681
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6050870019535496e-06,
+      "loss": 0.2151,
+      "step": 682
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5771130978099896e-06,
+      "loss": 0.2601,
+      "step": 683
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.549182155634076e-06,
+      "loss": 0.217,
+      "step": 684
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.521294446734637e-06,
+      "loss": 0.2154,
+      "step": 685
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.493450242000546e-06,
+      "loss": 0.2126,
+      "step": 686
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.465649811898098e-06,
+      "loss": 0.1736,
+      "step": 687
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.43789342646837e-06,
+      "loss": 0.1717,
+      "step": 688
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.410181355324622e-06,
+      "loss": 0.2172,
+      "step": 689
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.382513867649663e-06,
+      "loss": 0.1749,
+      "step": 690
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.354891232193225e-06,
+      "loss": 0.1913,
+      "step": 691
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.32731371726938e-06,
+      "loss": 0.1716,
+      "step": 692
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.299781590753916e-06,
+      "loss": 0.1678,
+      "step": 693
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2722951200817315e-06,
+      "loss": 0.2042,
+      "step": 694
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.244854572244249e-06,
+      "loss": 0.2333,
+      "step": 695
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.217460213786822e-06,
+      "loss": 0.2132,
+      "step": 696
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.190112310806126e-06,
+      "loss": 0.174,
+      "step": 697
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1628111289476025e-06,
+      "loss": 0.2333,
+      "step": 698
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.135556933402862e-06,
+      "loss": 0.1917,
+      "step": 699
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.108349988907111e-06,
+      "loss": 0.1775,
+      "step": 700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.081190559736569e-06,
+      "loss": 0.1912,
+      "step": 701
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.054078909705926e-06,
+      "loss": 0.1913,
+      "step": 702
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.027015302165768e-06,
+      "loss": 0.1585,
+      "step": 703
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.1887,
+      "step": 704
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.973033265623333e-06,
+      "loss": 0.2051,
+      "step": 705
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.946115360978696e-06,
+      "loss": 0.2369,
+      "step": 706
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.919246547534709e-06,
+      "loss": 0.2128,
+      "step": 707
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.892427086283147e-06,
+      "loss": 0.2136,
+      "step": 708
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.865657237736397e-06,
+      "loss": 0.1993,
+      "step": 709
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.838937261924933e-06,
+      "loss": 0.1996,
+      "step": 710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.812267418394784e-06,
+      "loss": 0.1674,
+      "step": 711
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.78564796620502e-06,
+      "loss": 0.1856,
+      "step": 712
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.759079163925223e-06,
+      "loss": 0.23,
+      "step": 713
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.732561269632992e-06,
+      "loss": 0.2363,
+      "step": 714
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.706094540911429e-06,
+      "loss": 0.1998,
+      "step": 715
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.2292,
+      "step": 716
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6533156080252076e-06,
+      "loss": 0.1936,
+      "step": 717
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.627003916531761e-06,
+      "loss": 0.2148,
+      "step": 718
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.600744415946438e-06,
+      "loss": 0.1721,
+      "step": 719
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5745373613424075e-06,
+      "loss": 0.229,
+      "step": 720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.548383007283412e-06,
+      "loss": 0.1631,
+      "step": 721
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.522281607821288e-06,
+      "loss": 0.1901,
+      "step": 722
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.496233416493481e-06,
+      "loss": 0.2303,
+      "step": 723
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.470238686320606e-06,
+      "loss": 0.2108,
+      "step": 724
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.444297669803981e-06,
+      "loss": 0.2034,
+      "step": 725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.418410618923163e-06,
+      "loss": 0.1761,
+      "step": 726
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.392577785133521e-06,
+      "loss": 0.2433,
+      "step": 727
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3667994193637794e-06,
+      "loss": 0.2341,
+      "step": 728
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3410757720135886e-06,
+      "loss": 0.1733,
+      "step": 729
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.315407092951078e-06,
+      "loss": 0.1906,
+      "step": 730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.289793631510449e-06,
+      "loss": 0.194,
+      "step": 731
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.264235636489542e-06,
+      "loss": 0.192,
+      "step": 732
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.238733356147414e-06,
+      "loss": 0.1992,
+      "step": 733
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213287038201943e-06,
+      "loss": 0.1599,
+      "step": 734
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.187896929827414e-06,
+      "loss": 0.2,
+      "step": 735
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.162563277652104e-06,
+      "loss": 0.1963,
+      "step": 736
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.137286327755913e-06,
+      "loss": 0.2077,
+      "step": 737
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.112066325667954e-06,
+      "loss": 0.2046,
+      "step": 738
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.086903516364179e-06,
+      "loss": 0.1972,
+      "step": 739
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.061798144264986e-06,
+      "loss": 0.1413,
+      "step": 740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.03675045323286e-06,
+      "loss": 0.2233,
+      "step": 741
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0117606865699975e-06,
+      "loss": 0.218,
+      "step": 742
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.986829087015941e-06,
+      "loss": 0.1825,
+      "step": 743
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.961955896745224e-06,
+      "loss": 0.1993,
+      "step": 744
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.937141357365023e-06,
+      "loss": 0.1426,
+      "step": 745
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.912385709912794e-06,
+      "loss": 0.191,
+      "step": 746
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.887689194853951e-06,
+      "loss": 0.2164,
+      "step": 747
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8630520520795275e-06,
+      "loss": 0.2245,
+      "step": 748
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.838474520903825e-06,
+      "loss": 0.2123,
+      "step": 749
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8139568400621184e-06,
+      "loss": 0.1846,
+      "step": 750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7894992477083226e-06,
+      "loss": 0.212,
+      "step": 751
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.1747,
+      "step": 752
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7407652781594094e-06,
+      "loss": 0.1706,
+      "step": 753
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7164893743445274e-06,
+      "loss": 0.186,
+      "step": 754
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.692274505773419e-06,
+      "loss": 0.144,
+      "step": 755
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6681209076586035e-06,
+      "loss": 0.181,
+      "step": 756
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.644028814617454e-06,
+      "loss": 0.1896,
+      "step": 757
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.619998460669916e-06,
+      "loss": 0.1952,
+      "step": 758
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5960300792362124e-06,
+      "loss": 0.2055,
+      "step": 759
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.1877,
+      "step": 760
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5482801645791266e-06,
+      "loss": 0.1732,
+      "step": 761
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5244990951772972e-06,
+      "loss": 0.2262,
+      "step": 762
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5007809259279146e-06,
+      "loss": 0.1873,
+      "step": 763
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4771258872187917e-06,
+      "loss": 0.1888,
+      "step": 764
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.453534208824507e-06,
+      "loss": 0.1957,
+      "step": 765
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4300061199041967e-06,
+      "loss": 0.1768,
+      "step": 766
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4065418489993118e-06,
+      "loss": 0.1889,
+      "step": 767
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3831416240314085e-06,
+      "loss": 0.179,
+      "step": 768
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3598056722999185e-06,
+      "loss": 0.2129,
+      "step": 769
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3365342204799613e-06,
+      "loss": 0.1804,
+      "step": 770
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3133274946201333e-06,
+      "loss": 0.2096,
+      "step": 771
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.290185720140301e-06,
+      "loss": 0.1598,
+      "step": 772
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.267109121829428e-06,
+      "loss": 0.1989,
+      "step": 773
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2440979238433977e-06,
+      "loss": 0.1883,
+      "step": 774
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.221152349702802e-06,
+      "loss": 0.211,
+      "step": 775
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1982726222908046e-06,
+      "loss": 0.1586,
+      "step": 776
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1754589638509647e-06,
+      "loss": 0.2006,
+      "step": 777
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.152711595985065e-06,
+      "loss": 0.2364,
+      "step": 778
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1300307396509833e-06,
+      "loss": 0.1697,
+      "step": 779
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.10741661516053e-06,
+      "loss": 0.1951,
+      "step": 780
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0848694421773075e-06,
+      "loss": 0.1979,
+      "step": 781
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0623894397145837e-06,
+      "loss": 0.2081,
+      "step": 782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0399768261331664e-06,
+      "loss": 0.2051,
+      "step": 783
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.017631819139273e-06,
+      "loss": 0.1897,
+      "step": 784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.995354635782417e-06,
+      "loss": 0.1888,
+      "step": 785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9731454924533086e-06,
+      "loss": 0.2013,
+      "step": 786
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.95100460488175e-06,
+      "loss": 0.1524,
+      "step": 787
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.2057,
+      "step": 788
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.906928456613336e-06,
+      "loss": 0.2109,
+      "step": 789
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.884993624052701e-06,
+      "loss": 0.201,
+      "step": 790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8631279035178796e-06,
+      "loss": 0.1755,
+      "step": 791
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8413315074028157e-06,
+      "loss": 0.1988,
+      "step": 792
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.819604647428067e-06,
+      "loss": 0.2055,
+      "step": 793
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7979475346387363e-06,
+      "loss": 0.1913,
+      "step": 794
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.776360379402445e-06,
+      "loss": 0.2225,
+      "step": 795
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7548433914072736e-06,
+      "loss": 0.1769,
+      "step": 796
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7333967796597317e-06,
+      "loss": 0.162,
+      "step": 797
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.712020752482717e-06,
+      "loss": 0.2094,
+      "step": 798
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.690715517513508e-06,
+      "loss": 0.1883,
+      "step": 799
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 0.1695,
+      "step": 800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6483182513073835e-06,
+      "loss": 0.1784,
+      "step": 801
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6272266318987606e-06,
+      "loss": 0.181,
+      "step": 802
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6062066283505404e-06,
+      "loss": 0.1841,
+      "step": 803
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5852584448417327e-06,
+      "loss": 0.1839,
+      "step": 804
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.564382284853738e-06,
+      "loss": 0.2004,
+      "step": 805
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5435783511683444e-06,
+      "loss": 0.1948,
+      "step": 806
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5228468458657585e-06,
+      "loss": 0.2216,
+      "step": 807
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.502187970322657e-06,
+      "loss": 0.1912,
+      "step": 808
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4816019252102274e-06,
+      "loss": 0.2035,
+      "step": 809
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.461088910492202e-06,
+      "loss": 0.2033,
+      "step": 810
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.440649125422937e-06,
+      "loss": 0.175,
+      "step": 811
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.420282768545469e-06,
+      "loss": 0.1906,
+      "step": 812
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3999900376895844e-06,
+      "loss": 0.2035,
+      "step": 813
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3797711299698924e-06,
+      "loss": 0.2323,
+      "step": 814
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3596262417839256e-06,
+      "loss": 0.1857,
+      "step": 815
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.2047,
+      "step": 816
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.319559306006417e-06,
+      "loss": 0.2609,
+      "step": 817
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2996376476073724e-06,
+      "loss": 0.164,
+      "step": 818
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2797907871232673e-06,
+      "loss": 0.1576,
+      "step": 819
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2600189173377263e-06,
+      "loss": 0.1792,
+      "step": 820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.240322230305951e-06,
+      "loss": 0.1791,
+      "step": 821
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2207009173528528e-06,
+      "loss": 0.1871,
+      "step": 822
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.201155169071184e-06,
+      "loss": 0.1857,
+      "step": 823
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.2209,
+      "step": 824
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1622911252213195e-06,
+      "loss": 0.1837,
+      "step": 825
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1429732071612653e-06,
+      "loss": 0.1861,
+      "step": 826
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1237316087852465e-06,
+      "loss": 0.2274,
+      "step": 827
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.104566516997647e-06,
+      "loss": 0.1811,
+      "step": 828
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0854781179596937e-06,
+      "loss": 0.1937,
+      "step": 829
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0664665970876496e-06,
+      "loss": 0.2116,
+      "step": 830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0475321390510262e-06,
+      "loss": 0.1793,
+      "step": 831
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0286749277707783e-06,
+      "loss": 0.1877,
+      "step": 832
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.009895146417512e-06,
+      "loss": 0.2079,
+      "step": 833
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9911929774097216e-06,
+      "loss": 0.1863,
+      "step": 834
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9725686024120093e-06,
+      "loss": 0.1799,
+      "step": 835
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9540222023333165e-06,
+      "loss": 0.1986,
+      "step": 836
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9355539573251737e-06,
+      "loss": 0.2047,
+      "step": 837
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9171640467799478e-06,
+      "loss": 0.1995,
+      "step": 838
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8988526493290948e-06,
+      "loss": 0.1667,
+      "step": 839
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.880619942841435e-06,
+      "loss": 0.216,
+      "step": 840
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8624661044214154e-06,
+      "loss": 0.2285,
+      "step": 841
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8443913104073984e-06,
+      "loss": 0.1693,
+      "step": 842
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.826395736369937e-06,
+      "loss": 0.2317,
+      "step": 843
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.808479557110081e-06,
+      "loss": 0.2045,
+      "step": 844
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7906429466576768e-06,
+      "loss": 0.1893,
+      "step": 845
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7728860782696666e-06,
+      "loss": 0.1855,
+      "step": 846
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7552091244284197e-06,
+      "loss": 0.1853,
+      "step": 847
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 0.1925,
+      "step": 848
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7200956464327512e-06,
+      "loss": 0.1824,
+      "step": 849
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7026594633551252e-06,
+      "loss": 0.184,
+      "step": 850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6853038769745466e-06,
+      "loss": 0.2051,
+      "step": 851
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6680290558755119e-06,
+      "loss": 0.217,
+      "step": 852
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6508351678579882e-06,
+      "loss": 0.1731,
+      "step": 853
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6337223799358025e-06,
+      "loss": 0.2135,
+      "step": 854
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6166908583350138e-06,
+      "loss": 0.1927,
+      "step": 855
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.599740768492286e-06,
+      "loss": 0.2015,
+      "step": 856
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.582872275053301e-06,
+      "loss": 0.2133,
+      "step": 857
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.566085541871145e-06,
+      "loss": 0.1621,
+      "step": 858
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5493807320047183e-06,
+      "loss": 0.1993,
+      "step": 859
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5327580077171589e-06,
+      "loss": 0.1822,
+      "step": 860
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5162175304742633e-06,
+      "loss": 0.1861,
+      "step": 861
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.499759460942909e-06,
+      "loss": 0.1826,
+      "step": 862
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4833839589895072e-06,
+      "loss": 0.2332,
+      "step": 863
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.467091183678444e-06,
+      "loss": 0.1965,
+      "step": 864
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4508812932705364e-06,
+      "loss": 0.1706,
+      "step": 865
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4347544452214869e-06,
+      "loss": 0.1527,
+      "step": 866
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4187107961803704e-06,
+      "loss": 0.1816,
+      "step": 867
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4027505019880972e-06,
+      "loss": 0.2167,
+      "step": 868
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3868737176759105e-06,
+      "loss": 0.1854,
+      "step": 869
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3710805974638697e-06,
+      "loss": 0.2114,
+      "step": 870
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3553712947593655e-06,
+      "loss": 0.1707,
+      "step": 871
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.1876,
+      "step": 872
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.324204751430186e-06,
+      "loss": 0.2214,
+      "step": 873
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3087478135435361e-06,
+      "loss": 0.1993,
+      "step": 874
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.293375298637518e-06,
+      "loss": 0.1672,
+      "step": 875
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.278087356033947e-06,
+      "loss": 0.1968,
+      "step": 876
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2628841342331389e-06,
+      "loss": 0.1739,
+      "step": 877
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2477657809124632e-06,
+      "loss": 0.1611,
+      "step": 878
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2327324429249232e-06,
+      "loss": 0.1711,
+      "step": 879
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2177842662977136e-06,
+      "loss": 0.2045,
+      "step": 880
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2029213962308172e-06,
+      "loss": 0.1839,
+      "step": 881
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.188143977095576e-06,
+      "loss": 0.1725,
+      "step": 882
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1734521524333087e-06,
+      "loss": 0.1644,
+      "step": 883
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1588460649539036e-06,
+      "loss": 0.1676,
+      "step": 884
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1443258565344329e-06,
+      "loss": 0.1542,
+      "step": 885
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.129891668217783e-06,
+      "loss": 0.2249,
+      "step": 886
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1155436402112785e-06,
+      "loss": 0.1742,
+      "step": 887
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1012819118853147e-06,
+      "loss": 0.1868,
+      "step": 888
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0871066217720173e-06,
+      "loss": 0.1788,
+      "step": 889
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.073017907563887e-06,
+      "loss": 0.1971,
+      "step": 890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0590159061124606e-06,
+      "loss": 0.2081,
+      "step": 891
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0451007534269908e-06,
+      "loss": 0.1941,
+      "step": 892
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0312725846731174e-06,
+      "loss": 0.2039,
+      "step": 893
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0175315341715598e-06,
+      "loss": 0.1995,
+      "step": 894
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.003877735396801e-06,
+      "loss": 0.2382,
+      "step": 895
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.1926,
+      "step": 896
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.768324226867353e-07,
+      "loss": 0.2006,
+      "step": 897
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.634411714576353e-07,
+      "loss": 0.1786,
+      "step": 898
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.501376973651999e-07,
+      "loss": 0.1645,
+      "step": 899
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.369221296335007e-07,
+      "loss": 0.1837,
+      "step": 900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.237945966327133e-07,
+      "loss": 0.2088,
+      "step": 901
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.107552258778907e-07,
+      "loss": 0.1978,
+      "step": 902
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.978041440277163e-07,
+      "loss": 0.232,
+      "step": 903
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.849414768832687e-07,
+      "loss": 0.2096,
+      "step": 904
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.721673493868111e-07,
+      "loss": 0.2535,
+      "step": 905
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.5948188562057e-07,
+      "loss": 0.2034,
+      "step": 906
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.468852088055291e-07,
+      "loss": 0.25,
+      "step": 907
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.343774413002382e-07,
+      "loss": 0.214,
+      "step": 908
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.219587045996203e-07,
+      "loss": 0.1842,
+      "step": 909
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.096291193337935e-07,
+      "loss": 0.1769,
+      "step": 910
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.973888052668943e-07,
+      "loss": 0.2118,
+      "step": 911
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.852378812959227e-07,
+      "loss": 0.1779,
+      "step": 912
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.731764654495832e-07,
+      "loss": 0.2335,
+      "step": 913
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.1741,
+      "step": 914
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.493226258972519e-07,
+      "loss": 0.1666,
+      "step": 915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.375304338969135e-07,
+      "loss": 0.19,
+      "step": 916
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.258282134302519e-07,
+      "loss": 0.1602,
+      "step": 917
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.142160781674645e-07,
+      "loss": 0.1518,
+      "step": 918
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.026941409036991e-07,
+      "loss": 0.2394,
+      "step": 919
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.912625135579587e-07,
+      "loss": 0.2053,
+      "step": 920
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.799213071720156e-07,
+      "loss": 0.1823,
+      "step": 921
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.68670631909335e-07,
+      "loss": 0.1832,
+      "step": 922
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.57510597054003e-07,
+      "loss": 0.165,
+      "step": 923
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.464413110096601e-07,
+      "loss": 0.205,
+      "step": 924
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.354628812984576e-07,
+      "loss": 0.1882,
+      "step": 925
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.245754145600091e-07,
+      "loss": 0.1562,
+      "step": 926
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.137790165503499e-07,
+      "loss": 0.2059,
+      "step": 927
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.2044,
+      "step": 928
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.924598453175278e-07,
+      "loss": 0.1807,
+      "step": 929
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.819372791793654e-07,
+      "loss": 0.1813,
+      "step": 930
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.715061959379875e-07,
+      "loss": 0.1779,
+      "step": 931
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.611666969163243e-07,
+      "loss": 0.2346,
+      "step": 932
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.509188825476964e-07,
+      "loss": 0.201,
+      "step": 933
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.407628523748398e-07,
+      "loss": 0.1885,
+      "step": 934
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.306987050489442e-07,
+      "loss": 0.2107,
+      "step": 935
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.207265383286831e-07,
+      "loss": 0.1596,
+      "step": 936
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.108464490792753e-07,
+      "loss": 0.201,
+      "step": 937
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.010585332715401e-07,
+      "loss": 0.2232,
+      "step": 938
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.913628859809638e-07,
+      "loss": 0.2225,
+      "step": 939
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.817596013867765e-07,
+      "loss": 0.2014,
+      "step": 940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7224877277103673e-07,
+      "loss": 0.2343,
+      "step": 941
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.628304925177318e-07,
+      "loss": 0.2072,
+      "step": 942
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.535048521118668e-07,
+      "loss": 0.2328,
+      "step": 943
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 0.1782,
+      "step": 944
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.351318522823134e-07,
+      "loss": 0.1995,
+      "step": 945
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2608467132581934e-07,
+      "loss": 0.1822,
+      "step": 946
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.171304871494264e-07,
+      "loss": 0.1847,
+      "step": 947
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.082693867301224e-07,
+      "loss": 0.173,
+      "step": 948
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.99501456140714e-07,
+      "loss": 0.2425,
+      "step": 949
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.908267805490051e-07,
+      "loss": 0.2184,
+      "step": 950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8224544421695766e-07,
+      "loss": 0.2342,
+      "step": 951
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7375753049987974e-07,
+      "loss": 0.2128,
+      "step": 952
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6536312184560996e-07,
+      "loss": 0.1572,
+      "step": 953
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.570622997937234e-07,
+      "loss": 0.1766,
+      "step": 954
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4885514497473574e-07,
+      "loss": 0.2025,
+      "step": 955
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.2631,
+      "step": 956
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.327221550075266e-07,
+      "loss": 0.1534,
+      "step": 957
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.247964765680389e-07,
+      "loss": 0.2218,
+      "step": 958
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1696477877738664e-07,
+      "loss": 0.2127,
+      "step": 959
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0922713770922155e-07,
+      "loss": 0.214,
+      "step": 960
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0158362852356627e-07,
+      "loss": 0.201,
+      "step": 961
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.940343254660905e-07,
+      "loss": 0.1895,
+      "step": 962
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.865793018673857e-07,
+      "loss": 0.1845,
+      "step": 963
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7921863014225504e-07,
+      "loss": 0.1834,
+      "step": 964
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7195238178900685e-07,
+      "loss": 0.2098,
+      "step": 965
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6478062738876654e-07,
+      "loss": 0.215,
+      "step": 966
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.577034366047848e-07,
+      "loss": 0.1811,
+      "step": 967
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.1579,
+      "step": 968
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4383301994518773e-07,
+      "loss": 0.2576,
+      "step": 969
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.370399288006664e-07,
+      "loss": 0.1737,
+      "step": 970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3034167073328283e-07,
+      "loss": 0.1905,
+      "step": 971
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2373831080695463e-07,
+      "loss": 0.17,
+      "step": 972
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1722991316380005e-07,
+      "loss": 0.1898,
+      "step": 973
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1081654102351634e-07,
+      "loss": 0.1597,
+      "step": 974
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0449825668276246e-07,
+      "loss": 0.2054,
+      "step": 975
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9827512151456175e-07,
+      "loss": 0.2174,
+      "step": 976
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.921471959676957e-07,
+      "loss": 0.1657,
+      "step": 977
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8611453956612346e-07,
+      "loss": 0.1931,
+      "step": 978
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8017721090840324e-07,
+      "loss": 0.1814,
+      "step": 979
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7433526766711727e-07,
+      "loss": 0.1865,
+      "step": 980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6858876658832235e-07,
+      "loss": 0.201,
+      "step": 981
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.629377634909868e-07,
+      "loss": 0.2141,
+      "step": 982
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5738231326645758e-07,
+      "loss": 0.2043,
+      "step": 983
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.2361,
+      "step": 984
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.465582863598791e-07,
+      "loss": 0.1925,
+      "step": 985
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4128981481764115e-07,
+      "loss": 0.196,
+      "step": 986
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3611710642681076e-07,
+      "loss": 0.1608,
+      "step": 987
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3104021143278911e-07,
+      "loss": 0.1536,
+      "step": 988
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2605917915028743e-07,
+      "loss": 0.1811,
+      "step": 989
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2117405796285286e-07,
+      "loss": 0.2412,
+      "step": 990
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1638489532239339e-07,
+      "loss": 0.1897,
+      "step": 991
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 0.1526,
+      "step": 992
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0709463082907545e-07,
+      "loss": 0.1847,
+      "step": 993
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0259361921774014e-07,
+      "loss": 0.188,
+      "step": 994
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.818874663554356e-08,
+      "loss": 0.2109,
+      "step": 995
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.388005586947191e-08,
+      "loss": 0.1913,
+      "step": 996
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.966758877224202e-08,
+      "loss": 0.2325,
+      "step": 997
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.555138626189619e-08,
+      "loss": 0.1621,
+      "step": 998
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.153148832140467e-08,
+      "loss": 0.1824,
+      "step": 999
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.760793399827937e-08,
+      "loss": 0.1909,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1040,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 100,
+  "total_flos": 52178597019648.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}