diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15330 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 2550,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.3919,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.3877,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 0.5515,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.4575,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.4384,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5909,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4e-05,
+      "loss": 0.5512,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6439,
+      "step": 8
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4596,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3938,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999992351042268e-05,
+      "loss": 0.5131,
+      "step": 11
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9999969404180768e-05,
+      "loss": 0.2911,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9999931159450608e-05,
+      "loss": 0.4085,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999987761691029e-05,
+      "loss": 0.5575,
+      "step": 14
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999808776641724e-05,
+      "loss": 0.4824,
+      "step": 15
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999724638750225e-05,
+      "loss": 0.3096,
+      "step": 16
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999625203364505e-05,
+      "loss": 0.366,
+      "step": 17
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999510470636677e-05,
+      "loss": 0.4442,
+      "step": 18
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999938044074226e-05,
+      "loss": 0.5573,
+      "step": 19
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999923511388017e-05,
+      "loss": 0.3941,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999907449027273e-05,
+      "loss": 0.4739,
+      "step": 21
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9998898570165663e-05,
+      "loss": 0.4873,
+      "step": 22
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9998707353828083e-05,
+      "loss": 0.3344,
+      "step": 23
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9998500841552517e-05,
+      "loss": 0.3786,
+      "step": 24
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998279033654883e-05,
+      "loss": 0.3359,
+      "step": 25
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998041930474498e-05,
+      "loss": 0.4744,
+      "step": 26
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999778953237409e-05,
+      "loss": 0.4769,
+      "step": 27
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997521839739767e-05,
+      "loss": 0.5549,
+      "step": 28
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997238852981043e-05,
+      "loss": 0.4316,
+      "step": 29
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999694057253083e-05,
+      "loss": 0.4973,
+      "step": 30
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999662699884544e-05,
+      "loss": 0.4756,
+      "step": 31
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999629813240457e-05,
+      "loss": 0.371,
+      "step": 32
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999595397371132e-05,
+      "loss": 0.3762,
+      "step": 33
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999559452329218e-05,
+      "loss": 0.3605,
+      "step": 34
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999521978169703e-05,
+      "loss": 0.2511,
+      "step": 35
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9994829749499153e-05,
+      "loss": 0.2381,
+      "step": 36
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9994424427295215e-05,
+      "loss": 0.4014,
+      "step": 37
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999400381570527e-05,
+      "loss": 0.5437,
+      "step": 38
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9993567915372772e-05,
+      "loss": 0.4884,
+      "step": 39
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9993116726964554e-05,
+      "loss": 0.3813,
+      "step": 40
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9992650251170845e-05,
+      "loss": 0.2525,
+      "step": 41
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9992168488705248e-05,
+      "loss": 0.3392,
+      "step": 42
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9991671440304766e-05,
+      "loss": 0.5718,
+      "step": 43
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9991159106729777e-05,
+      "loss": 0.5557,
+      "step": 44
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9990631488764044e-05,
+      "loss": 0.4046,
+      "step": 45
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9990088587214712e-05,
+      "loss": 0.3928,
+      "step": 46
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9989530402912307e-05,
+      "loss": 0.2806,
+      "step": 47
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998895693671074e-05,
+      "loss": 0.3569,
+      "step": 48
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9988368189487287e-05,
+      "loss": 0.3077,
+      "step": 49
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9987764162142615e-05,
+      "loss": 0.3304,
+      "step": 50
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9987144855600754e-05,
+      "loss": 0.546,
+      "step": 51
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998651027080912e-05,
+      "loss": 0.3096,
+      "step": 52
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9985860408738488e-05,
+      "loss": 0.3508,
+      "step": 53
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.998519527038302e-05,
+      "loss": 0.2271,
+      "step": 54
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9984514856760233e-05,
+      "loss": 0.2994,
+      "step": 55
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9983819168911023e-05,
+      "loss": 0.3178,
+      "step": 56
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9983108207899643e-05,
+      "loss": 0.4092,
+      "step": 57
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9982381974813714e-05,
+      "loss": 0.2608,
+      "step": 58
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9981640470764227e-05,
+      "loss": 0.4931,
+      "step": 59
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998088369688552e-05,
+      "loss": 0.2412,
+      "step": 60
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998011165433531e-05,
+      "loss": 0.3089,
+      "step": 61
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9979324344294654e-05,
+      "loss": 0.3841,
+      "step": 62
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978521767967974e-05,
+      "loss": 0.2336,
+      "step": 63
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977703926583044e-05,
+      "loss": 0.3285,
+      "step": 64
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997687082139099e-05,
+      "loss": 0.2754,
+      "step": 65
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976022453666293e-05,
+      "loss": 0.2812,
+      "step": 66
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975158824706775e-05,
+      "loss": 0.2783,
+      "step": 67
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997427993583361e-05,
+      "loss": 0.2243,
+      "step": 68
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9973385788391315e-05,
+      "loss": 0.497,
+      "step": 69
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972476383747748e-05,
+      "loss": 0.3033,
+      "step": 70
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.997155172329411e-05,
+      "loss": 0.3444,
+      "step": 71
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9970611808444937e-05,
+      "loss": 0.3265,
+      "step": 72
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9969656640638105e-05,
+      "loss": 0.2752,
+      "step": 73
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9968686221334818e-05,
+      "loss": 0.3881,
+      "step": 74
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996770055201962e-05,
+      "loss": 0.2812,
+      "step": 75
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9966699634200375e-05,
+      "loss": 0.3541,
+      "step": 76
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9965683469408283e-05,
+      "loss": 0.3206,
+      "step": 77
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9964652059197863e-05,
+      "loss": 0.3032,
+      "step": 78
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.996360540514695e-05,
+      "loss": 0.205,
+      "step": 79
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9962543508856722e-05,
+      "loss": 0.2291,
+      "step": 80
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9961466371951644e-05,
+      "loss": 0.1969,
+      "step": 81
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.996037399607952e-05,
+      "loss": 0.2856,
+      "step": 82
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9959266382911456e-05,
+      "loss": 0.2507,
+      "step": 83
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9958143534141865e-05,
+      "loss": 0.433,
+      "step": 84
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9957005451488476e-05,
+      "loss": 0.3655,
+      "step": 85
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9955852136692315e-05,
+      "loss": 0.3383,
+      "step": 86
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9954683591517718e-05,
+      "loss": 0.417,
+      "step": 87
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9953499817752308e-05,
+      "loss": 0.3099,
+      "step": 88
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.995230081720702e-05,
+      "loss": 0.2062,
+      "step": 89
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.995108659171607e-05,
+      "loss": 0.2597,
+      "step": 90
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.994985714313697e-05,
+      "loss": 0.2237,
+      "step": 91
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.994861247335052e-05,
+      "loss": 0.2311,
+      "step": 92
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.994735258426081e-05,
+      "loss": 0.4204,
+      "step": 93
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99460774777952e-05,
+      "loss": 0.2367,
+      "step": 94
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9944787155904346e-05,
+      "loss": 0.2404,
+      "step": 95
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9943481620562166e-05,
+      "loss": 0.3445,
+      "step": 96
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9942160873765853e-05,
+      "loss": 0.2469,
+      "step": 97
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9940824917535887e-05,
+      "loss": 0.2128,
+      "step": 98
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9939473753915987e-05,
+      "loss": 0.2582,
+      "step": 99
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9938107384973165e-05,
+      "loss": 0.2013,
+      "step": 100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.993672581279767e-05,
+      "loss": 0.2682,
+      "step": 101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.993532903950303e-05,
+      "loss": 0.2931,
+      "step": 102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.993391706722601e-05,
+      "loss": 0.2307,
+      "step": 103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.993248989812663e-05,
+      "loss": 0.2068,
+      "step": 104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.993104753438817e-05,
+      "loss": 0.1919,
+      "step": 105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.992958997821714e-05,
+      "loss": 0.2927,
+      "step": 106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9928117231843302e-05,
+      "loss": 0.2586,
+      "step": 107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9926629297519648e-05,
+      "loss": 0.1824,
+      "step": 108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9925126177522403e-05,
+      "loss": 0.3001,
+      "step": 109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.992360787415103e-05,
+      "loss": 0.173,
+      "step": 110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9922074389728223e-05,
+      "loss": 0.2554,
+      "step": 111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9920525726599887e-05,
+      "loss": 0.2395,
+      "step": 112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9918961887135153e-05,
+      "loss": 0.2335,
+      "step": 113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9917382873726375e-05,
+      "loss": 0.2211,
+      "step": 114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9915788688789107e-05,
+      "loss": 0.2676,
+      "step": 115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9914179334762128e-05,
+      "loss": 0.1553,
+      "step": 116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.991255481410741e-05,
+      "loss": 0.2312,
+      "step": 117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.991091512931013e-05,
+      "loss": 0.1897,
+      "step": 118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9909260282878664e-05,
+      "loss": 0.2161,
+      "step": 119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9907590277344582e-05,
+      "loss": 0.1262,
+      "step": 120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9905905115262647e-05,
+      "loss": 0.2103,
+      "step": 121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.99042047992108e-05,
+      "loss": 0.1608,
+      "step": 122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.990248933179018e-05,
+      "loss": 0.1966,
+      "step": 123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9900758715625085e-05,
+      "loss": 0.2381,
+      "step": 124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9899012953363002e-05,
+      "loss": 0.1829,
+      "step": 125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.989725204767458e-05,
+      "loss": 0.1687,
+      "step": 126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9895476001253643e-05,
+      "loss": 0.2391,
+      "step": 127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9893684816817166e-05,
+      "loss": 0.1743,
+      "step": 128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.989187849710529e-05,
+      "loss": 0.2493,
+      "step": 129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9890057044881308e-05,
+      "loss": 0.2592,
+      "step": 130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9888220462931663e-05,
+      "loss": 0.2039,
+      "step": 131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.988636875406594e-05,
+      "loss": 0.252,
+      "step": 132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9884501921116873e-05,
+      "loss": 0.2028,
+      "step": 133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.988261996694032e-05,
+      "loss": 0.1747,
+      "step": 134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9880722894415284e-05,
+      "loss": 0.1362,
+      "step": 135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9878810706443887e-05,
+      "loss": 0.1191,
+      "step": 136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 0.2447,
+      "step": 137
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.987494099588613e-05,
+      "loss": 0.2442,
+      "step": 138
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.987298347921962e-05,
+      "loss": 0.2539,
+      "step": 139
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9871010858946443e-05,
+      "loss": 0.1214,
+      "step": 140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9869023138084296e-05,
+      "loss": 0.1313,
+      "step": 141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.986702031967398e-05,
+      "loss": 0.2094,
+      "step": 142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9865002406779383e-05,
+      "loss": 0.135,
+      "step": 143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.98629694024875e-05,
+      "loss": 0.1852,
+      "step": 144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9860921309908395e-05,
+      "loss": 0.1648,
+      "step": 145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.985885813217523e-05,
+      "loss": 0.2224,
+      "step": 146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9856779872444236e-05,
+      "loss": 0.1995,
+      "step": 147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9854686533894712e-05,
+      "loss": 0.2593,
+      "step": 148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9852578119729036e-05,
+      "loss": 0.2347,
+      "step": 149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9850454633172632e-05,
+      "loss": 0.1637,
+      "step": 150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9848316077474006e-05,
+      "loss": 0.2673,
+      "step": 151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9846162455904695e-05,
+      "loss": 0.1251,
+      "step": 152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9843993771759287e-05,
+      "loss": 0.2463,
+      "step": 153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.984181002835542e-05,
+      "loss": 0.1979,
+      "step": 154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9839611229033774e-05,
+      "loss": 0.1902,
+      "step": 155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.983739737715804e-05,
+      "loss": 0.1127,
+      "step": 156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.983516847611496e-05,
+      "loss": 0.1056,
+      "step": 157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9832924529314284e-05,
+      "loss": 0.2298,
+      "step": 158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.983066554018878e-05,
+      "loss": 0.1166,
+      "step": 159
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.982839151219424e-05,
+      "loss": 0.1694,
+      "step": 160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9826102448809442e-05,
+      "loss": 0.0688,
+      "step": 161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9823798353536183e-05,
+      "loss": 0.1681,
+      "step": 162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9821479229899246e-05,
+      "loss": 0.1814,
+      "step": 163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9819145081446403e-05,
+      "loss": 0.0869,
+      "step": 164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9816795911748422e-05,
+      "loss": 0.0741,
+      "step": 165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.981443172439904e-05,
+      "loss": 0.1162,
+      "step": 166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.981205252301497e-05,
+      "loss": 0.2276,
+      "step": 167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.980965831123589e-05,
+      "loss": 0.0856,
+      "step": 168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9807249092724453e-05,
+      "loss": 0.1338,
+      "step": 169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9804824871166254e-05,
+      "loss": 0.15,
+      "step": 170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9802385650269854e-05,
+      "loss": 0.1879,
+      "step": 171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9799931433766746e-05,
+      "loss": 0.1755,
+      "step": 172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9797462225411372e-05,
+      "loss": 0.1776,
+      "step": 173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9794978028981106e-05,
+      "loss": 0.1756,
+      "step": 174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.979247884827625e-05,
+      "loss": 0.1859,
+      "step": 175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9789964687120032e-05,
+      "loss": 0.0871,
+      "step": 176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9787435549358592e-05,
+      "loss": 0.1152,
+      "step": 177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9784891438860987e-05,
+      "loss": 0.0962,
+      "step": 178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9782332359519173e-05,
+      "loss": 0.1387,
+      "step": 179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9779758315248006e-05,
+      "loss": 0.0723,
+      "step": 180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.977716930998524e-05,
+      "loss": 0.0984,
+      "step": 181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.977456534769151e-05,
+      "loss": 0.1847,
+      "step": 182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.977194643235034e-05,
+      "loss": 0.1281,
+      "step": 183
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9769312567968124e-05,
+      "loss": 0.0519,
+      "step": 184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9766663758574122e-05,
+      "loss": 0.1285,
+      "step": 185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9764000008220464e-05,
+      "loss": 0.0915,
+      "step": 186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9761321320982133e-05,
+      "loss": 0.1971,
+      "step": 187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9758627700956957e-05,
+      "loss": 0.1629,
+      "step": 188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9755919152265618e-05,
+      "loss": 0.1345,
+      "step": 189
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.975319567905163e-05,
+      "loss": 0.1092,
+      "step": 190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9750457285481334e-05,
+      "loss": 0.2356,
+      "step": 191
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.974770397574391e-05,
+      "loss": 0.2002,
+      "step": 192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9744935754051342e-05,
+      "loss": 0.1391,
+      "step": 193
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9742152624638437e-05,
+      "loss": 0.0867,
+      "step": 194
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9739354591762798e-05,
+      "loss": 0.1581,
+      "step": 195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9736541659704833e-05,
+      "loss": 0.0758,
+      "step": 196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9733713832767743e-05,
+      "loss": 0.1565,
+      "step": 197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9730871115277513e-05,
+      "loss": 0.1556,
+      "step": 198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.972801351158291e-05,
+      "loss": 0.1403,
+      "step": 199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9725141026055473e-05,
+      "loss": 0.1265,
+      "step": 200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9722253663089498e-05,
+      "loss": 0.1459,
+      "step": 201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9719351427102054e-05,
+      "loss": 0.1058,
+      "step": 202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.971643432253296e-05,
+      "loss": 0.1711,
+      "step": 203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9713502353844777e-05,
+      "loss": 0.1078,
+      "step": 204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9710555525522802e-05,
+      "loss": 0.1217,
+      "step": 205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.970759384207507e-05,
+      "loss": 0.1879,
+      "step": 206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9704617308032338e-05,
+      "loss": 0.0504,
+      "step": 207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9701625927948083e-05,
+      "loss": 0.1014,
+      "step": 208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9698619706398495e-05,
+      "loss": 0.1908,
+      "step": 209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9695598647982467e-05,
+      "loss": 0.2199,
+      "step": 210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.969256275732159e-05,
+      "loss": 0.098,
+      "step": 211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9689512039060133e-05,
+      "loss": 0.1466,
+      "step": 212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.968644649786507e-05,
+      "loss": 0.1121,
+      "step": 213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9683366138426033e-05,
+      "loss": 0.1815,
+      "step": 214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9680270965455343e-05,
+      "loss": 0.1825,
+      "step": 215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9677160983687953e-05,
+      "loss": 0.1633,
+      "step": 216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9674036197881493e-05,
+      "loss": 0.0803,
+      "step": 217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9670896612816236e-05,
+      "loss": 0.1235,
+      "step": 218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9667742233295093e-05,
+      "loss": 0.1826,
+      "step": 219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9664573064143604e-05,
+      "loss": 0.0946,
+      "step": 220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.966138911020994e-05,
+      "loss": 0.0971,
+      "step": 221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9658190376364884e-05,
+      "loss": 0.1731,
+      "step": 222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9654976867501836e-05,
+      "loss": 0.1088,
+      "step": 223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9651748588536793e-05,
+      "loss": 0.0874,
+      "step": 224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9648505544408343e-05,
+      "loss": 0.1233,
+      "step": 225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.964524774007768e-05,
+      "loss": 0.1223,
+      "step": 226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.964197518052855e-05,
+      "loss": 0.1017,
+      "step": 227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9638687870767298e-05,
+      "loss": 0.1261,
+      "step": 228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.963538581582282e-05,
+      "loss": 0.1445,
+      "step": 229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9632069020746574e-05,
+      "loss": 0.0581,
+      "step": 230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.962873749061256e-05,
+      "loss": 0.0917,
+      "step": 231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9625391230517325e-05,
+      "loss": 0.1257,
+      "step": 232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9622030245579956e-05,
+      "loss": 0.0842,
+      "step": 233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.961865454094205e-05,
+      "loss": 0.15,
+      "step": 234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9615264121767742e-05,
+      "loss": 0.1203,
+      "step": 235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9611858993243658e-05,
+      "loss": 0.1194,
+      "step": 236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9608439160578938e-05,
+      "loss": 0.1056,
+      "step": 237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9605004629005214e-05,
+      "loss": 0.1095,
+      "step": 238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.96015554037766e-05,
+      "loss": 0.0791,
+      "step": 239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9598091490169696e-05,
+      "loss": 0.1023,
+      "step": 240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9594612893483563e-05,
+      "loss": 0.0913,
+      "step": 241
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.959111961903973e-05,
+      "loss": 0.1944,
+      "step": 242
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9587611672182184e-05,
+      "loss": 0.073,
+      "step": 243
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.958408905827735e-05,
+      "loss": 0.1257,
+      "step": 244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.958055178271409e-05,
+      "loss": 0.1372,
+      "step": 245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9576999850903698e-05,
+      "loss": 0.1009,
+      "step": 246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.957343326827989e-05,
+      "loss": 0.0898,
+      "step": 247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9569852040298798e-05,
+      "loss": 0.1426,
+      "step": 248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.956625617243895e-05,
+      "loss": 0.1383,
+      "step": 249
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9562645670201278e-05,
+      "loss": 0.124,
+      "step": 250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9559020539109093e-05,
+      "loss": 0.1039,
+      "step": 251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.955538078470809e-05,
+      "loss": 0.1366,
+      "step": 252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9551726412566342e-05,
+      "loss": 0.1481,
+      "step": 253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9548057428274267e-05,
+      "loss": 0.0805,
+      "step": 254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.954437383744465e-05,
+      "loss": 0.1313,
+      "step": 255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9540675645712622e-05,
+      "loss": 0.1083,
+      "step": 256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.953696285873564e-05,
+      "loss": 0.1213,
+      "step": 257
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9533235482193492e-05,
+      "loss": 0.1307,
+      "step": 258
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9529493521788292e-05,
+      "loss": 0.1444,
+      "step": 259
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9525736983244458e-05,
+      "loss": 0.086,
+      "step": 260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9521965872308713e-05,
+      "loss": 0.1107,
+      "step": 261
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9518180194750063e-05,
+      "loss": 0.1495,
+      "step": 262
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.951437995635982e-05,
+      "loss": 0.1734,
+      "step": 263
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.0602,
+      "step": 264
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9506735820361065e-05,
+      "loss": 0.0648,
+      "step": 265
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9502891934446505e-05,
+      "loss": 0.1865,
+      "step": 266
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.949903351108819e-05,
+      "loss": 0.1199,
+      "step": 267
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9495160556188704e-05,
+      "loss": 0.1436,
+      "step": 268
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9491273075672867e-05,
+      "loss": 0.1472,
+      "step": 269
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.948737107548771e-05,
+      "loss": 0.0762,
+      "step": 270
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9483454561602487e-05,
+      "loss": 0.1376,
+      "step": 271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9479523540008635e-05,
+      "loss": 0.0892,
+      "step": 272
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9475578016719807e-05,
+      "loss": 0.0447,
+      "step": 273
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.947161799777183e-05,
+      "loss": 0.1476,
+      "step": 274
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9467643489222704e-05,
+      "loss": 0.1137,
+      "step": 275
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.94636544971526e-05,
+      "loss": 0.0809,
+      "step": 276
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9459651027663847e-05,
+      "loss": 0.111,
+      "step": 277
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9455633086880916e-05,
+      "loss": 0.1086,
+      "step": 278
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9451600680950422e-05,
+      "loss": 0.2091,
+      "step": 279
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.94475538160411e-05,
+      "loss": 0.0955,
+      "step": 280
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9443492498343817e-05,
+      "loss": 0.1457,
+      "step": 281
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9439416734071535e-05,
+      "loss": 0.0693,
+      "step": 282
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9435326529459327e-05,
+      "loss": 0.0794,
+      "step": 283
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9431221890764352e-05,
+      "loss": 0.0884,
+      "step": 284
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9427102824265858e-05,
+      "loss": 0.0554,
+      "step": 285
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.942296933626515e-05,
+      "loss": 0.1054,
+      "step": 286
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.941882143308561e-05,
+      "loss": 0.0549,
+      "step": 287
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9414659121072657e-05,
+      "loss": 0.1097,
+      "step": 288
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9410482406593766e-05,
+      "loss": 0.1662,
+      "step": 289
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.940629129603844e-05,
+      "loss": 0.1489,
+      "step": 290
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9402085795818202e-05,
+      "loss": 0.0576,
+      "step": 291
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9397865912366594e-05,
+      "loss": 0.1252,
+      "step": 292
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.939363165213915e-05,
+      "loss": 0.0917,
+      "step": 293
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9389383021613422e-05,
+      "loss": 0.0805,
+      "step": 294
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9385120027288914e-05,
+      "loss": 0.0844,
+      "step": 295
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.938084267568712e-05,
+      "loss": 0.1263,
+      "step": 296
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9376550973351503e-05,
+      "loss": 0.1149,
+      "step": 297
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9372244926847467e-05,
+      "loss": 0.0656,
+      "step": 298
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.936792454276237e-05,
+      "loss": 0.0456,
+      "step": 299
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9363589827705494e-05,
+      "loss": 0.0788,
+      "step": 300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9359240788308057e-05,
+      "loss": 0.1399,
+      "step": 301
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9354877431223173e-05,
+      "loss": 0.1597,
+      "step": 302
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9350499763125875e-05,
+      "loss": 0.1422,
+      "step": 303
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9346107790713086e-05,
+      "loss": 0.0797,
+      "step": 304
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.93417015207036e-05,
+      "loss": 0.0886,
+      "step": 305
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.93372809598381e-05,
+      "loss": 0.0373,
+      "step": 306
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9332846114879113e-05,
+      "loss": 0.0953,
+      "step": 307
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9328396992611035e-05,
+      "loss": 0.116,
+      "step": 308
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9323933599840093e-05,
+      "loss": 0.1247,
+      "step": 309
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9319455943394347e-05,
+      "loss": 0.0961,
+      "step": 310
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.931496403012368e-05,
+      "loss": 0.0402,
+      "step": 311
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.931045786689978e-05,
+      "loss": 0.1388,
+      "step": 312
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.930593746061614e-05,
+      "loss": 0.0546,
+      "step": 313
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9301402818188036e-05,
+      "loss": 0.127,
+      "step": 314
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9296853946552532e-05,
+      "loss": 0.0605,
+      "step": 315
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9292290852668448e-05,
+      "loss": 0.0812,
+      "step": 316
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9287713543516367e-05,
+      "loss": 0.0823,
+      "step": 317
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.928312202609862e-05,
+      "loss": 0.0708,
+      "step": 318
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.927851630743927e-05,
+      "loss": 0.0824,
+      "step": 319
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9273896394584103e-05,
+      "loss": 0.1486,
+      "step": 320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.926926229460063e-05,
+      "loss": 0.1461,
+      "step": 321
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9264614014578057e-05,
+      "loss": 0.1395,
+      "step": 322
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9259951561627274e-05,
+      "loss": 0.044,
+      "step": 323
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9255274942880872e-05,
+      "loss": 0.0879,
+      "step": 324
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9250584165493102e-05,
+      "loss": 0.1222,
+      "step": 325
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.924587923663987e-05,
+      "loss": 0.156,
+      "step": 326
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.924116016351874e-05,
+      "loss": 0.041,
+      "step": 327
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9236426953348913e-05,
+      "loss": 0.1211,
+      "step": 328
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9231679613371208e-05,
+      "loss": 0.1382,
+      "step": 329
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9226918150848067e-05,
+      "loss": 0.081,
+      "step": 330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.922214257306354e-05,
+      "loss": 0.1077,
+      "step": 331
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.921735288732326e-05,
+      "loss": 0.087,
+      "step": 332
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9212549100954452e-05,
+      "loss": 0.1352,
+      "step": 333
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9207731221305905e-05,
+      "loss": 0.0516,
+      "step": 334
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9202899255747967e-05,
+      "loss": 0.1657,
+      "step": 335
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9198053211672544e-05,
+      "loss": 0.1094,
+      "step": 336
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9193193096493074e-05,
+      "loss": 0.0386,
+      "step": 337
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9188318917644514e-05,
+      "loss": 0.1166,
+      "step": 338
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.918343068258335e-05,
+      "loss": 0.0676,
+      "step": 339
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9178528398787553e-05,
+      "loss": 0.1164,
+      "step": 340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.91736120737566e-05,
+      "loss": 0.0721,
+      "step": 341
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9168681715011444e-05,
+      "loss": 0.09,
+      "step": 342
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9163737330094508e-05,
+      "loss": 0.1571,
+      "step": 343
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9158778926569666e-05,
+      "loss": 0.1207,
+      "step": 344
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9153806512022248e-05,
+      "loss": 0.0383,
+      "step": 345
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9148820094059004e-05,
+      "loss": 0.0906,
+      "step": 346
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9143819680308118e-05,
+      "loss": 0.1266,
+      "step": 347
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9138805278419182e-05,
+      "loss": 0.0671,
+      "step": 348
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9133776896063182e-05,
+      "loss": 0.1038,
+      "step": 349
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9128734540932494e-05,
+      "loss": 0.0753,
+      "step": 350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9123678220740877e-05,
+      "loss": 0.1383,
+      "step": 351
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9118607943223438e-05,
+      "loss": 0.1066,
+      "step": 352
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9113523716136655e-05,
+      "loss": 0.0892,
+      "step": 353
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9108425547258328e-05,
+      "loss": 0.0605,
+      "step": 354
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9103313444387595e-05,
+      "loss": 0.0627,
+      "step": 355
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.909818741534491e-05,
+      "loss": 0.0594,
+      "step": 356
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9093047467972022e-05,
+      "loss": 0.0932,
+      "step": 357
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9087893610131986e-05,
+      "loss": 0.1587,
+      "step": 358
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9082725849709125e-05,
+      "loss": 0.1035,
+      "step": 359
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.907754419460904e-05,
+      "loss": 0.0765,
+      "step": 360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.907234865275858e-05,
+      "loss": 0.1306,
+      "step": 361
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9067139232105844e-05,
+      "loss": 0.1227,
+      "step": 362
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9061915940620156e-05,
+      "loss": 0.1382,
+      "step": 363
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9056678786292065e-05,
+      "loss": 0.0361,
+      "step": 364
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9051427777133328e-05,
+      "loss": 0.0652,
+      "step": 365
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.904616292117689e-05,
+      "loss": 0.1012,
+      "step": 366
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9040884226476883e-05,
+      "loss": 0.1308,
+      "step": 367
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.903559170110861e-05,
+      "loss": 0.0499,
+      "step": 368
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9030285353168536e-05,
+      "loss": 0.1324,
+      "step": 369
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9024965190774262e-05,
+      "loss": 0.0695,
+      "step": 370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9019631222064535e-05,
+      "loss": 0.1324,
+      "step": 371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9014283455199204e-05,
+      "loss": 0.1224,
+      "step": 372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.900892189835925e-05,
+      "loss": 0.0933,
+      "step": 373
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.900354655974672e-05,
+      "loss": 0.1159,
+      "step": 374
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.899815744758478e-05,
+      "loss": 0.135,
+      "step": 375
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.899275457011764e-05,
+      "loss": 0.0607,
+      "step": 376
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.8987337935610576e-05,
+      "loss": 0.0418,
+      "step": 377
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.898190755234991e-05,
+      "loss": 0.0736,
+      "step": 378
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8976463428642997e-05,
+      "loss": 0.1622,
+      "step": 379
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8971005572818213e-05,
+      "loss": 0.1026,
+      "step": 380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8965533993224935e-05,
+      "loss": 0.094,
+      "step": 381
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8960048698233544e-05,
+      "loss": 0.1309,
+      "step": 382
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8954549696235397e-05,
+      "loss": 0.1353,
+      "step": 383
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.894903699564282e-05,
+      "loss": 0.079,
+      "step": 384
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.8943510604889094e-05,
+      "loss": 0.0833,
+      "step": 385
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.8937970532428446e-05,
+      "loss": 0.0593,
+      "step": 386
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8932416786736035e-05,
+      "loss": 0.0726,
+      "step": 387
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8926849376307933e-05,
+      "loss": 0.0686,
+      "step": 388
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.892126830966111e-05,
+      "loss": 0.1248,
+      "step": 389
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.8915673595333443e-05,
+      "loss": 0.0614,
+      "step": 390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 0.0783,
+      "step": 391
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.8904443257891428e-05,
+      "loss": 0.0517,
+      "step": 392
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.889880765195715e-05,
+      "loss": 0.0365,
+      "step": 393
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.889315843270215e-05,
+      "loss": 0.0782,
+      "step": 394
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.8887495608768557e-05,
+      "loss": 0.1206,
+      "step": 395
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.888181918881931e-05,
+      "loss": 0.1181,
+      "step": 396
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.8876129181538147e-05,
+      "loss": 0.1113,
+      "step": 397
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.8870425595629597e-05,
+      "loss": 0.0795,
+      "step": 398
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.8864708439818952e-05,
+      "loss": 0.0604,
+      "step": 399
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.8858977722852273e-05,
+      "loss": 0.1092,
+      "step": 400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.8853233453496362e-05,
+      "loss": 0.0434,
+      "step": 401
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.8847475640538753e-05,
+      "loss": 0.065,
+      "step": 402
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.8841704292787696e-05,
+      "loss": 0.1296,
+      "step": 403
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.883591941907216e-05,
+      "loss": 0.1289,
+      "step": 404
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.883012102824178e-05,
+      "loss": 0.0657,
+      "step": 405
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8824309129166897e-05,
+      "loss": 0.1089,
+      "step": 406
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8818483730738508e-05,
+      "loss": 0.1143,
+      "step": 407
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.881264484186825e-05,
+      "loss": 0.0449,
+      "step": 408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8806792471488404e-05,
+      "loss": 0.1095,
+      "step": 409
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8800926628551884e-05,
+      "loss": 0.1603,
+      "step": 410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8795047322032207e-05,
+      "loss": 0.1394,
+      "step": 411
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.878915456092348e-05,
+      "loss": 0.1892,
+      "step": 412
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8783248354240405e-05,
+      "loss": 0.0253,
+      "step": 413
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8777328711018244e-05,
+      "loss": 0.1942,
+      "step": 414
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.877139564031282e-05,
+      "loss": 0.0829,
+      "step": 415
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.876544915120049e-05,
+      "loss": 0.041,
+      "step": 416
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8759489252778144e-05,
+      "loss": 0.1207,
+      "step": 417
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8753515954163188e-05,
+      "loss": 0.0423,
+      "step": 418
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.874752926449352e-05,
+      "loss": 0.0582,
+      "step": 419
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8741529192927528e-05,
+      "loss": 0.0891,
+      "step": 420
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.873551574864407e-05,
+      "loss": 0.1276,
+      "step": 421
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8729488940842468e-05,
+      "loss": 0.0681,
+      "step": 422
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8723448778742473e-05,
+      "loss": 0.0255,
+      "step": 423
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.871739527158428e-05,
+      "loss": 0.1012,
+      "step": 424
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8711328428628492e-05,
+      "loss": 0.1349,
+      "step": 425
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.870524825915611e-05,
+      "loss": 0.0447,
+      "step": 426
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.869915477246853e-05,
+      "loss": 0.0569,
+      "step": 427
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.8693047977887517e-05,
+      "loss": 0.0753,
+      "step": 428
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.868692788475519e-05,
+      "loss": 0.069,
+      "step": 429
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8680794502434018e-05,
+      "loss": 0.1148,
+      "step": 430
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.86746478403068e-05,
+      "loss": 0.0786,
+      "step": 431
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.866848790777664e-05,
+      "loss": 0.1318,
+      "step": 432
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.866231471426696e-05,
+      "loss": 0.0843,
+      "step": 433
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8656128269221452e-05,
+      "loss": 0.1357,
+      "step": 434
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8649928582104097e-05,
+      "loss": 0.1273,
+      "step": 435
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8643715662399108e-05,
+      "loss": 0.0843,
+      "step": 436
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.863748951961097e-05,
+      "loss": 0.0452,
+      "step": 437
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8631250163264383e-05,
+      "loss": 0.0524,
+      "step": 438
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8624997602904256e-05,
+      "loss": 0.0838,
+      "step": 439
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8618731848095706e-05,
+      "loss": 0.132,
+      "step": 440
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.861245290842403e-05,
+      "loss": 0.0337,
+      "step": 441
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8606160793494697e-05,
+      "loss": 0.0486,
+      "step": 442
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.859985551293334e-05,
+      "loss": 0.1146,
+      "step": 443
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8593537076385704e-05,
+      "loss": 0.0983,
+      "step": 444
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8587205493517703e-05,
+      "loss": 0.0716,
+      "step": 445
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.858086077401532e-05,
+      "loss": 0.0814,
+      "step": 446
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8574502927584664e-05,
+      "loss": 0.1238,
+      "step": 447
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8568131963951918e-05,
+      "loss": 0.1534,
+      "step": 448
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8561747892863312e-05,
+      "loss": 0.0741,
+      "step": 449
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.855535072408516e-05,
+      "loss": 0.0846,
+      "step": 450
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8548940467403792e-05,
+      "loss": 0.1242,
+      "step": 451
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.854251713262556e-05,
+      "loss": 0.0985,
+      "step": 452
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8536080729576832e-05,
+      "loss": 0.0334,
+      "step": 453
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8529631268103964e-05,
+      "loss": 0.1528,
+      "step": 454
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8523168758073283e-05,
+      "loss": 0.0844,
+      "step": 455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8516693209371084e-05,
+      "loss": 0.0918,
+      "step": 456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.851020463190361e-05,
+      "loss": 0.097,
+      "step": 457
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8503703035597026e-05,
+      "loss": 0.1199,
+      "step": 458
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8497188430397424e-05,
+      "loss": 0.0407,
+      "step": 459
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.849066082627079e-05,
+      "loss": 0.081,
+      "step": 460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8484120233202996e-05,
+      "loss": 0.0857,
+      "step": 461
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8477566661199788e-05,
+      "loss": 0.2011,
+      "step": 462
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.847100012028677e-05,
+      "loss": 0.0534,
+      "step": 463
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.846442062050937e-05,
+      "loss": 0.0919,
+      "step": 464
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.845782817193286e-05,
+      "loss": 0.0836,
+      "step": 465
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8451222784642307e-05,
+      "loss": 0.0414,
+      "step": 466
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.844460446874258e-05,
+      "loss": 0.0386,
+      "step": 467
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.843797323435832e-05,
+      "loss": 0.0713,
+      "step": 468
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8431329091633935e-05,
+      "loss": 0.122,
+      "step": 469
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8424672050733577e-05,
+      "loss": 0.0619,
+      "step": 470
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8418002121841133e-05,
+      "loss": 0.0871,
+      "step": 471
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.84113193151602e-05,
+      "loss": 0.0376,
+      "step": 472
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8404623640914083e-05,
+      "loss": 0.0635,
+      "step": 473
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.839791510934577e-05,
+      "loss": 0.0704,
+      "step": 474
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.839119373071791e-05,
+      "loss": 0.0705,
+      "step": 475
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8384459515312813e-05,
+      "loss": 0.047,
+      "step": 476
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8377712473432425e-05,
+      "loss": 0.0379,
+      "step": 477
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.837095261539832e-05,
+      "loss": 0.097,
+      "step": 478
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8364179951551658e-05,
+      "loss": 0.0346,
+      "step": 479
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8357394492253216e-05,
+      "loss": 0.0783,
+      "step": 480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8350596247883324e-05,
+      "loss": 0.0587,
+      "step": 481
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8343785228841883e-05,
+      "loss": 0.0397,
+      "step": 482
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8336961445548333e-05,
+      "loss": 0.1604,
+      "step": 483
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8330124908441638e-05,
+      "loss": 0.0601,
+      "step": 484
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8323275627980272e-05,
+      "loss": 0.1035,
+      "step": 485
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8316413614642213e-05,
+      "loss": 0.0675,
+      "step": 486
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8309538878924905e-05,
+      "loss": 0.093,
+      "step": 487
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.830265143134526e-05,
+      "loss": 0.1055,
+      "step": 488
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.829575128243964e-05,
+      "loss": 0.1454,
+      "step": 489
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8288838442763838e-05,
+      "loss": 0.0682,
+      "step": 490
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8281912922893053e-05,
+      "loss": 0.0359,
+      "step": 491
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.827497473342189e-05,
+      "loss": 0.0482,
+      "step": 492
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8268023884964328e-05,
+      "loss": 0.0795,
+      "step": 493
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.826106038815372e-05,
+      "loss": 0.1432,
+      "step": 494
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.825408425364276e-05,
+      "loss": 0.1288,
+      "step": 495
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8247095492103486e-05,
+      "loss": 0.0918,
+      "step": 496
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8240094114227243e-05,
+      "loss": 0.0667,
+      "step": 497
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8233080130724677e-05,
+      "loss": 0.0519,
+      "step": 498
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8226053552325726e-05,
+      "loss": 0.096,
+      "step": 499
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8219014389779586e-05,
+      "loss": 0.0714,
+      "step": 500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.821196265385471e-05,
+      "loss": 0.0603,
+      "step": 501
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8204898355338785e-05,
+      "loss": 0.0906,
+      "step": 502
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8197821505038714e-05,
+      "loss": 0.0941,
+      "step": 503
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8190732113780604e-05,
+      "loss": 0.066,
+      "step": 504
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8183630192409746e-05,
+      "loss": 0.1328,
+      "step": 505
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8176515751790593e-05,
+      "loss": 0.0645,
+      "step": 506
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8169388802806765e-05,
+      "loss": 0.077,
+      "step": 507
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8162249356361002e-05,
+      "loss": 0.0907,
+      "step": 508
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8155097423375175e-05,
+      "loss": 0.066,
+      "step": 509
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8147933014790245e-05,
+      "loss": 0.0872,
+      "step": 510
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8140756141566267e-05,
+      "loss": 0.0689,
+      "step": 511
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.813356681468236e-05,
+      "loss": 0.071,
+      "step": 512
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8126365045136692e-05,
+      "loss": 0.0689,
+      "step": 513
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8119150843946475e-05,
+      "loss": 0.0605,
+      "step": 514
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8111924222147927e-05,
+      "loss": 0.0388,
+      "step": 515
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.810468519079628e-05,
+      "loss": 0.0458,
+      "step": 516
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8097433760965737e-05,
+      "loss": 0.0551,
+      "step": 517
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.1633,
+      "step": 518
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8082893750259626e-05,
+      "loss": 0.1008,
+      "step": 519
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8075605191627242e-05,
+      "loss": 0.1338,
+      "step": 520
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.80683042790023e-05,
+      "loss": 0.1378,
+      "step": 521
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8060991023553676e-05,
+      "loss": 0.0761,
+      "step": 522
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.805366543646913e-05,
+      "loss": 0.0555,
+      "step": 523
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.804632752895527e-05,
+      "loss": 0.1346,
+      "step": 524
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8038977312237583e-05,
+      "loss": 0.0439,
+      "step": 525
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8031614797560352e-05,
+      "loss": 0.0417,
+      "step": 526
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8024239996186704e-05,
+      "loss": 0.052,
+      "step": 527
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.801685291939854e-05,
+      "loss": 0.1331,
+      "step": 528
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8009453578496546e-05,
+      "loss": 0.0933,
+      "step": 529
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8002041984800173e-05,
+      "loss": 0.0414,
+      "step": 530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7994618149647616e-05,
+      "loss": 0.133,
+      "step": 531
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7987182084395795e-05,
+      "loss": 0.0284,
+      "step": 532
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7979733800420338e-05,
+      "loss": 0.0603,
+      "step": 533
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.797227330911557e-05,
+      "loss": 0.0519,
+      "step": 534
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.796480062189448e-05,
+      "loss": 0.0634,
+      "step": 535
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7957315750188736e-05,
+      "loss": 0.0906,
+      "step": 536
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7949818705448617e-05,
+      "loss": 0.0318,
+      "step": 537
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7942309499143045e-05,
+      "loss": 0.0687,
+      "step": 538
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.793478814275954e-05,
+      "loss": 0.0843,
+      "step": 539
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.792725464780421e-05,
+      "loss": 0.0759,
+      "step": 540
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.791970902580173e-05,
+      "loss": 0.0521,
+      "step": 541
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7912151288295328e-05,
+      "loss": 0.0998,
+      "step": 542
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7904581446846772e-05,
+      "loss": 0.0636,
+      "step": 543
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7896999513036335e-05,
+      "loss": 0.0722,
+      "step": 544
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.78894054984628e-05,
+      "loss": 0.1129,
+      "step": 545
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.7881799414743428e-05,
+      "loss": 0.0624,
+      "step": 546
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7874181273513932e-05,
+      "loss": 0.0626,
+      "step": 547
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7866551086428493e-05,
+      "loss": 0.067,
+      "step": 548
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7858908865159694e-05,
+      "loss": 0.1507,
+      "step": 549
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.785125462139855e-05,
+      "loss": 0.0607,
+      "step": 550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7843588366854456e-05,
+      "loss": 0.0854,
+      "step": 551
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.783591011325518e-05,
+      "loss": 0.1161,
+      "step": 552
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.782821987234685e-05,
+      "loss": 0.0767,
+      "step": 553
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.782051765589394e-05,
+      "loss": 0.0968,
+      "step": 554
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7812803475679224e-05,
+      "loss": 0.0624,
+      "step": 555
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7805077343503794e-05,
+      "loss": 0.0115,
+      "step": 556
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.779733927118702e-05,
+      "loss": 0.0423,
+      "step": 557
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7789589270566548e-05,
+      "loss": 0.0675,
+      "step": 558
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7781827353498257e-05,
+      "loss": 0.0609,
+      "step": 559
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7774053531856258e-05,
+      "loss": 0.0617,
+      "step": 560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7766267817532886e-05,
+      "loss": 0.1028,
+      "step": 561
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7758470222438656e-05,
+      "loss": 0.0796,
+      "step": 562
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7750660758502266e-05,
+      "loss": 0.0289,
+      "step": 563
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7742839437670563e-05,
+      "loss": 0.0447,
+      "step": 564
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.773500627190854e-05,
+      "loss": 0.0904,
+      "step": 565
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.772716127319931e-05,
+      "loss": 0.0613,
+      "step": 566
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7719304453544086e-05,
+      "loss": 0.0706,
+      "step": 567
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.771143582496216e-05,
+      "loss": 0.1159,
+      "step": 568
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7703555399490894e-05,
+      "loss": 0.0854,
+      "step": 569
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7695663189185703e-05,
+      "loss": 0.0728,
+      "step": 570
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.768775920612001e-05,
+      "loss": 0.0426,
+      "step": 571
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.767984346238528e-05,
+      "loss": 0.0126,
+      "step": 572
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7671915970090934e-05,
+      "loss": 0.0441,
+      "step": 573
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7663976741364394e-05,
+      "loss": 0.051,
+      "step": 574
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.765602578835102e-05,
+      "loss": 0.1054,
+      "step": 575
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7648063123214112e-05,
+      "loss": 0.0899,
+      "step": 576
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.764008875813489e-05,
+      "loss": 0.0907,
+      "step": 577
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7632102705312472e-05,
+      "loss": 0.0405,
+      "step": 578
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7624104976963848e-05,
+      "loss": 0.0979,
+      "step": 579
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7616095585323882e-05,
+      "loss": 0.0846,
+      "step": 580
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7608074542645268e-05,
+      "loss": 0.2115,
+      "step": 581
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7600041861198533e-05,
+      "loss": 0.0782,
+      "step": 582
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7591997553272003e-05,
+      "loss": 0.0437,
+      "step": 583
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.758394163117179e-05,
+      "loss": 0.0838,
+      "step": 584
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7575874107221785e-05,
+      "loss": 0.0682,
+      "step": 585
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7567794993763606e-05,
+      "loss": 0.078,
+      "step": 586
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7559704303156624e-05,
+      "loss": 0.0843,
+      "step": 587
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.75516020477779e-05,
+      "loss": 0.0919,
+      "step": 588
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.75434882400222e-05,
+      "loss": 0.0836,
+      "step": 589
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7535362892301953e-05,
+      "loss": 0.0956,
+      "step": 590
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.752722601704726e-05,
+      "loss": 0.0648,
+      "step": 591
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.751907762670583e-05,
+      "loss": 0.0451,
+      "step": 592
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.751091773374301e-05,
+      "loss": 0.0687,
+      "step": 593
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.750274635064173e-05,
+      "loss": 0.062,
+      "step": 594
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.749456348990251e-05,
+      "loss": 0.0856,
+      "step": 595
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.748636916404341e-05,
+      "loss": 0.0588,
+      "step": 596
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7478163385600054e-05,
+      "loss": 0.1596,
+      "step": 597
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7469946167125563e-05,
+      "loss": 0.0485,
+      "step": 598
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7461717521190573e-05,
+      "loss": 0.0884,
+      "step": 599
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.745347746038319e-05,
+      "loss": 0.0605,
+      "step": 600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7445225997308998e-05,
+      "loss": 0.0468,
+      "step": 601
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.743696314459101e-05,
+      "loss": 0.0681,
+      "step": 602
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.742868891486967e-05,
+      "loss": 0.0898,
+      "step": 603
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7420403320802827e-05,
+      "loss": 0.0628,
+      "step": 604
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.741210637506571e-05,
+      "loss": 0.1006,
+      "step": 605
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7403798090350913e-05,
+      "loss": 0.0968,
+      "step": 606
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.739547847936839e-05,
+      "loss": 0.0217,
+      "step": 607
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7387147554845402e-05,
+      "loss": 0.0612,
+      "step": 608
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.737880532952653e-05,
+      "loss": 0.0756,
+      "step": 609
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.737045181617364e-05,
+      "loss": 0.0521,
+      "step": 610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7362087027565866e-05,
+      "loss": 0.033,
+      "step": 611
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7353710976499592e-05,
+      "loss": 0.0792,
+      "step": 612
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.734532367578843e-05,
+      "loss": 0.0692,
+      "step": 613
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7336925138263195e-05,
+      "loss": 0.0692,
+      "step": 614
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.732851537677191e-05,
+      "loss": 0.0552,
+      "step": 615
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.732009440417975e-05,
+      "loss": 0.0449,
+      "step": 616
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.731166223336905e-05,
+      "loss": 0.0982,
+      "step": 617
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.730321887723927e-05,
+      "loss": 0.0616,
+      "step": 618
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7294764348706992e-05,
+      "loss": 0.0719,
+      "step": 619
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7286298660705877e-05,
+      "loss": 0.0415,
+      "step": 620
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.727782182618666e-05,
+      "loss": 0.0459,
+      "step": 621
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7269333858117137e-05,
+      "loss": 0.0565,
+      "step": 622
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.726083476948213e-05,
+      "loss": 0.0207,
+      "step": 623
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.725232457328347e-05,
+      "loss": 0.0702,
+      "step": 624
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.724380328253998e-05,
+      "loss": 0.1005,
+      "step": 625
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7235270910287462e-05,
+      "loss": 0.0633,
+      "step": 626
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.722672746957867e-05,
+      "loss": 0.0525,
+      "step": 627
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7218172973483285e-05,
+      "loss": 0.1131,
+      "step": 628
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.72096074350879e-05,
+      "loss": 0.0718,
+      "step": 629
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7201030867496005e-05,
+      "loss": 0.0841,
+      "step": 630
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7192443283827963e-05,
+      "loss": 0.0327,
+      "step": 631
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7183844697220984e-05,
+      "loss": 0.0503,
+      "step": 632
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7175235120829118e-05,
+      "loss": 0.0469,
+      "step": 633
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7166614567823214e-05,
+      "loss": 0.0377,
+      "step": 634
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7157983051390926e-05,
+      "loss": 0.0511,
+      "step": 635
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7149340584736677e-05,
+      "loss": 0.0708,
+      "step": 636
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7140687181081635e-05,
+      "loss": 0.0308,
+      "step": 637
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7132022853663706e-05,
+      "loss": 0.0511,
+      "step": 638
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7123347615737504e-05,
+      "loss": 0.0472,
+      "step": 639
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.711466148057433e-05,
+      "loss": 0.1232,
+      "step": 640
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.710596446146217e-05,
+      "loss": 0.0449,
+      "step": 641
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.709725657170565e-05,
+      "loss": 0.0594,
+      "step": 642
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7088537824626014e-05,
+      "loss": 0.1081,
+      "step": 643
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.707980823356114e-05,
+      "loss": 0.0527,
+      "step": 644
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.0274,
+      "step": 645
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7062316572910048e-05,
+      "loss": 0.0214,
+      "step": 646
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7053554530082427e-05,
+      "loss": 0.0435,
+      "step": 647
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.704478169678671e-05,
+      "loss": 0.0483,
+      "step": 648
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7035998086443502e-05,
+      "loss": 0.0772,
+      "step": 649
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7027203712489902e-05,
+      "loss": 0.0667,
+      "step": 650
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7018398588379464e-05,
+      "loss": 0.0798,
+      "step": 651
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7009582727582197e-05,
+      "loss": 0.0521,
+      "step": 652
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7000756143584524e-05,
+      "loss": 0.0688,
+      "step": 653
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6991918849889283e-05,
+      "loss": 0.061,
+      "step": 654
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.698307086001569e-05,
+      "loss": 0.0372,
+      "step": 655
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6974212187499325e-05,
+      "loss": 0.0752,
+      "step": 656
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6965342845892112e-05,
+      "loss": 0.0659,
+      "step": 657
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.695646284876229e-05,
+      "loss": 0.0547,
+      "step": 658
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.694757220969441e-05,
+      "loss": 0.0743,
+      "step": 659
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.6938670942289292e-05,
+      "loss": 0.039,
+      "step": 660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.6929759060164018e-05,
+      "loss": 0.0751,
+      "step": 661
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6920836576951915e-05,
+      "loss": 0.0592,
+      "step": 662
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6911903506302524e-05,
+      "loss": 0.0441,
+      "step": 663
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6902959861881572e-05,
+      "loss": 0.0495,
+      "step": 664
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.689400565737098e-05,
+      "loss": 0.0968,
+      "step": 665
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6885040906468807e-05,
+      "loss": 0.0406,
+      "step": 666
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.687606562288926e-05,
+      "loss": 0.07,
+      "step": 667
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6867079820362644e-05,
+      "loss": 0.035,
+      "step": 668
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6858083512635372e-05,
+      "loss": 0.0714,
+      "step": 669
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6849076713469914e-05,
+      "loss": 0.0524,
+      "step": 670
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6840059436644796e-05,
+      "loss": 0.0662,
+      "step": 671
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6831031695954572e-05,
+      "loss": 0.0945,
+      "step": 672
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6821993505209806e-05,
+      "loss": 0.0609,
+      "step": 673
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.681294487823704e-05,
+      "loss": 0.03,
+      "step": 674
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6803885828878798e-05,
+      "loss": 0.0434,
+      "step": 675
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6794816370993527e-05,
+      "loss": 0.0578,
+      "step": 676
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6785736518455605e-05,
+      "loss": 0.0414,
+      "step": 677
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6776646285155326e-05,
+      "loss": 0.1031,
+      "step": 678
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.676754568499884e-05,
+      "loss": 0.07,
+      "step": 679
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6758434731908178e-05,
+      "loss": 0.0419,
+      "step": 680
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.674931343982119e-05,
+      "loss": 0.0477,
+      "step": 681
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6740181822691556e-05,
+      "loss": 0.0882,
+      "step": 682
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6731039894488747e-05,
+      "loss": 0.0667,
+      "step": 683
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6721887669198005e-05,
+      "loss": 0.0702,
+      "step": 684
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.671272516082033e-05,
+      "loss": 0.0509,
+      "step": 685
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6703552383372447e-05,
+      "loss": 0.0811,
+      "step": 686
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6694369350886796e-05,
+      "loss": 0.0265,
+      "step": 687
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.66851760774115e-05,
+      "loss": 0.0717,
+      "step": 688
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6675972577010354e-05,
+      "loss": 0.0501,
+      "step": 689
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6666758863762796e-05,
+      "loss": 0.0412,
+      "step": 690
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.665753495176388e-05,
+      "loss": 0.0316,
+      "step": 691
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6648300855124275e-05,
+      "loss": 0.0717,
+      "step": 692
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6639056587970222e-05,
+      "loss": 0.0195,
+      "step": 693
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6629802164443522e-05,
+      "loss": 0.0895,
+      "step": 694
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.662053759870151e-05,
+      "loss": 0.0543,
+      "step": 695
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.661126290491705e-05,
+      "loss": 0.0785,
+      "step": 696
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6601978097278483e-05,
+      "loss": 0.045,
+      "step": 697
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.659268318998963e-05,
+      "loss": 0.0745,
+      "step": 698
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6583378197269763e-05,
+      "loss": 0.0623,
+      "step": 699
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.657406313335358e-05,
+      "loss": 0.0868,
+      "step": 700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6564738012491193e-05,
+      "loss": 0.0598,
+      "step": 701
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6555402848948085e-05,
+      "loss": 0.0498,
+      "step": 702
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6546057657005113e-05,
+      "loss": 0.1038,
+      "step": 703
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6536702450958473e-05,
+      "loss": 0.1163,
+      "step": 704
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6527337245119678e-05,
+      "loss": 0.0325,
+      "step": 705
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6517962053815547e-05,
+      "loss": 0.0601,
+      "step": 706
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.650857689138816e-05,
+      "loss": 0.042,
+      "step": 707
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6499181772194867e-05,
+      "loss": 0.0543,
+      "step": 708
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6489776710608235e-05,
+      "loss": 0.0629,
+      "step": 709
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6480361721016053e-05,
+      "loss": 0.0561,
+      "step": 710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6470936817821292e-05,
+      "loss": 0.0692,
+      "step": 711
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6461502015442085e-05,
+      "loss": 0.025,
+      "step": 712
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6452057328311716e-05,
+      "loss": 0.0791,
+      "step": 713
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6442602770878586e-05,
+      "loss": 0.0636,
+      "step": 714
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6433138357606198e-05,
+      "loss": 0.0714,
+      "step": 715
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6423664102973133e-05,
+      "loss": 0.1096,
+      "step": 716
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6414180021473025e-05,
+      "loss": 0.048,
+      "step": 717
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.640468612761454e-05,
+      "loss": 0.0572,
+      "step": 718
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.639518243592136e-05,
+      "loss": 0.044,
+      "step": 719
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6385668960932143e-05,
+      "loss": 0.115,
+      "step": 720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6376145717200537e-05,
+      "loss": 0.0313,
+      "step": 721
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6366612719295106e-05,
+      "loss": 0.1093,
+      "step": 722
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.635706998179936e-05,
+      "loss": 0.0651,
+      "step": 723
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6347517519311694e-05,
+      "loss": 0.0444,
+      "step": 724
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.633795534644538e-05,
+      "loss": 0.0291,
+      "step": 725
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6328383477828556e-05,
+      "loss": 0.0395,
+      "step": 726
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6318801928104185e-05,
+      "loss": 0.0965,
+      "step": 727
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6309210711930037e-05,
+      "loss": 0.1418,
+      "step": 728
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6299609843978675e-05,
+      "loss": 0.0891,
+      "step": 729
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6289999338937427e-05,
+      "loss": 0.0214,
+      "step": 730
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6280379211508362e-05,
+      "loss": 0.1033,
+      "step": 731
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6270749476408264e-05,
+      "loss": 0.0945,
+      "step": 732
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6261110148368632e-05,
+      "loss": 0.142,
+      "step": 733
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.625146124213562e-05,
+      "loss": 0.0873,
+      "step": 734
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6241802772470043e-05,
+      "loss": 0.0767,
+      "step": 735
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6232134754147348e-05,
+      "loss": 0.0503,
+      "step": 736
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.622245720195759e-05,
+      "loss": 0.0515,
+      "step": 737
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6212770130705403e-05,
+      "loss": 0.0856,
+      "step": 738
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.620307355520999e-05,
+      "loss": 0.0538,
+      "step": 739
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.619336749030509e-05,
+      "loss": 0.0366,
+      "step": 740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.618365195083895e-05,
+      "loss": 0.0311,
+      "step": 741
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6173926951674336e-05,
+      "loss": 0.0417,
+      "step": 742
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.616419250768846e-05,
+      "loss": 0.0196,
+      "step": 743
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6154448633772993e-05,
+      "loss": 0.039,
+      "step": 744
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6144695344834026e-05,
+      "loss": 0.0663,
+      "step": 745
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.613493265579207e-05,
+      "loss": 0.0145,
+      "step": 746
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.612516058158199e-05,
+      "loss": 0.0725,
+      "step": 747
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.611537913715304e-05,
+      "loss": 0.0399,
+      "step": 748
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6105588337468773e-05,
+      "loss": 0.0625,
+      "step": 749
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.609578819750708e-05,
+      "loss": 0.0307,
+      "step": 750
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6085978732260133e-05,
+      "loss": 0.0098,
+      "step": 751
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6076159956734368e-05,
+      "loss": 0.05,
+      "step": 752
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6066331885950466e-05,
+      "loss": 0.0463,
+      "step": 753
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6056494534943324e-05,
+      "loss": 0.0551,
+      "step": 754
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.604664791876204e-05,
+      "loss": 0.077,
+      "step": 755
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6036792052469886e-05,
+      "loss": 0.0765,
+      "step": 756
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.6026926951144276e-05,
+      "loss": 0.0473,
+      "step": 757
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.6017052629876768e-05,
+      "loss": 0.0315,
+      "step": 758
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6007169103773005e-05,
+      "loss": 0.038,
+      "step": 759
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5997276387952733e-05,
+      "loss": 0.0568,
+      "step": 760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5987374497549736e-05,
+      "loss": 0.0227,
+      "step": 761
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5977463447711844e-05,
+      "loss": 0.0384,
+      "step": 762
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5967543253600902e-05,
+      "loss": 0.0547,
+      "step": 763
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5957613930392736e-05,
+      "loss": 0.0365,
+      "step": 764
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.594767549327714e-05,
+      "loss": 0.0486,
+      "step": 765
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5937727957457853e-05,
+      "loss": 0.0913,
+      "step": 766
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5927771338152528e-05,
+      "loss": 0.0397,
+      "step": 767
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.591780565059272e-05,
+      "loss": 0.0423,
+      "step": 768
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.590783091002385e-05,
+      "loss": 0.0308,
+      "step": 769
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5897847131705194e-05,
+      "loss": 0.026,
+      "step": 770
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.588785433090986e-05,
+      "loss": 0.0504,
+      "step": 771
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.0317,
+      "step": 772
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5867841723050505e-05,
+      "loss": 0.0727,
+      "step": 773
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5857821946601616e-05,
+      "loss": 0.068,
+      "step": 774
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5847793208906228e-05,
+      "loss": 0.0579,
+      "step": 775
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.583775552530622e-05,
+      "loss": 0.0859,
+      "step": 776
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.582770891115716e-05,
+      "loss": 0.0418,
+      "step": 777
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.581765338182827e-05,
+      "loss": 0.0146,
+      "step": 778
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5807588952702416e-05,
+      "loss": 0.0395,
+      "step": 779
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5797515639176077e-05,
+      "loss": 0.0248,
+      "step": 780
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.578743345665932e-05,
+      "loss": 0.0698,
+      "step": 781
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5777342420575782e-05,
+      "loss": 0.0359,
+      "step": 782
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5767242546362643e-05,
+      "loss": 0.066,
+      "step": 783
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5757133849470615e-05,
+      "loss": 0.0874,
+      "step": 784
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.5747016345363885e-05,
+      "loss": 0.0383,
+      "step": 785
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.5736890049520136e-05,
+      "loss": 0.0703,
+      "step": 786
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.5726754977430485e-05,
+      "loss": 0.0781,
+      "step": 787
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.5716611144599476e-05,
+      "loss": 0.0635,
+      "step": 788
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.5706458566545062e-05,
+      "loss": 0.0659,
+      "step": 789
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.5696297258798573e-05,
+      "loss": 0.0242,
+      "step": 790
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.568612723690469e-05,
+      "loss": 0.014,
+      "step": 791
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5675948516421424e-05,
+      "loss": 0.0504,
+      "step": 792
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.56657611129201e-05,
+      "loss": 0.0649,
+      "step": 793
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.565556504198532e-05,
+      "loss": 0.0839,
+      "step": 794
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5645360319214946e-05,
+      "loss": 0.0399,
+      "step": 795
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5635146960220074e-05,
+      "loss": 0.0663,
+      "step": 796
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.562492498062502e-05,
+      "loss": 0.0783,
+      "step": 797
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5614694396067285e-05,
+      "loss": 0.0205,
+      "step": 798
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.560445522219752e-05,
+      "loss": 0.0256,
+      "step": 799
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5594207474679533e-05,
+      "loss": 0.0908,
+      "step": 800
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.558395116919024e-05,
+      "loss": 0.0233,
+      "step": 801
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5573686321419653e-05,
+      "loss": 0.0315,
+      "step": 802
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5563412947070845e-05,
+      "loss": 0.0267,
+      "step": 803
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5553131061859944e-05,
+      "loss": 0.071,
+      "step": 804
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.554284068151608e-05,
+      "loss": 0.0402,
+      "step": 805
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.55325418217814e-05,
+      "loss": 0.0472,
+      "step": 806
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.552223449841101e-05,
+      "loss": 0.045,
+      "step": 807
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5511918727172966e-05,
+      "loss": 0.0385,
+      "step": 808
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5501594523848242e-05,
+      "loss": 0.0611,
+      "step": 809
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.549126190423073e-05,
+      "loss": 0.0793,
+      "step": 810
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5480920884127167e-05,
+      "loss": 0.0512,
+      "step": 811
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5470571479357173e-05,
+      "loss": 0.1304,
+      "step": 812
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5460213705753168e-05,
+      "loss": 0.0399,
+      "step": 813
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5449847579160398e-05,
+      "loss": 0.0198,
+      "step": 814
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5439473115436872e-05,
+      "loss": 0.0397,
+      "step": 815
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.542909033045335e-05,
+      "loss": 0.0367,
+      "step": 816
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5418699240093338e-05,
+      "loss": 0.0363,
+      "step": 817
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5408299860253034e-05,
+      "loss": 0.0589,
+      "step": 818
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5397892206841317e-05,
+      "loss": 0.0476,
+      "step": 819
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5387476295779737e-05,
+      "loss": 0.0473,
+      "step": 820
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.537705214300246e-05,
+      "loss": 0.0268,
+      "step": 821
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.536661976445627e-05,
+      "loss": 0.0194,
+      "step": 822
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5356179176100528e-05,
+      "loss": 0.045,
+      "step": 823
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.534573039390716e-05,
+      "loss": 0.0403,
+      "step": 824
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.533527343386062e-05,
+      "loss": 0.0415,
+      "step": 825
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.532480831195789e-05,
+      "loss": 0.0236,
+      "step": 826
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.531433504420841e-05,
+      "loss": 0.0551,
+      "step": 827
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5303853646634102e-05,
+      "loss": 0.0578,
+      "step": 828
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5293364135269325e-05,
+      "loss": 0.0419,
+      "step": 829
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5282866526160837e-05,
+      "loss": 0.076,
+      "step": 830
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5272360835367796e-05,
+      "loss": 0.0746,
+      "step": 831
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5261847078961718e-05,
+      "loss": 0.0589,
+      "step": 832
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5251325273026458e-05,
+      "loss": 0.0595,
+      "step": 833
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5240795433658187e-05,
+      "loss": 0.0584,
+      "step": 834
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5230257576965363e-05,
+      "loss": 0.0772,
+      "step": 835
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5219711719068708e-05,
+      "loss": 0.0868,
+      "step": 836
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5209157876101191e-05,
+      "loss": 0.0381,
+      "step": 837
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5198596064207989e-05,
+      "loss": 0.0821,
+      "step": 838
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5188026299546475e-05,
+      "loss": 0.0222,
+      "step": 839
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5177448598286182e-05,
+      "loss": 0.0444,
+      "step": 840
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.516686297660879e-05,
+      "loss": 0.0532,
+      "step": 841
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5156269450708094e-05,
+      "loss": 0.0489,
+      "step": 842
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.514566803678998e-05,
+      "loss": 0.0841,
+      "step": 843
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.51350587510724e-05,
+      "loss": 0.0292,
+      "step": 844
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5124441609785347e-05,
+      "loss": 0.0698,
+      "step": 845
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.5113816629170843e-05,
+      "loss": 0.0828,
+      "step": 846
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.5103183825482887e-05,
+      "loss": 0.0452,
+      "step": 847
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.509254321498745e-05,
+      "loss": 0.0482,
+      "step": 848
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.5081894813962454e-05,
+      "loss": 0.0486,
+      "step": 849
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.5071238638697731e-05,
+      "loss": 0.0614,
+      "step": 850
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.5060574705495007e-05,
+      "loss": 0.0688,
+      "step": 851
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.5049903030667876e-05,
+      "loss": 0.0499,
+      "step": 852
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5039223630541778e-05,
+      "loss": 0.0275,
+      "step": 853
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5028536521453969e-05,
+      "loss": 0.0339,
+      "step": 854
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5017841719753495e-05,
+      "loss": 0.0309,
+      "step": 855
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5007139241801176e-05,
+      "loss": 0.0676,
+      "step": 856
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.499642910396957e-05,
+      "loss": 0.0236,
+      "step": 857
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.4985711322642957e-05,
+      "loss": 0.0465,
+      "step": 858
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4974985914217309e-05,
+      "loss": 0.0278,
+      "step": 859
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4964252895100265e-05,
+      "loss": 0.0188,
+      "step": 860
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4953512281711105e-05,
+      "loss": 0.0603,
+      "step": 861
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.494276409048073e-05,
+      "loss": 0.0683,
+      "step": 862
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4932008337851633e-05,
+      "loss": 0.0685,
+      "step": 863
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4921245040277872e-05,
+      "loss": 0.0203,
+      "step": 864
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.491047421422505e-05,
+      "loss": 0.0489,
+      "step": 865
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4899695876170283e-05,
+      "loss": 0.0401,
+      "step": 866
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4888910042602182e-05,
+      "loss": 0.025,
+      "step": 867
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4878116730020829e-05,
+      "loss": 0.0228,
+      "step": 868
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4867315954937735e-05,
+      "loss": 0.0389,
+      "step": 869
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4856507733875837e-05,
+      "loss": 0.0466,
+      "step": 870
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4845692083369466e-05,
+      "loss": 0.0388,
+      "step": 871
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4834869019964303e-05,
+      "loss": 0.0425,
+      "step": 872
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4824038560217386e-05,
+      "loss": 0.0989,
+      "step": 873
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4813200720697056e-05,
+      "loss": 0.0681,
+      "step": 874
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4802355517982956e-05,
+      "loss": 0.009,
+      "step": 875
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.4791502968665977e-05,
+      "loss": 0.07,
+      "step": 876
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.4780643089348259e-05,
+      "loss": 0.0293,
+      "step": 877
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.4769775896643157e-05,
+      "loss": 0.0194,
+      "step": 878
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.4758901407175209e-05,
+      "loss": 0.0312,
+      "step": 879
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.4748019637580116e-05,
+      "loss": 0.0224,
+      "step": 880
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.4737130604504714e-05,
+      "loss": 0.0152,
+      "step": 881
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.472623432460696e-05,
+      "loss": 0.0713,
+      "step": 882
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.471533081455589e-05,
+      "loss": 0.045,
+      "step": 883
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.4704420091031599e-05,
+      "loss": 0.0535,
+      "step": 884
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.469350217072522e-05,
+      "loss": 0.0963,
+      "step": 885
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.4682577070338898e-05,
+      "loss": 0.0746,
+      "step": 886
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.4671644806585752e-05,
+      "loss": 0.0851,
+      "step": 887
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.4660705396189877e-05,
+      "loss": 0.0257,
+      "step": 888
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4649758855886283e-05,
+      "loss": 0.0407,
+      "step": 889
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4638805202420896e-05,
+      "loss": 0.0843,
+      "step": 890
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4627844452550523e-05,
+      "loss": 0.0446,
+      "step": 891
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4616876623042826e-05,
+      "loss": 0.0501,
+      "step": 892
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.46059017306763e-05,
+      "loss": 0.0469,
+      "step": 893
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4594919792240244e-05,
+      "loss": 0.0675,
+      "step": 894
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4583930824534729e-05,
+      "loss": 0.0478,
+      "step": 895
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.457293484437059e-05,
+      "loss": 0.0254,
+      "step": 896
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4561931868569378e-05,
+      "loss": 0.0506,
+      "step": 897
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4550921913963364e-05,
+      "loss": 0.0479,
+      "step": 898
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 0.045,
+      "step": 899
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.452888113571929e-05,
+      "loss": 0.0729,
+      "step": 900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4517850345799031e-05,
+      "loss": 0.0531,
+      "step": 901
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4506812644509505e-05,
+      "loss": 0.064,
+      "step": 902
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4495768048736092e-05,
+      "loss": 0.0295,
+      "step": 903
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4484716575374725e-05,
+      "loss": 0.0486,
+      "step": 904
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.447365824133185e-05,
+      "loss": 0.0371,
+      "step": 905
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.446259306352441e-05,
+      "loss": 0.0685,
+      "step": 906
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.445152105887983e-05,
+      "loss": 0.0344,
+      "step": 907
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4440442244335962e-05,
+      "loss": 0.0384,
+      "step": 908
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4429356636841085e-05,
+      "loss": 0.1131,
+      "step": 909
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4418264253353869e-05,
+      "loss": 0.07,
+      "step": 910
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4407165110843346e-05,
+      "loss": 0.0309,
+      "step": 911
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4396059226288893e-05,
+      "loss": 0.026,
+      "step": 912
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4384946616680194e-05,
+      "loss": 0.0184,
+      "step": 913
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4373827299017226e-05,
+      "loss": 0.0631,
+      "step": 914
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4362701290310234e-05,
+      "loss": 0.0921,
+      "step": 915
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4351568607579687e-05,
+      "loss": 0.0528,
+      "step": 916
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4340429267856263e-05,
+      "loss": 0.018,
+      "step": 917
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4329283288180841e-05,
+      "loss": 0.0635,
+      "step": 918
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.431813068560444e-05,
+      "loss": 0.0487,
+      "step": 919
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4306971477188223e-05,
+      "loss": 0.0309,
+      "step": 920
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4295805680003442e-05,
+      "loss": 0.0931,
+      "step": 921
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4284633311131448e-05,
+      "loss": 0.0448,
+      "step": 922
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4273454387663633e-05,
+      "loss": 0.0326,
+      "step": 923
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4262268926701422e-05,
+      "loss": 0.0782,
+      "step": 924
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4251076945356233e-05,
+      "loss": 0.0427,
+      "step": 925
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4239878460749474e-05,
+      "loss": 0.0444,
+      "step": 926
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4228673490012483e-05,
+      "loss": 0.0355,
+      "step": 927
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4217462050286532e-05,
+      "loss": 0.0451,
+      "step": 928
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.420624415872279e-05,
+      "loss": 0.026,
+      "step": 929
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.419501983248229e-05,
+      "loss": 0.0681,
+      "step": 930
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4183789088735915e-05,
+      "loss": 0.0376,
+      "step": 931
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4172551944664355e-05,
+      "loss": 0.0338,
+      "step": 932
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4161308417458102e-05,
+      "loss": 0.0646,
+      "step": 933
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.415005852431741e-05,
+      "loss": 0.0226,
+      "step": 934
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4138802282452269e-05,
+      "loss": 0.0439,
+      "step": 935
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4127539709082381e-05,
+      "loss": 0.0501,
+      "step": 936
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.411627082143714e-05,
+      "loss": 0.0492,
+      "step": 937
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.410499563675559e-05,
+      "loss": 0.0131,
+      "step": 938
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4093714172286415e-05,
+      "loss": 0.0813,
+      "step": 939
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4082426445287904e-05,
+      "loss": 0.0729,
+      "step": 940
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4071132473027929e-05,
+      "loss": 0.0629,
+      "step": 941
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4059832272783913e-05,
+      "loss": 0.0585,
+      "step": 942
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4048525861842801e-05,
+      "loss": 0.0426,
+      "step": 943
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.403721325750105e-05,
+      "loss": 0.0053,
+      "step": 944
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4025894477064586e-05,
+      "loss": 0.067,
+      "step": 945
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4014569537848784e-05,
+      "loss": 0.0343,
+      "step": 946
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4003238457178436e-05,
+      "loss": 0.0639,
+      "step": 947
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.3991901252387741e-05,
+      "loss": 0.0481,
+      "step": 948
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.3980557940820252e-05,
+      "loss": 0.0494,
+      "step": 949
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.3969208539828873e-05,
+      "loss": 0.0518,
+      "step": 950
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.3957853066775824e-05,
+      "loss": 0.0362,
+      "step": 951
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.3946491539032611e-05,
+      "loss": 0.0521,
+      "step": 952
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3935123973980002e-05,
+      "loss": 0.0196,
+      "step": 953
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3923750389008e-05,
+      "loss": 0.0502,
+      "step": 954
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.3912370801515821e-05,
+      "loss": 0.0799,
+      "step": 955
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.3900985228911865e-05,
+      "loss": 0.0537,
+      "step": 956
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.3889593688613682e-05,
+      "loss": 0.0495,
+      "step": 957
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3878196198047956e-05,
+      "loss": 0.0785,
+      "step": 958
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3866792774650466e-05,
+      "loss": 0.0822,
+      "step": 959
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3855383435866076e-05,
+      "loss": 0.0254,
+      "step": 960
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3843968199148697e-05,
+      "loss": 0.0706,
+      "step": 961
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.3832547081961261e-05,
+      "loss": 0.0494,
+      "step": 962
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.3821120101775698e-05,
+      "loss": 0.0599,
+      "step": 963
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.3809687276072903e-05,
+      "loss": 0.0353,
+      "step": 964
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.3798248622342719e-05,
+      "loss": 0.0561,
+      "step": 965
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.3786804158083898e-05,
+      "loss": 0.0317,
+      "step": 966
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.3775353900804088e-05,
+      "loss": 0.0495,
+      "step": 967
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3763897868019793e-05,
+      "loss": 0.0301,
+      "step": 968
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.375243607725636e-05,
+      "loss": 0.0744,
+      "step": 969
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3740968546047935e-05,
+      "loss": 0.0414,
+      "step": 970
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.372949529193745e-05,
+      "loss": 0.0448,
+      "step": 971
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.3718016332476593e-05,
+      "loss": 0.0676,
+      "step": 972
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3706531685225781e-05,
+      "loss": 0.0119,
+      "step": 973
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3695041367754134e-05,
+      "loss": 0.0404,
+      "step": 974
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3683545397639434e-05,
+      "loss": 0.0337,
+      "step": 975
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.367204379246812e-05,
+      "loss": 0.058,
+      "step": 976
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3660536569835257e-05,
+      "loss": 0.0491,
+      "step": 977
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3649023747344492e-05,
+      "loss": 0.0292,
+      "step": 978
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3637505342608044e-05,
+      "loss": 0.0272,
+      "step": 979
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.362598137324667e-05,
+      "loss": 0.06,
+      "step": 980
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3614451856889645e-05,
+      "loss": 0.0459,
+      "step": 981
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3602916811174724e-05,
+      "loss": 0.0399,
+      "step": 982
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3591376253748119e-05,
+      "loss": 0.0241,
+      "step": 983
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3579830202264484e-05,
+      "loss": 0.0419,
+      "step": 984
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3568278674386863e-05,
+      "loss": 0.0489,
+      "step": 985
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3556721687786692e-05,
+      "loss": 0.0303,
+      "step": 986
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3545159260143745e-05,
+      "loss": 0.0365,
+      "step": 987
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3533591409146133e-05,
+      "loss": 0.0597,
+      "step": 988
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3522018152490253e-05,
+      "loss": 0.0827,
+      "step": 989
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3510439507880778e-05,
+      "loss": 0.0746,
+      "step": 990
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3498855493030617e-05,
+      "loss": 0.0824,
+      "step": 991
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3487266125660898e-05,
+      "loss": 0.078,
+      "step": 992
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3475671423500942e-05,
+      "loss": 0.0202,
+      "step": 993
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3464071404288219e-05,
+      "loss": 0.1307,
+      "step": 994
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3452466085768348e-05,
+      "loss": 0.0182,
+      "step": 995
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3440855485695044e-05,
+      "loss": 0.0286,
+      "step": 996
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.34292396218301e-05,
+      "loss": 0.0377,
+      "step": 997
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3417618511943374e-05,
+      "loss": 0.0101,
+      "step": 998
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3405992173812738e-05,
+      "loss": 0.0199,
+      "step": 999
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3394360625224067e-05,
+      "loss": 0.0557,
+      "step": 1000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3382723883971203e-05,
+      "loss": 0.0083,
+      "step": 1001
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3371081967855939e-05,
+      "loss": 0.0697,
+      "step": 1002
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3359434894687976e-05,
+      "loss": 0.0505,
+      "step": 1003
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3347782682284909e-05,
+      "loss": 0.0702,
+      "step": 1004
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3336125348472193e-05,
+      "loss": 0.0359,
+      "step": 1005
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3324462911083123e-05,
+      "loss": 0.0602,
+      "step": 1006
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3312795387958795e-05,
+      "loss": 0.0996,
+      "step": 1007
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3301122796948083e-05,
+      "loss": 0.0391,
+      "step": 1008
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.328944515590762e-05,
+      "loss": 0.0414,
+      "step": 1009
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3277762482701769e-05,
+      "loss": 0.028,
+      "step": 1010
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.326607479520258e-05,
+      "loss": 0.0127,
+      "step": 1011
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3254382111289772e-05,
+      "loss": 0.0469,
+      "step": 1012
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3242684448850725e-05,
+      "loss": 0.0133,
+      "step": 1013
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.323098182578042e-05,
+      "loss": 0.021,
+      "step": 1014
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.321927425998143e-05,
+      "loss": 0.1061,
+      "step": 1015
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3207561769363889e-05,
+      "loss": 0.0579,
+      "step": 1016
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3195844371845467e-05,
+      "loss": 0.057,
+      "step": 1017
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3184122085351343e-05,
+      "loss": 0.0254,
+      "step": 1018
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3172394927814167e-05,
+      "loss": 0.0529,
+      "step": 1019
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3160662917174045e-05,
+      "loss": 0.1047,
+      "step": 1020
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3148926071378514e-05,
+      "loss": 0.0208,
+      "step": 1021
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.31371844083825e-05,
+      "loss": 0.0265,
+      "step": 1022
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3125437946148292e-05,
+      "loss": 0.0262,
+      "step": 1023
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3113686702645535e-05,
+      "loss": 0.0326,
+      "step": 1024
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3101930695851186e-05,
+      "loss": 0.0442,
+      "step": 1025
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.0455,
+      "step": 1026
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3078404464331909e-05,
+      "loss": 0.0503,
+      "step": 1027
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3066634275597214e-05,
+      "loss": 0.0455,
+      "step": 1028
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3054859395551329e-05,
+      "loss": 0.0601,
+      "step": 1029
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3043079842207363e-05,
+      "loss": 0.0735,
+      "step": 1030
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3031295633585577e-05,
+      "loss": 0.0468,
+      "step": 1031
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3019506787713359e-05,
+      "loss": 0.0642,
+      "step": 1032
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.300771332262518e-05,
+      "loss": 0.0309,
+      "step": 1033
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.2995915256362583e-05,
+      "loss": 0.0654,
+      "step": 1034
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.2984112606974155e-05,
+      "loss": 0.0098,
+      "step": 1035
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.2972305392515486e-05,
+      "loss": 0.0403,
+      "step": 1036
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.2960493631049151e-05,
+      "loss": 0.0831,
+      "step": 1037
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.2948677340644687e-05,
+      "loss": 0.0906,
+      "step": 1038
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.2936856539378552e-05,
+      "loss": 0.0342,
+      "step": 1039
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.2925031245334112e-05,
+      "loss": 0.0697,
+      "step": 1040
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.2913201476601595e-05,
+      "loss": 0.0564,
+      "step": 1041
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.2901367251278085e-05,
+      "loss": 0.0201,
+      "step": 1042
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.2889528587467486e-05,
+      "loss": 0.0521,
+      "step": 1043
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.2877685503280475e-05,
+      "loss": 0.0399,
+      "step": 1044
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.2865838016834506e-05,
+      "loss": 0.0206,
+      "step": 1045
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.2853986146253768e-05,
+      "loss": 0.0285,
+      "step": 1046
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2842129909669146e-05,
+      "loss": 0.0683,
+      "step": 1047
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2830269325218214e-05,
+      "loss": 0.0768,
+      "step": 1048
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2818404411045196e-05,
+      "loss": 0.0237,
+      "step": 1049
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2806535185300931e-05,
+      "loss": 0.0654,
+      "step": 1050
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2794661666142868e-05,
+      "loss": 0.0496,
+      "step": 1051
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2782783871735009e-05,
+      "loss": 0.0361,
+      "step": 1052
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2770901820247905e-05,
+      "loss": 0.0682,
+      "step": 1053
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2759015529858625e-05,
+      "loss": 0.0245,
+      "step": 1054
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2747125018750708e-05,
+      "loss": 0.0703,
+      "step": 1055
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2735230305114156e-05,
+      "loss": 0.0268,
+      "step": 1056
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2723331407145408e-05,
+      "loss": 0.012,
+      "step": 1057
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2711428343047293e-05,
+      "loss": 0.0406,
+      "step": 1058
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2699521131029023e-05,
+      "loss": 0.0211,
+      "step": 1059
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2687609789306144e-05,
+      "loss": 0.0204,
+      "step": 1060
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2675694336100529e-05,
+      "loss": 0.0269,
+      "step": 1061
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2663774789640335e-05,
+      "loss": 0.0504,
+      "step": 1062
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2651851168159985e-05,
+      "loss": 0.0413,
+      "step": 1063
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2639923489900138e-05,
+      "loss": 0.0371,
+      "step": 1064
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2627991773107651e-05,
+      "loss": 0.0715,
+      "step": 1065
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2616056036035564e-05,
+      "loss": 0.0277,
+      "step": 1066
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2604116296943065e-05,
+      "loss": 0.0249,
+      "step": 1067
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2592172574095472e-05,
+      "loss": 0.0562,
+      "step": 1068
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2580224885764185e-05,
+      "loss": 0.0269,
+      "step": 1069
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2568273250226681e-05,
+      "loss": 0.0522,
+      "step": 1070
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2556317685766468e-05,
+      "loss": 0.0325,
+      "step": 1071
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.254435821067307e-05,
+      "loss": 0.0222,
+      "step": 1072
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2532394843241988e-05,
+      "loss": 0.0203,
+      "step": 1073
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2520427601774683e-05,
+      "loss": 0.0521,
+      "step": 1074
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2508456504578538e-05,
+      "loss": 0.0251,
+      "step": 1075
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2496481569966837e-05,
+      "loss": 0.0335,
+      "step": 1076
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2484502816258735e-05,
+      "loss": 0.0192,
+      "step": 1077
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2472520261779221e-05,
+      "loss": 0.0425,
+      "step": 1078
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2460533924859118e-05,
+      "loss": 0.0649,
+      "step": 1079
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2448543823835016e-05,
+      "loss": 0.0431,
+      "step": 1080
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2436549977049267e-05,
+      "loss": 0.0322,
+      "step": 1081
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.242455240284996e-05,
+      "loss": 0.055,
+      "step": 1082
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2412551119590886e-05,
+      "loss": 0.0449,
+      "step": 1083
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.24005461456315e-05,
+      "loss": 0.0327,
+      "step": 1084
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2388537499336915e-05,
+      "loss": 0.0232,
+      "step": 1085
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2376525199077853e-05,
+      "loss": 0.0337,
+      "step": 1086
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2364509263230631e-05,
+      "loss": 0.0196,
+      "step": 1087
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.2352489710177126e-05,
+      "loss": 0.029,
+      "step": 1088
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.2340466558304749e-05,
+      "loss": 0.0471,
+      "step": 1089
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.2328439826006415e-05,
+      "loss": 0.0179,
+      "step": 1090
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.2316409531680522e-05,
+      "loss": 0.0137,
+      "step": 1091
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.2304375693730904e-05,
+      "loss": 0.0113,
+      "step": 1092
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.2292338330566828e-05,
+      "loss": 0.0409,
+      "step": 1093
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.2280297460602958e-05,
+      "loss": 0.0171,
+      "step": 1094
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.2268253102259302e-05,
+      "loss": 0.0404,
+      "step": 1095
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.2256205273961228e-05,
+      "loss": 0.064,
+      "step": 1096
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.2244153994139399e-05,
+      "loss": 0.0321,
+      "step": 1097
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.223209928122976e-05,
+      "loss": 0.0563,
+      "step": 1098
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.2220041153673507e-05,
+      "loss": 0.0515,
+      "step": 1099
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.2207979629917061e-05,
+      "loss": 0.056,
+      "step": 1100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.2195914728412046e-05,
+      "loss": 0.0653,
+      "step": 1101
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.2183846467615244e-05,
+      "loss": 0.0193,
+      "step": 1102
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.2171774865988572e-05,
+      "loss": 0.0417,
+      "step": 1103
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.2159699941999073e-05,
+      "loss": 0.0199,
+      "step": 1104
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.2147621714118856e-05,
+      "loss": 0.0961,
+      "step": 1105
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2135540200825095e-05,
+      "loss": 0.038,
+      "step": 1106
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.212345542059999e-05,
+      "loss": 0.0556,
+      "step": 1107
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.211136739193073e-05,
+      "loss": 0.0363,
+      "step": 1108
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2099276133309483e-05,
+      "loss": 0.0079,
+      "step": 1109
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2087181663233354e-05,
+      "loss": 0.0383,
+      "step": 1110
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2075084000204355e-05,
+      "loss": 0.0187,
+      "step": 1111
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.20629831627294e-05,
+      "loss": 0.0463,
+      "step": 1112
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.205087916932024e-05,
+      "loss": 0.0251,
+      "step": 1113
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.203877203849346e-05,
+      "loss": 0.0109,
+      "step": 1114
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2026661788770453e-05,
+      "loss": 0.0168,
+      "step": 1115
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2014548438677374e-05,
+      "loss": 0.023,
+      "step": 1116
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2002432006745121e-05,
+      "loss": 0.083,
+      "step": 1117
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1990312511509313e-05,
+      "loss": 0.0505,
+      "step": 1118
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1978189971510248e-05,
+      "loss": 0.0167,
+      "step": 1119
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1966064405292887e-05,
+      "loss": 0.0429,
+      "step": 1120
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1953935831406817e-05,
+      "loss": 0.0464,
+      "step": 1121
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1941804268406232e-05,
+      "loss": 0.0377,
+      "step": 1122
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1929669734849892e-05,
+      "loss": 0.0375,
+      "step": 1123
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1917532249301102e-05,
+      "loss": 0.0435,
+      "step": 1124
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1905391830327685e-05,
+      "loss": 0.0566,
+      "step": 1125
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1893248496501955e-05,
+      "loss": 0.0273,
+      "step": 1126
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1881102266400679e-05,
+      "loss": 0.0175,
+      "step": 1127
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1868953158605058e-05,
+      "loss": 0.0183,
+      "step": 1128
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1856801191700694e-05,
+      "loss": 0.0569,
+      "step": 1129
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.184464638427756e-05,
+      "loss": 0.0197,
+      "step": 1130
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1832488754929982e-05,
+      "loss": 0.0441,
+      "step": 1131
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.18203283222566e-05,
+      "loss": 0.0299,
+      "step": 1132
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1808165104860332e-05,
+      "loss": 0.0255,
+      "step": 1133
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1795999121348379e-05,
+      "loss": 0.0263,
+      "step": 1134
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.178383039033215e-05,
+      "loss": 0.0297,
+      "step": 1135
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1771658930427266e-05,
+      "loss": 0.0456,
+      "step": 1136
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1759484760253528e-05,
+      "loss": 0.0659,
+      "step": 1137
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.1747307898434877e-05,
+      "loss": 0.0167,
+      "step": 1138
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1735128363599376e-05,
+      "loss": 0.0244,
+      "step": 1139
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1722946174379168e-05,
+      "loss": 0.0263,
+      "step": 1140
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.1710761349410469e-05,
+      "loss": 0.0782,
+      "step": 1141
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1698573907333519e-05,
+      "loss": 0.0387,
+      "step": 1142
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.1686383866792563e-05,
+      "loss": 0.02,
+      "step": 1143
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1674191246435826e-05,
+      "loss": 0.036,
+      "step": 1144
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.166199606491547e-05,
+      "loss": 0.073,
+      "step": 1145
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.1649798340887584e-05,
+      "loss": 0.0138,
+      "step": 1146
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.163759809301214e-05,
+      "loss": 0.1064,
+      "step": 1147
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.162539533995298e-05,
+      "loss": 0.0312,
+      "step": 1148
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1613190100377767e-05,
+      "loss": 0.0257,
+      "step": 1149
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1600982392957978e-05,
+      "loss": 0.0209,
+      "step": 1150
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.1588772236368852e-05,
+      "loss": 0.0223,
+      "step": 1151
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1576559649289394e-05,
+      "loss": 0.0226,
+      "step": 1152
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 0.0398,
+      "step": 1153
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1552127258394004e-05,
+      "loss": 0.0496,
+      "step": 1154
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1539907491954539e-05,
+      "loss": 0.0176,
+      "step": 1155
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.152768536977761e-05,
+      "loss": 0.053,
+      "step": 1156
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1515460910560514e-05,
+      "loss": 0.0197,
+      "step": 1157
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1503234133004129e-05,
+      "loss": 0.0088,
+      "step": 1158
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.1491005055812874e-05,
+      "loss": 0.0314,
+      "step": 1159
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.1478773697694691e-05,
+      "loss": 0.0986,
+      "step": 1160
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.1466540077361005e-05,
+      "loss": 0.0303,
+      "step": 1161
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.1454304213526703e-05,
+      "loss": 0.0545,
+      "step": 1162
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.1442066124910111e-05,
+      "loss": 0.0649,
+      "step": 1163
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.142982583023295e-05,
+      "loss": 0.0486,
+      "step": 1164
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.1417583348220322e-05,
+      "loss": 0.0436,
+      "step": 1165
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.1405338697600672e-05,
+      "loss": 0.0835,
+      "step": 1166
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.1393091897105762e-05,
+      "loss": 0.0309,
+      "step": 1167
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.1380842965470641e-05,
+      "loss": 0.0057,
+      "step": 1168
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.1368591921433625e-05,
+      "loss": 0.0519,
+      "step": 1169
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.1356338783736256e-05,
+      "loss": 0.0422,
+      "step": 1170
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.1344083571123285e-05,
+      "loss": 0.0443,
+      "step": 1171
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.1331826302342629e-05,
+      "loss": 0.025,
+      "step": 1172
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.1319566996145348e-05,
+      "loss": 0.0175,
+      "step": 1173
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.1307305671285633e-05,
+      "loss": 0.021,
+      "step": 1174
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1295042346520755e-05,
+      "loss": 0.0341,
+      "step": 1175
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1282777040611044e-05,
+      "loss": 0.0379,
+      "step": 1176
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.1270509772319857e-05,
+      "loss": 0.0264,
+      "step": 1177
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.125824056041356e-05,
+      "loss": 0.0314,
+      "step": 1178
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.1245969423661491e-05,
+      "loss": 0.0667,
+      "step": 1179
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.123369638083593e-05,
+      "loss": 0.0413,
+      "step": 1180
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.1221421450712076e-05,
+      "loss": 0.0597,
+      "step": 1181
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.120914465206801e-05,
+      "loss": 0.0187,
+      "step": 1182
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.119686600368468e-05,
+      "loss": 0.0466,
+      "step": 1183
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.1184585524345848e-05,
+      "loss": 0.0236,
+      "step": 1184
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.11723032328381e-05,
+      "loss": 0.0303,
+      "step": 1185
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.1160019147950774e-05,
+      "loss": 0.0103,
+      "step": 1186
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.1147733288475961e-05,
+      "loss": 0.0217,
+      "step": 1187
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.1135445673208468e-05,
+      "loss": 0.0558,
+      "step": 1188
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.112315632094578e-05,
+      "loss": 0.0196,
+      "step": 1189
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1110865250488047e-05,
+      "loss": 0.0357,
+      "step": 1190
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1098572480638044e-05,
+      "loss": 0.0212,
+      "step": 1191
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.108627803020115e-05,
+      "loss": 0.0456,
+      "step": 1192
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.1073981917985303e-05,
+      "loss": 0.0196,
+      "step": 1193
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.1061684162801e-05,
+      "loss": 0.0279,
+      "step": 1194
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.1049384783461237e-05,
+      "loss": 0.0195,
+      "step": 1195
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.1037083798781502e-05,
+      "loss": 0.0369,
+      "step": 1196
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.1024781227579741e-05,
+      "loss": 0.0516,
+      "step": 1197
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1012477088676322e-05,
+      "loss": 0.0623,
+      "step": 1198
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1000171400894008e-05,
+      "loss": 0.0428,
+      "step": 1199
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0987864183057943e-05,
+      "loss": 0.0125,
+      "step": 1200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0975555453995601e-05,
+      "loss": 0.0675,
+      "step": 1201
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0963245232536768e-05,
+      "loss": 0.0284,
+      "step": 1202
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0950933537513526e-05,
+      "loss": 0.0646,
+      "step": 1203
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0938620387760194e-05,
+      "loss": 0.037,
+      "step": 1204
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.092630580211333e-05,
+      "loss": 0.036,
+      "step": 1205
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0913989799411675e-05,
+      "loss": 0.0277,
+      "step": 1206
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0901672398496155e-05,
+      "loss": 0.0253,
+      "step": 1207
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0889353618209823e-05,
+      "loss": 0.0282,
+      "step": 1208
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.087703347739784e-05,
+      "loss": 0.0263,
+      "step": 1209
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0864711994907457e-05,
+      "loss": 0.0289,
+      "step": 1210
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0852389189587977e-05,
+      "loss": 0.044,
+      "step": 1211
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.084006508029072e-05,
+      "loss": 0.0664,
+      "step": 1212
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0827739685869001e-05,
+      "loss": 0.035,
+      "step": 1213
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0815413025178112e-05,
+      "loss": 0.0641,
+      "step": 1214
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.080308511707527e-05,
+      "loss": 0.0381,
+      "step": 1215
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0790755980419603e-05,
+      "loss": 0.0617,
+      "step": 1216
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0778425634072119e-05,
+      "loss": 0.0242,
+      "step": 1217
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.076609409689568e-05,
+      "loss": 0.0564,
+      "step": 1218
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0753761387754975e-05,
+      "loss": 0.039,
+      "step": 1219
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0741427525516463e-05,
+      "loss": 0.0164,
+      "step": 1220
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0729092529048393e-05,
+      "loss": 0.0168,
+      "step": 1221
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0716756417220735e-05,
+      "loss": 0.0453,
+      "step": 1222
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.070441920890517e-05,
+      "loss": 0.0262,
+      "step": 1223
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0692080922975052e-05,
+      "loss": 0.0394,
+      "step": 1224
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.067974157830539e-05,
+      "loss": 0.0127,
+      "step": 1225
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0667401193772807e-05,
+      "loss": 0.0188,
+      "step": 1226
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.065505978825552e-05,
+      "loss": 0.0594,
+      "step": 1227
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0642717380633303e-05,
+      "loss": 0.0578,
+      "step": 1228
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0630373989787472e-05,
+      "loss": 0.0435,
+      "step": 1229
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0618029634600843e-05,
+      "loss": 0.0162,
+      "step": 1230
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.06056843339577e-05,
+      "loss": 0.041,
+      "step": 1231
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.0593338106743785e-05,
+      "loss": 0.0508,
+      "step": 1232
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0580990971846247e-05,
+      "loss": 0.0225,
+      "step": 1233
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0568642948153635e-05,
+      "loss": 0.022,
+      "step": 1234
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.0556294054555847e-05,
+      "loss": 0.0388,
+      "step": 1235
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0543944309944117e-05,
+      "loss": 0.0302,
+      "step": 1236
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0531593733210979e-05,
+      "loss": 0.0158,
+      "step": 1237
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0519242343250241e-05,
+      "loss": 0.0345,
+      "step": 1238
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0506890158956954e-05,
+      "loss": 0.0277,
+      "step": 1239
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0494537199227393e-05,
+      "loss": 0.0122,
+      "step": 1240
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0482183482959e-05,
+      "loss": 0.0153,
+      "step": 1241
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0469829029050391e-05,
+      "loss": 0.0171,
+      "step": 1242
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0457473856401305e-05,
+      "loss": 0.0194,
+      "step": 1243
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.0445117983912583e-05,
+      "loss": 0.0692,
+      "step": 1244
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.043276143048613e-05,
+      "loss": 0.0397,
+      "step": 1245
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.04204042150249e-05,
+      "loss": 0.0478,
+      "step": 1246
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0408046356432855e-05,
+      "loss": 0.0128,
+      "step": 1247
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0395687873614944e-05,
+      "loss": 0.0162,
+      "step": 1248
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0383328785477067e-05,
+      "loss": 0.0326,
+      "step": 1249
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0370969110926052e-05,
+      "loss": 0.0402,
+      "step": 1250
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0358608868869632e-05,
+      "loss": 0.0685,
+      "step": 1251
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0346248078216397e-05,
+      "loss": 0.0433,
+      "step": 1252
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0333886757875773e-05,
+      "loss": 0.0405,
+      "step": 1253
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0321524926758013e-05,
+      "loss": 0.0456,
+      "step": 1254
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0309162603774137e-05,
+      "loss": 0.07,
+      "step": 1255
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0296799807835925e-05,
+      "loss": 0.0336,
+      "step": 1256
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0284436557855873e-05,
+      "loss": 0.041,
+      "step": 1257
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0272072872747181e-05,
+      "loss": 0.0151,
+      "step": 1258
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0259708771423708e-05,
+      "loss": 0.035,
+      "step": 1259
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.024734427279995e-05,
+      "loss": 0.0156,
+      "step": 1260
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0234979395791015e-05,
+      "loss": 0.0648,
+      "step": 1261
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0222614159312589e-05,
+      "loss": 0.104,
+      "step": 1262
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.0210248582280904e-05,
+      "loss": 0.0379,
+      "step": 1263
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0197882683612712e-05,
+      "loss": 0.0294,
+      "step": 1264
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0185516482225264e-05,
+      "loss": 0.0314,
+      "step": 1265
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0173149997036272e-05,
+      "loss": 0.0557,
+      "step": 1266
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0160783246963877e-05,
+      "loss": 0.016,
+      "step": 1267
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0148416250926629e-05,
+      "loss": 0.0316,
+      "step": 1268
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0136049027843457e-05,
+      "loss": 0.0045,
+      "step": 1269
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.012368159663363e-05,
+      "loss": 0.0494,
+      "step": 1270
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0111313976216744e-05,
+      "loss": 0.0173,
+      "step": 1271
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0098946185512677e-05,
+      "loss": 0.005,
+      "step": 1272
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0086578243441571e-05,
+      "loss": 0.0325,
+      "step": 1273
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0074210168923804e-05,
+      "loss": 0.0269,
+      "step": 1274
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0061841980879941e-05,
+      "loss": 0.0251,
+      "step": 1275
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0049473698230743e-05,
+      "loss": 0.0226,
+      "step": 1276
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.00371053398971e-05,
+      "loss": 0.037,
+      "step": 1277
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.002473692480002e-05,
+      "loss": 0.0298,
+      "step": 1278
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.0012368471860603e-05,
+      "loss": 0.0278,
+      "step": 1279
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1e-05,
+      "loss": 0.033,
+      "step": 1280
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.987631528139402e-06,
+      "loss": 0.0561,
+      "step": 1281
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.975263075199983e-06,
+      "loss": 0.0165,
+      "step": 1282
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.962894660102902e-06,
+      "loss": 0.0076,
+      "step": 1283
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.950526301769257e-06,
+      "loss": 0.0358,
+      "step": 1284
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.93815801912006e-06,
+      "loss": 0.0462,
+      "step": 1285
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.925789831076202e-06,
+      "loss": 0.0397,
+      "step": 1286
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.91342175655843e-06,
+      "loss": 0.0457,
+      "step": 1287
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.901053814487328e-06,
+      "loss": 0.0419,
+      "step": 1288
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.888686023783258e-06,
+      "loss": 0.0131,
+      "step": 1289
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.876318403366371e-06,
+      "loss": 0.0192,
+      "step": 1290
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.863950972156548e-06,
+      "loss": 0.015,
+      "step": 1291
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.851583749073373e-06,
+      "loss": 0.0091,
+      "step": 1292
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.839216753036127e-06,
+      "loss": 0.0328,
+      "step": 1293
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.82685000296373e-06,
+      "loss": 0.0311,
+      "step": 1294
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.814483517774738e-06,
+      "loss": 0.0592,
+      "step": 1295
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.802117316387289e-06,
+      "loss": 0.0364,
+      "step": 1296
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.789751417719099e-06,
+      "loss": 0.0314,
+      "step": 1297
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.777385840687415e-06,
+      "loss": 0.029,
+      "step": 1298
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.765020604208986e-06,
+      "loss": 0.0505,
+      "step": 1299
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.752655727200051e-06,
+      "loss": 0.009,
+      "step": 1300
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.740291228576294e-06,
+      "loss": 0.0215,
+      "step": 1301
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.727927127252822e-06,
+      "loss": 0.0166,
+      "step": 1302
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.715563442144132e-06,
+      "loss": 0.044,
+      "step": 1303
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.70320019216408e-06,
+      "loss": 0.0197,
+      "step": 1304
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.690837396225867e-06,
+      "loss": 0.0227,
+      "step": 1305
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.678475073241987e-06,
+      "loss": 0.0385,
+      "step": 1306
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.66611324212423e-06,
+      "loss": 0.0415,
+      "step": 1307
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.653751921783608e-06,
+      "loss": 0.0217,
+      "step": 1308
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.64139113113037e-06,
+      "loss": 0.0394,
+      "step": 1309
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.62903088907395e-06,
+      "loss": 0.0386,
+      "step": 1310
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.616671214522935e-06,
+      "loss": 0.0201,
+      "step": 1311
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.604312126385061e-06,
+      "loss": 0.0336,
+      "step": 1312
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.591953643567152e-06,
+      "loss": 0.0519,
+      "step": 1313
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.579595784975102e-06,
+      "loss": 0.0425,
+      "step": 1314
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.567238569513872e-06,
+      "loss": 0.0136,
+      "step": 1315
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.554882016087419e-06,
+      "loss": 0.0124,
+      "step": 1316
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.542526143598697e-06,
+      "loss": 0.0327,
+      "step": 1317
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.530170970949612e-06,
+      "loss": 0.0075,
+      "step": 1318
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.517816517041003e-06,
+      "loss": 0.0208,
+      "step": 1319
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.505462800772612e-06,
+      "loss": 0.0347,
+      "step": 1320
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.493109841043046e-06,
+      "loss": 0.0292,
+      "step": 1321
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.48075765674976e-06,
+      "loss": 0.0671,
+      "step": 1322
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.468406266789026e-06,
+      "loss": 0.0202,
+      "step": 1323
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.456055690055886e-06,
+      "loss": 0.0403,
+      "step": 1324
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.443705945444158e-06,
+      "loss": 0.0124,
+      "step": 1325
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.431357051846367e-06,
+      "loss": 0.0298,
+      "step": 1326
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.419009028153754e-06,
+      "loss": 0.0238,
+      "step": 1327
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.406661893256215e-06,
+      "loss": 0.0276,
+      "step": 1328
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.394315666042303e-06,
+      "loss": 0.0093,
+      "step": 1329
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.381970365399162e-06,
+      "loss": 0.0961,
+      "step": 1330
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.369626010212528e-06,
+      "loss": 0.0387,
+      "step": 1331
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.3572826193667e-06,
+      "loss": 0.0096,
+      "step": 1332
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.344940211744483e-06,
+      "loss": 0.0142,
+      "step": 1333
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.332598806227195e-06,
+      "loss": 0.0282,
+      "step": 1334
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.320258421694615e-06,
+      "loss": 0.0073,
+      "step": 1335
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.307919077024951e-06,
+      "loss": 0.0266,
+      "step": 1336
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.295580791094834e-06,
+      "loss": 0.0285,
+      "step": 1337
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.283243582779266e-06,
+      "loss": 0.0758,
+      "step": 1338
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.27090747095161e-06,
+      "loss": 0.0351,
+      "step": 1339
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.25857247448354e-06,
+      "loss": 0.0435,
+      "step": 1340
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.246238612245028e-06,
+      "loss": 0.0511,
+      "step": 1341
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.233905903104321e-06,
+      "loss": 0.0276,
+      "step": 1342
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.221574365927881e-06,
+      "loss": 0.0166,
+      "step": 1343
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.209244019580402e-06,
+      "loss": 0.0077,
+      "step": 1344
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.196914882924737e-06,
+      "loss": 0.0361,
+      "step": 1345
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.184586974821891e-06,
+      "loss": 0.0553,
+      "step": 1346
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.172260314131002e-06,
+      "loss": 0.032,
+      "step": 1347
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.159934919709284e-06,
+      "loss": 0.0396,
+      "step": 1348
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.147610810412026e-06,
+      "loss": 0.0778,
+      "step": 1349
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.135288005092546e-06,
+      "loss": 0.0224,
+      "step": 1350
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.122966522602163e-06,
+      "loss": 0.0627,
+      "step": 1351
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.110646381790182e-06,
+      "loss": 0.0313,
+      "step": 1352
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.098327601503847e-06,
+      "loss": 0.0866,
+      "step": 1353
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.086010200588328e-06,
+      "loss": 0.0226,
+      "step": 1354
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.073694197886676e-06,
+      "loss": 0.013,
+      "step": 1355
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.061379612239807e-06,
+      "loss": 0.0184,
+      "step": 1356
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.049066462486479e-06,
+      "loss": 0.033,
+      "step": 1357
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.036754767463233e-06,
+      "loss": 0.0512,
+      "step": 1358
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.024444546004402e-06,
+      "loss": 0.0256,
+      "step": 1359
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.012135816942058e-06,
+      "loss": 0.0118,
+      "step": 1360
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 8.999828599105993e-06,
+      "loss": 0.0363,
+      "step": 1361
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 8.987522911323681e-06,
+      "loss": 0.0312,
+      "step": 1362
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 8.975218772420259e-06,
+      "loss": 0.0314,
+      "step": 1363
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 8.9629162012185e-06,
+      "loss": 0.0258,
+      "step": 1364
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 8.950615216538765e-06,
+      "loss": 0.0297,
+      "step": 1365
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 8.938315837199003e-06,
+      "loss": 0.0143,
+      "step": 1366
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 8.926018082014702e-06,
+      "loss": 0.0101,
+      "step": 1367
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 8.913721969798855e-06,
+      "loss": 0.0181,
+      "step": 1368
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 8.90142751936196e-06,
+      "loss": 0.0205,
+      "step": 1369
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 8.889134749511956e-06,
+      "loss": 0.028,
+      "step": 1370
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 8.876843679054222e-06,
+      "loss": 0.035,
+      "step": 1371
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 8.864554326791537e-06,
+      "loss": 0.0105,
+      "step": 1372
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 8.85226671152404e-06,
+      "loss": 0.0264,
+      "step": 1373
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 8.83998085204923e-06,
+      "loss": 0.0162,
+      "step": 1374
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 8.827696767161902e-06,
+      "loss": 0.0532,
+      "step": 1375
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 8.815414475654156e-06,
+      "loss": 0.0656,
+      "step": 1376
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 8.803133996315326e-06,
+      "loss": 0.0185,
+      "step": 1377
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 8.790855347931991e-06,
+      "loss": 0.0406,
+      "step": 1378
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 8.778578549287928e-06,
+      "loss": 0.0526,
+      "step": 1379
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 8.76630361916407e-06,
+      "loss": 0.0364,
+      "step": 1380
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 8.75403057633851e-06,
+      "loss": 0.0483,
+      "step": 1381
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 8.741759439586443e-06,
+      "loss": 0.0123,
+      "step": 1382
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 8.729490227680147e-06,
+      "loss": 0.0192,
+      "step": 1383
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 8.71722295938896e-06,
+      "loss": 0.0283,
+      "step": 1384
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 8.704957653479245e-06,
+      "loss": 0.0136,
+      "step": 1385
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 8.692694328714369e-06,
+      "loss": 0.024,
+      "step": 1386
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 8.680433003854657e-06,
+      "loss": 0.022,
+      "step": 1387
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 8.668173697657376e-06,
+      "loss": 0.0599,
+      "step": 1388
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 8.65591642887672e-06,
+      "loss": 0.0159,
+      "step": 1389
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 8.643661216263744e-06,
+      "loss": 0.0368,
+      "step": 1390
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 8.631408078566379e-06,
+      "loss": 0.0392,
+      "step": 1391
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 8.619157034529362e-06,
+      "loss": 0.0247,
+      "step": 1392
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 8.606908102894243e-06,
+      "loss": 0.0319,
+      "step": 1393
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.594661302399333e-06,
+      "loss": 0.0719,
+      "step": 1394
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.58241665177968e-06,
+      "loss": 0.0214,
+      "step": 1395
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.570174169767051e-06,
+      "loss": 0.0389,
+      "step": 1396
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 8.55793387508989e-06,
+      "loss": 0.0167,
+      "step": 1397
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 8.5456957864733e-06,
+      "loss": 0.0185,
+      "step": 1398
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.533459922639e-06,
+      "loss": 0.0308,
+      "step": 1399
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.52122630230531e-06,
+      "loss": 0.0159,
+      "step": 1400
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.508994944187128e-06,
+      "loss": 0.0307,
+      "step": 1401
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.496765866995871e-06,
+      "loss": 0.0217,
+      "step": 1402
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.484539089439488e-06,
+      "loss": 0.0276,
+      "step": 1403
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.472314630222396e-06,
+      "loss": 0.0144,
+      "step": 1404
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.460092508045465e-06,
+      "loss": 0.0246,
+      "step": 1405
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.447872741606e-06,
+      "loss": 0.0286,
+      "step": 1406
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 0.0155,
+      "step": 1407
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.42344035071061e-06,
+      "loss": 0.0375,
+      "step": 1408
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.41122776363115e-06,
+      "loss": 0.0324,
+      "step": 1409
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.399017607042025e-06,
+      "loss": 0.0943,
+      "step": 1410
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.386809899622235e-06,
+      "loss": 0.0697,
+      "step": 1411
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.374604660047021e-06,
+      "loss": 0.0998,
+      "step": 1412
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.362401906987863e-06,
+      "loss": 0.0367,
+      "step": 1413
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.35020165911242e-06,
+      "loss": 0.0528,
+      "step": 1414
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.338003935084531e-06,
+      "loss": 0.0643,
+      "step": 1415
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.325808753564179e-06,
+      "loss": 0.0317,
+      "step": 1416
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.313616133207439e-06,
+      "loss": 0.0344,
+      "step": 1417
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.301426092666483e-06,
+      "loss": 0.0074,
+      "step": 1418
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.289238650589536e-06,
+      "loss": 0.0513,
+      "step": 1419
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.277053825620836e-06,
+      "loss": 0.0227,
+      "step": 1420
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.264871636400629e-06,
+      "loss": 0.017,
+      "step": 1421
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.252692101565123e-06,
+      "loss": 0.0551,
+      "step": 1422
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.240515239746477e-06,
+      "loss": 0.0154,
+      "step": 1423
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.228341069572736e-06,
+      "loss": 0.0332,
+      "step": 1424
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.216169609667854e-06,
+      "loss": 0.0386,
+      "step": 1425
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.204000878651626e-06,
+      "loss": 0.0539,
+      "step": 1426
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.191834895139668e-06,
+      "loss": 0.0108,
+      "step": 1427
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.179671677743404e-06,
+      "loss": 0.0216,
+      "step": 1428
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.167511245070018e-06,
+      "loss": 0.0305,
+      "step": 1429
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.155353615722442e-06,
+      "loss": 0.0418,
+      "step": 1430
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.143198808299311e-06,
+      "loss": 0.0087,
+      "step": 1431
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.131046841394943e-06,
+      "loss": 0.017,
+      "step": 1432
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.118897733599323e-06,
+      "loss": 0.0326,
+      "step": 1433
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.106751503498045e-06,
+      "loss": 0.0349,
+      "step": 1434
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.094608169672318e-06,
+      "loss": 0.0476,
+      "step": 1435
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.082467750698903e-06,
+      "loss": 0.0541,
+      "step": 1436
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.07033026515011e-06,
+      "loss": 0.0175,
+      "step": 1437
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.058195731593773e-06,
+      "loss": 0.039,
+      "step": 1438
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.046064168593184e-06,
+      "loss": 0.0371,
+      "step": 1439
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.033935594707116e-06,
+      "loss": 0.0164,
+      "step": 1440
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.021810028489757e-06,
+      "loss": 0.0442,
+      "step": 1441
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.009687488490692e-06,
+      "loss": 0.028,
+      "step": 1442
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.997567993254882e-06,
+      "loss": 0.0335,
+      "step": 1443
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.985451561322628e-06,
+      "loss": 0.0354,
+      "step": 1444
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.97333821122955e-06,
+      "loss": 0.0167,
+      "step": 1445
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.961227961506543e-06,
+      "loss": 0.0404,
+      "step": 1446
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.949120830679763e-06,
+      "loss": 0.0333,
+      "step": 1447
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.937016837270605e-06,
+      "loss": 0.0611,
+      "step": 1448
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 7.924915999795645e-06,
+      "loss": 0.0186,
+      "step": 1449
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.91281833676665e-06,
+      "loss": 0.0148,
+      "step": 1450
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.900723866690522e-06,
+      "loss": 0.0392,
+      "step": 1451
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.888632608069273e-06,
+      "loss": 0.035,
+      "step": 1452
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.876544579400015e-06,
+      "loss": 0.0119,
+      "step": 1453
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.864459799174905e-06,
+      "loss": 0.0236,
+      "step": 1454
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.852378285881148e-06,
+      "loss": 0.0423,
+      "step": 1455
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.840300058000928e-06,
+      "loss": 0.0187,
+      "step": 1456
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 7.82822513401143e-06,
+      "loss": 0.0229,
+      "step": 1457
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.81615353238476e-06,
+      "loss": 0.0162,
+      "step": 1458
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.804085271587953e-06,
+      "loss": 0.0775,
+      "step": 1459
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.79202037008294e-06,
+      "loss": 0.0287,
+      "step": 1460
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.779958846326497e-06,
+      "loss": 0.0294,
+      "step": 1461
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.767900718770245e-06,
+      "loss": 0.0099,
+      "step": 1462
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.755846005860606e-06,
+      "loss": 0.031,
+      "step": 1463
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 7.743794726038774e-06,
+      "loss": 0.0305,
+      "step": 1464
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.7317468977407e-06,
+      "loss": 0.0236,
+      "step": 1465
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.719702539397044e-06,
+      "loss": 0.0377,
+      "step": 1466
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.707661669433173e-06,
+      "loss": 0.0344,
+      "step": 1467
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.695624306269101e-06,
+      "loss": 0.0319,
+      "step": 1468
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.683590468319482e-06,
+      "loss": 0.0149,
+      "step": 1469
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 7.671560173993588e-06,
+      "loss": 0.0281,
+      "step": 1470
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.659533441695253e-06,
+      "loss": 0.0206,
+      "step": 1471
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.647510289822875e-06,
+      "loss": 0.0163,
+      "step": 1472
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.635490736769374e-06,
+      "loss": 0.0411,
+      "step": 1473
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.62347480092215e-06,
+      "loss": 0.0498,
+      "step": 1474
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.6114625006630885e-06,
+      "loss": 0.0175,
+      "step": 1475
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.5994538543685e-06,
+      "loss": 0.0228,
+      "step": 1476
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.5874488804091184e-06,
+      "loss": 0.0287,
+      "step": 1477
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.575447597150045e-06,
+      "loss": 0.0243,
+      "step": 1478
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.563450022950737e-06,
+      "loss": 0.021,
+      "step": 1479
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.551456176164989e-06,
+      "loss": 0.0091,
+      "step": 1480
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.539466075140885e-06,
+      "loss": 0.0305,
+      "step": 1481
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.527479738220781e-06,
+      "loss": 0.0231,
+      "step": 1482
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.5154971837412715e-06,
+      "loss": 0.0496,
+      "step": 1483
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.503518430033167e-06,
+      "loss": 0.0146,
+      "step": 1484
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.491543495421468e-06,
+      "loss": 0.0168,
+      "step": 1485
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.479572398225321e-06,
+      "loss": 0.0356,
+      "step": 1486
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.467605156758014e-06,
+      "loss": 0.0395,
+      "step": 1487
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.45564178932693e-06,
+      "loss": 0.0122,
+      "step": 1488
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.443682314233533e-06,
+      "loss": 0.0039,
+      "step": 1489
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.431726749773322e-06,
+      "loss": 0.0141,
+      "step": 1490
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.419775114235815e-06,
+      "loss": 0.0386,
+      "step": 1491
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.4078274259045316e-06,
+      "loss": 0.0321,
+      "step": 1492
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.3958837030569355e-06,
+      "loss": 0.0266,
+      "step": 1493
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.383943963964439e-06,
+      "loss": 0.0307,
+      "step": 1494
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.372008226892354e-06,
+      "loss": 0.0279,
+      "step": 1495
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.360076510099865e-06,
+      "loss": 0.0126,
+      "step": 1496
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.348148831840016e-06,
+      "loss": 0.0303,
+      "step": 1497
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.336225210359666e-06,
+      "loss": 0.0466,
+      "step": 1498
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.3243056638994755e-06,
+      "loss": 0.0083,
+      "step": 1499
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.312390210693863e-06,
+      "loss": 0.0152,
+      "step": 1500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.300478868970982e-06,
+      "loss": 0.0463,
+      "step": 1501
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.288571656952708e-06,
+      "loss": 0.0281,
+      "step": 1502
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.2766685928545924e-06,
+      "loss": 0.0372,
+      "step": 1503
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.264769694885847e-06,
+      "loss": 0.0202,
+      "step": 1504
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.252874981249297e-06,
+      "loss": 0.0245,
+      "step": 1505
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.240984470141378e-06,
+      "loss": 0.024,
+      "step": 1506
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.229098179752097e-06,
+      "loss": 0.0457,
+      "step": 1507
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.2172161282649944e-06,
+      "loss": 0.0163,
+      "step": 1508
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.205338333857136e-06,
+      "loss": 0.0426,
+      "step": 1509
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.193464814699073e-06,
+      "loss": 0.0501,
+      "step": 1510
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.181595588954808e-06,
+      "loss": 0.0413,
+      "step": 1511
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.169730674781787e-06,
+      "loss": 0.0088,
+      "step": 1512
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.157870090330854e-06,
+      "loss": 0.0321,
+      "step": 1513
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.146013853746236e-06,
+      "loss": 0.0378,
+      "step": 1514
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.134161983165498e-06,
+      "loss": 0.0133,
+      "step": 1515
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.122314496719528e-06,
+      "loss": 0.0299,
+      "step": 1516
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.11047141253252e-06,
+      "loss": 0.0269,
+      "step": 1517
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.098632748721914e-06,
+      "loss": 0.0131,
+      "step": 1518
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.086798523398407e-06,
+      "loss": 0.0337,
+      "step": 1519
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.07496875466589e-06,
+      "loss": 0.0435,
+      "step": 1520
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.06314346062145e-06,
+      "loss": 0.0277,
+      "step": 1521
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.0513226593553175e-06,
+      "loss": 0.0176,
+      "step": 1522
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.039506368950852e-06,
+      "loss": 0.0237,
+      "step": 1523
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.027694607484519e-06,
+      "loss": 0.0654,
+      "step": 1524
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.015887393025847e-06,
+      "loss": 0.0305,
+      "step": 1525
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.00408474363742e-06,
+      "loss": 0.0196,
+      "step": 1526
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.992286677374825e-06,
+      "loss": 0.0143,
+      "step": 1527
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.9804932122866434e-06,
+      "loss": 0.0064,
+      "step": 1528
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.968704366414425e-06,
+      "loss": 0.0159,
+      "step": 1529
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.9569201577926395e-06,
+      "loss": 0.0668,
+      "step": 1530
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 6.9451406044486735e-06,
+      "loss": 0.0383,
+      "step": 1531
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.93336572440279e-06,
+      "loss": 0.0353,
+      "step": 1532
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 6.921595535668094e-06,
+      "loss": 0.0326,
+      "step": 1533
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.0224,
+      "step": 1534
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.898069304148816e-06,
+      "loss": 0.0317,
+      "step": 1535
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 6.886313297354465e-06,
+      "loss": 0.0125,
+      "step": 1536
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.87456205385171e-06,
+      "loss": 0.0176,
+      "step": 1537
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 6.862815591617503e-06,
+      "loss": 0.0297,
+      "step": 1538
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.851073928621488e-06,
+      "loss": 0.0397,
+      "step": 1539
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.839337082825954e-06,
+      "loss": 0.008,
+      "step": 1540
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 6.827605072185837e-06,
+      "loss": 0.0505,
+      "step": 1541
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.815877914648663e-06,
+      "loss": 0.0255,
+      "step": 1542
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.804155628154535e-06,
+      "loss": 0.013,
+      "step": 1543
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 6.792438230636116e-06,
+      "loss": 0.0217,
+      "step": 1544
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.7807257400185745e-06,
+      "loss": 0.0341,
+      "step": 1545
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 6.769018174219584e-06,
+      "loss": 0.0482,
+      "step": 1546
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.757315551149279e-06,
+      "loss": 0.0287,
+      "step": 1547
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.7456178887102294e-06,
+      "loss": 0.0222,
+      "step": 1548
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 6.733925204797426e-06,
+      "loss": 0.0498,
+      "step": 1549
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.722237517298232e-06,
+      "loss": 0.0259,
+      "step": 1550
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.7105548440923815e-06,
+      "loss": 0.0311,
+      "step": 1551
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.698877203051919e-06,
+      "loss": 0.021,
+      "step": 1552
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.687204612041209e-06,
+      "loss": 0.015,
+      "step": 1553
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.6755370889168816e-06,
+      "loss": 0.0168,
+      "step": 1554
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.6638746515278086e-06,
+      "loss": 0.0231,
+      "step": 1555
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.652217317715093e-06,
+      "loss": 0.0297,
+      "step": 1556
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.6405651053120244e-06,
+      "loss": 0.0213,
+      "step": 1557
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.628918032144064e-06,
+      "loss": 0.0338,
+      "step": 1558
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.6172761160287995e-06,
+      "loss": 0.0299,
+      "step": 1559
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.605639374775934e-06,
+      "loss": 0.0446,
+      "step": 1560
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.5940078261872655e-06,
+      "loss": 0.0355,
+      "step": 1561
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.582381488056627e-06,
+      "loss": 0.023,
+      "step": 1562
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.5707603781699e-06,
+      "loss": 0.027,
+      "step": 1563
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.559144514304962e-06,
+      "loss": 0.0529,
+      "step": 1564
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.547533914231654e-06,
+      "loss": 0.0377,
+      "step": 1565
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.535928595711785e-06,
+      "loss": 0.0273,
+      "step": 1566
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.5243285764990615e-06,
+      "loss": 0.025,
+      "step": 1567
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.512733874339103e-06,
+      "loss": 0.0308,
+      "step": 1568
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.501144506969388e-06,
+      "loss": 0.0147,
+      "step": 1569
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.489560492119225e-06,
+      "loss": 0.0088,
+      "step": 1570
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.477981847509749e-06,
+      "loss": 0.0224,
+      "step": 1571
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.466408590853867e-06,
+      "loss": 0.0173,
+      "step": 1572
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.454840739856257e-06,
+      "loss": 0.0095,
+      "step": 1573
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.4432783122133124e-06,
+      "loss": 0.03,
+      "step": 1574
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.431721325613138e-06,
+      "loss": 0.0157,
+      "step": 1575
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.4201697977355215e-06,
+      "loss": 0.0215,
+      "step": 1576
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.408623746251882e-06,
+      "loss": 0.007,
+      "step": 1577
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.397083188825278e-06,
+      "loss": 0.0321,
+      "step": 1578
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.385548143110358e-06,
+      "loss": 0.031,
+      "step": 1579
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.374018626753331e-06,
+      "loss": 0.0282,
+      "step": 1580
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.36249465739196e-06,
+      "loss": 0.0292,
+      "step": 1581
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.35097625265551e-06,
+      "loss": 0.018,
+      "step": 1582
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.339463430164746e-06,
+      "loss": 0.0393,
+      "step": 1583
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.32795620753188e-06,
+      "loss": 0.0265,
+      "step": 1584
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.316454602360569e-06,
+      "loss": 0.0235,
+      "step": 1585
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.304958632245871e-06,
+      "loss": 0.0434,
+      "step": 1586
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.293468314774218e-06,
+      "loss": 0.0159,
+      "step": 1587
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.28198366752341e-06,
+      "loss": 0.0231,
+      "step": 1588
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.270504708062553e-06,
+      "loss": 0.0183,
+      "step": 1589
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.2590314539520695e-06,
+      "loss": 0.0259,
+      "step": 1590
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.247563922743647e-06,
+      "loss": 0.016,
+      "step": 1591
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.23610213198021e-06,
+      "loss": 0.0439,
+      "step": 1592
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.2246460991959155e-06,
+      "loss": 0.0204,
+      "step": 1593
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.213195841916104e-06,
+      "loss": 0.0821,
+      "step": 1594
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.2017513776572855e-06,
+      "loss": 0.0205,
+      "step": 1595
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.190312723927101e-06,
+      "loss": 0.0595,
+      "step": 1596
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.178879898224304e-06,
+      "loss": 0.0134,
+      "step": 1597
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.167452918038744e-06,
+      "loss": 0.0201,
+      "step": 1598
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.1560318008513056e-06,
+      "loss": 0.0442,
+      "step": 1599
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.144616564133927e-06,
+      "loss": 0.0337,
+      "step": 1600
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.13320722534954e-06,
+      "loss": 0.035,
+      "step": 1601
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.121803801952049e-06,
+      "loss": 0.0206,
+      "step": 1602
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.1104063113863196e-06,
+      "loss": 0.0189,
+      "step": 1603
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.0990147710881345e-06,
+      "loss": 0.0374,
+      "step": 1604
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.0876291984841795e-06,
+      "loss": 0.0444,
+      "step": 1605
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.076249610992003e-06,
+      "loss": 0.0358,
+      "step": 1606
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.064876026020001e-06,
+      "loss": 0.0203,
+      "step": 1607
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.053508460967393e-06,
+      "loss": 0.0171,
+      "step": 1608
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.0421469332241775e-06,
+      "loss": 0.052,
+      "step": 1609
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.03079146017113e-06,
+      "loss": 0.0245,
+      "step": 1610
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.019442059179752e-06,
+      "loss": 0.057,
+      "step": 1611
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.0080987476122624e-06,
+      "loss": 0.0279,
+      "step": 1612
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.996761542821567e-06,
+      "loss": 0.0269,
+      "step": 1613
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.985430462151219e-06,
+      "loss": 0.0205,
+      "step": 1614
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.974105522935416e-06,
+      "loss": 0.0379,
+      "step": 1615
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.962786742498949e-06,
+      "loss": 0.0244,
+      "step": 1616
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.951474138157201e-06,
+      "loss": 0.0677,
+      "step": 1617
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.940167727216091e-06,
+      "loss": 0.0306,
+      "step": 1618
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9288675269720705e-06,
+      "loss": 0.0302,
+      "step": 1619
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9175735547120975e-06,
+      "loss": 0.0314,
+      "step": 1620
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.906285827713587e-06,
+      "loss": 0.0159,
+      "step": 1621
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.895004363244413e-06,
+      "loss": 0.0164,
+      "step": 1622
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.883729178562866e-06,
+      "loss": 0.0047,
+      "step": 1623
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.872460290917621e-06,
+      "loss": 0.0322,
+      "step": 1624
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.8611977175477355e-06,
+      "loss": 0.0464,
+      "step": 1625
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.849941475682591e-06,
+      "loss": 0.0371,
+      "step": 1626
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.8386915825419e-06,
+      "loss": 0.0181,
+      "step": 1627
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.8274480553356485e-06,
+      "loss": 0.0073,
+      "step": 1628
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.816210911264088e-06,
+      "loss": 0.0195,
+      "step": 1629
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.804980167517712e-06,
+      "loss": 0.0277,
+      "step": 1630
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.793755841277211e-06,
+      "loss": 0.0207,
+      "step": 1631
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.782537949713471e-06,
+      "loss": 0.0109,
+      "step": 1632
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.771326509987522e-06,
+      "loss": 0.0324,
+      "step": 1633
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.760121539250532e-06,
+      "loss": 0.0174,
+      "step": 1634
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.748923054643767e-06,
+      "loss": 0.0282,
+      "step": 1635
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.7377310732985805e-06,
+      "loss": 0.0466,
+      "step": 1636
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.7265456123363695e-06,
+      "loss": 0.0432,
+      "step": 1637
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.7153666888685576e-06,
+      "loss": 0.0483,
+      "step": 1638
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.704194319996561e-06,
+      "loss": 0.0105,
+      "step": 1639
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.693028522811783e-06,
+      "loss": 0.0167,
+      "step": 1640
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.68186931439556e-06,
+      "loss": 0.0177,
+      "step": 1641
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.670716711819159e-06,
+      "loss": 0.0267,
+      "step": 1642
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.659570732143737e-06,
+      "loss": 0.0194,
+      "step": 1643
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.648431392420317e-06,
+      "loss": 0.0314,
+      "step": 1644
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.63729870968977e-06,
+      "loss": 0.0187,
+      "step": 1645
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.626172700982775e-06,
+      "loss": 0.0231,
+      "step": 1646
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.615053383319811e-06,
+      "loss": 0.0117,
+      "step": 1647
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.6039407737111115e-06,
+      "loss": 0.0529,
+      "step": 1648
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.592834889156655e-06,
+      "loss": 0.0418,
+      "step": 1649
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.581735746646134e-06,
+      "loss": 0.0269,
+      "step": 1650
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.570643363158916e-06,
+      "loss": 0.0184,
+      "step": 1651
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.55955775566404e-06,
+      "loss": 0.0062,
+      "step": 1652
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.548478941120171e-06,
+      "loss": 0.0231,
+      "step": 1653
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.537406936475592e-06,
+      "loss": 0.0237,
+      "step": 1654
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.526341758668158e-06,
+      "loss": 0.0328,
+      "step": 1655
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.515283424625279e-06,
+      "loss": 0.0104,
+      "step": 1656
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.5042319512639085e-06,
+      "loss": 0.0239,
+      "step": 1657
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.493187355490495e-06,
+      "loss": 0.0167,
+      "step": 1658
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.48214965420097e-06,
+      "loss": 0.0666,
+      "step": 1659
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.471118864280716e-06,
+      "loss": 0.009,
+      "step": 1660
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.460095002604533e-06,
+      "loss": 0.0348,
+      "step": 1661
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.449078086036643e-06,
+      "loss": 0.0136,
+      "step": 1662
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.4380681314306224e-06,
+      "loss": 0.015,
+      "step": 1663
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.4270651556294116e-06,
+      "loss": 0.0458,
+      "step": 1664
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.416069175465274e-06,
+      "loss": 0.0117,
+      "step": 1665
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.405080207759759e-06,
+      "loss": 0.0106,
+      "step": 1666
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.394098269323702e-06,
+      "loss": 0.0481,
+      "step": 1667
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.383123376957175e-06,
+      "loss": 0.0548,
+      "step": 1668
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.372155547449481e-06,
+      "loss": 0.012,
+      "step": 1669
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.361194797579108e-06,
+      "loss": 0.0198,
+      "step": 1670
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.3502411441137195e-06,
+      "loss": 0.0209,
+      "step": 1671
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.339294603810127e-06,
+      "loss": 0.0314,
+      "step": 1672
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.3283551934142475e-06,
+      "loss": 0.0164,
+      "step": 1673
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.3174229296611064e-06,
+      "loss": 0.0467,
+      "step": 1674
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.306497829274785e-06,
+      "loss": 0.0214,
+      "step": 1675
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.2955799089684044e-06,
+      "loss": 0.0257,
+      "step": 1676
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.284669185444115e-06,
+      "loss": 0.0211,
+      "step": 1677
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.273765675393042e-06,
+      "loss": 0.0228,
+      "step": 1678
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.262869395495287e-06,
+      "loss": 0.0427,
+      "step": 1679
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.2519803624198865e-06,
+      "loss": 0.0292,
+      "step": 1680
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.241098592824795e-06,
+      "loss": 0.0238,
+      "step": 1681
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.2302241033568476e-06,
+      "loss": 0.0166,
+      "step": 1682
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.219356910651743e-06,
+      "loss": 0.0222,
+      "step": 1683
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.208497031334029e-06,
+      "loss": 0.019,
+      "step": 1684
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.197644482017048e-06,
+      "loss": 0.0231,
+      "step": 1685
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.1867992793029435e-06,
+      "loss": 0.0384,
+      "step": 1686
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.1759614397826174e-06,
+      "loss": 0.0178,
+      "step": 1687
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.165130980035697e-06,
+      "loss": 0.0268,
+      "step": 1688
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.154307916630538e-06,
+      "loss": 0.0228,
+      "step": 1689
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.143492266124164e-06,
+      "loss": 0.0371,
+      "step": 1690
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.132684045062268e-06,
+      "loss": 0.0187,
+      "step": 1691
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.1218832699791745e-06,
+      "loss": 0.0389,
+      "step": 1692
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.111089957397817e-06,
+      "loss": 0.0529,
+      "step": 1693
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.10030412382972e-06,
+      "loss": 0.0223,
+      "step": 1694
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.089525785774951e-06,
+      "loss": 0.0428,
+      "step": 1695
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.07875495972213e-06,
+      "loss": 0.0146,
+      "step": 1696
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.0679916621483706e-06,
+      "loss": 0.0317,
+      "step": 1697
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.057235909519271e-06,
+      "loss": 0.0146,
+      "step": 1698
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.0464877182889e-06,
+      "loss": 0.0166,
+      "step": 1699
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.035747104899738e-06,
+      "loss": 0.0426,
+      "step": 1700
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.025014085782693e-06,
+      "loss": 0.0102,
+      "step": 1701
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.014288677357046e-06,
+      "loss": 0.0153,
+      "step": 1702
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.0035708960304315e-06,
+      "loss": 0.0097,
+      "step": 1703
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.992860758198829e-06,
+      "loss": 0.0146,
+      "step": 1704
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.982158280246508e-06,
+      "loss": 0.026,
+      "step": 1705
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.971463478546037e-06,
+      "loss": 0.0304,
+      "step": 1706
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.960776369458225e-06,
+      "loss": 0.036,
+      "step": 1707
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.950096969332123e-06,
+      "loss": 0.016,
+      "step": 1708
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.939425294504994e-06,
+      "loss": 0.0489,
+      "step": 1709
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.928761361302269e-06,
+      "loss": 0.0263,
+      "step": 1710
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.918105186037548e-06,
+      "loss": 0.0166,
+      "step": 1711
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.90745678501255e-06,
+      "loss": 0.0063,
+      "step": 1712
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.896816174517117e-06,
+      "loss": 0.0377,
+      "step": 1713
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.886183370829158e-06,
+      "loss": 0.0812,
+      "step": 1714
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.875558390214652e-06,
+      "loss": 0.0476,
+      "step": 1715
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.864941248927604e-06,
+      "loss": 0.0125,
+      "step": 1716
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.854331963210023e-06,
+      "loss": 0.0089,
+      "step": 1717
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.843730549291909e-06,
+      "loss": 0.0242,
+      "step": 1718
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.8331370233912145e-06,
+      "loss": 0.0225,
+      "step": 1719
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.8225514017138205e-06,
+      "loss": 0.061,
+      "step": 1720
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.811973700453531e-06,
+      "loss": 0.0246,
+      "step": 1721
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.801403935792013e-06,
+      "loss": 0.0505,
+      "step": 1722
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.79084212389881e-06,
+      "loss": 0.0206,
+      "step": 1723
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.7802882809312946e-06,
+      "loss": 0.0404,
+      "step": 1724
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.76974242303464e-06,
+      "loss": 0.032,
+      "step": 1725
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.7592045663418175e-06,
+      "loss": 0.0158,
+      "step": 1726
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.748674726973543e-06,
+      "loss": 0.0194,
+      "step": 1727
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.738152921038286e-06,
+      "loss": 0.0166,
+      "step": 1728
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.727639164632205e-06,
+      "loss": 0.0273,
+      "step": 1729
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.717133473839163e-06,
+      "loss": 0.0176,
+      "step": 1730
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.706635864730677e-06,
+      "loss": 0.0183,
+      "step": 1731
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.696146353365897e-06,
+      "loss": 0.0233,
+      "step": 1732
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.685664955791594e-06,
+      "loss": 0.011,
+      "step": 1733
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.675191688042116e-06,
+      "loss": 0.0181,
+      "step": 1734
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.6647265661393806e-06,
+      "loss": 0.0185,
+      "step": 1735
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.654269606092843e-06,
+      "loss": 0.0387,
+      "step": 1736
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.643820823899473e-06,
+      "loss": 0.0176,
+      "step": 1737
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.633380235543733e-06,
+      "loss": 0.0089,
+      "step": 1738
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.6229478569975424e-06,
+      "loss": 0.0409,
+      "step": 1739
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.612523704220264e-06,
+      "loss": 0.0095,
+      "step": 1740
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.602107793158684e-06,
+      "loss": 0.0219,
+      "step": 1741
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.59170013974697e-06,
+      "loss": 0.0192,
+      "step": 1742
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.581300759906666e-06,
+      "loss": 0.0278,
+      "step": 1743
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.570909669546651e-06,
+      "loss": 0.005,
+      "step": 1744
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.56052688456313e-06,
+      "loss": 0.0482,
+      "step": 1745
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.550152420839604e-06,
+      "loss": 0.0556,
+      "step": 1746
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.539786294246831e-06,
+      "loss": 0.0113,
+      "step": 1747
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.529428520642832e-06,
+      "loss": 0.0165,
+      "step": 1748
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.5190791158728355e-06,
+      "loss": 0.0201,
+      "step": 1749
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.508738095769278e-06,
+      "loss": 0.0219,
+      "step": 1750
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.4984054761517595e-06,
+      "loss": 0.015,
+      "step": 1751
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.488081272827036e-06,
+      "loss": 0.02,
+      "step": 1752
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.4777655015889935e-06,
+      "loss": 0.0188,
+      "step": 1753
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.4674581782186e-06,
+      "loss": 0.0198,
+      "step": 1754
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.457159318483922e-06,
+      "loss": 0.023,
+      "step": 1755
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.446868938140063e-06,
+      "loss": 0.0028,
+      "step": 1756
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.436587052929158e-06,
+      "loss": 0.0253,
+      "step": 1757
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.426313678580348e-06,
+      "loss": 0.0262,
+      "step": 1758
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.416048830809759e-06,
+      "loss": 0.0179,
+      "step": 1759
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.405792525320469e-06,
+      "loss": 0.0067,
+      "step": 1760
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.395544777802484e-06,
+      "loss": 0.0552,
+      "step": 1761
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.385305603932718e-06,
+      "loss": 0.0243,
+      "step": 1762
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.375075019374981e-06,
+      "loss": 0.0163,
+      "step": 1763
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.364853039779926e-06,
+      "loss": 0.023,
+      "step": 1764
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.354639680785059e-06,
+      "loss": 0.0125,
+      "step": 1765
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.3444349580146836e-06,
+      "loss": 0.0129,
+      "step": 1766
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.3342388870799015e-06,
+      "loss": 0.0496,
+      "step": 1767
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.324051483578579e-06,
+      "loss": 0.0524,
+      "step": 1768
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.313872763095312e-06,
+      "loss": 0.0093,
+      "step": 1769
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.30370274120143e-06,
+      "loss": 0.0265,
+      "step": 1770
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.293541433454943e-06,
+      "loss": 0.0144,
+      "step": 1771
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.283388855400529e-06,
+      "loss": 0.0186,
+      "step": 1772
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.273245022569519e-06,
+      "loss": 0.0142,
+      "step": 1773
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.263109950479863e-06,
+      "loss": 0.0109,
+      "step": 1774
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.252983654636115e-06,
+      "loss": 0.0405,
+      "step": 1775
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.242866150529387e-06,
+      "loss": 0.0343,
+      "step": 1776
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.2327574536373585e-06,
+      "loss": 0.0359,
+      "step": 1777
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.222657579424225e-06,
+      "loss": 0.0181,
+      "step": 1778
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.2125665433406855e-06,
+      "loss": 0.0056,
+      "step": 1779
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.202484360823925e-06,
+      "loss": 0.0249,
+      "step": 1780
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.192411047297582e-06,
+      "loss": 0.0408,
+      "step": 1781
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.18234661817173e-06,
+      "loss": 0.0175,
+      "step": 1782
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.172291088842842e-06,
+      "loss": 0.0061,
+      "step": 1783
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.162244474693781e-06,
+      "loss": 0.0195,
+      "step": 1784
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.152206791093777e-06,
+      "loss": 0.0134,
+      "step": 1785
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.1421780533983865e-06,
+      "loss": 0.008,
+      "step": 1786
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.1321582769494975e-06,
+      "loss": 0.0087,
+      "step": 1787
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.0319,
+      "step": 1788
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.112145669090144e-06,
+      "loss": 0.0368,
+      "step": 1789
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.1021528682948064e-06,
+      "loss": 0.0314,
+      "step": 1790
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.092169089976152e-06,
+      "loss": 0.0111,
+      "step": 1791
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.082194349407285e-06,
+      "loss": 0.016,
+      "step": 1792
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.072228661847478e-06,
+      "loss": 0.0094,
+      "step": 1793
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.062272042542151e-06,
+      "loss": 0.0175,
+      "step": 1794
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.052324506722861e-06,
+      "loss": 0.0102,
+      "step": 1795
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.042386069607264e-06,
+      "loss": 0.0137,
+      "step": 1796
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.0324567463990986e-06,
+      "loss": 0.0369,
+      "step": 1797
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.022536552288158e-06,
+      "loss": 0.0133,
+      "step": 1798
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.012625502450267e-06,
+      "loss": 0.0173,
+      "step": 1799
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.002723612047272e-06,
+      "loss": 0.0195,
+      "step": 1800
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.992830896226997e-06,
+      "loss": 0.0057,
+      "step": 1801
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.982947370123235e-06,
+      "loss": 0.0236,
+      "step": 1802
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.973073048855727e-06,
+      "loss": 0.0548,
+      "step": 1803
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 3.963207947530118e-06,
+      "loss": 0.013,
+      "step": 1804
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.953352081237963e-06,
+      "loss": 0.0347,
+      "step": 1805
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.943505465056677e-06,
+      "loss": 0.0166,
+      "step": 1806
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.933668114049537e-06,
+      "loss": 0.0181,
+      "step": 1807
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.923840043265633e-06,
+      "loss": 0.0238,
+      "step": 1808
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 3.91402126773987e-06,
+      "loss": 0.0239,
+      "step": 1809
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.904211802492922e-06,
+      "loss": 0.0129,
+      "step": 1810
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.894411662531229e-06,
+      "loss": 0.0236,
+      "step": 1811
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.884620862846967e-06,
+      "loss": 0.0254,
+      "step": 1812
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.874839418418009e-06,
+      "loss": 0.0205,
+      "step": 1813
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.865067344207935e-06,
+      "loss": 0.0212,
+      "step": 1814
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.855304655165978e-06,
+      "loss": 0.0308,
+      "step": 1815
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 3.8455513662270125e-06,
+      "loss": 0.0143,
+      "step": 1816
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.835807492311543e-06,
+      "loss": 0.0127,
+      "step": 1817
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.826073048325664e-06,
+      "loss": 0.0201,
+      "step": 1818
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.816348049161048e-06,
+      "loss": 0.0535,
+      "step": 1819
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.8066325096949153e-06,
+      "loss": 0.0039,
+      "step": 1820
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.796926444790012e-06,
+      "loss": 0.0151,
+      "step": 1821
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.7872298692945995e-06,
+      "loss": 0.0154,
+      "step": 1822
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.7775427980424116e-06,
+      "loss": 0.0044,
+      "step": 1823
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.767865245852652e-06,
+      "loss": 0.0132,
+      "step": 1824
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.7581972275299606e-06,
+      "loss": 0.0192,
+      "step": 1825
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.748538757864383e-06,
+      "loss": 0.0271,
+      "step": 1826
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.7388898516313708e-06,
+      "loss": 0.0202,
+      "step": 1827
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.7292505235917353e-06,
+      "loss": 0.0051,
+      "step": 1828
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.7196207884916424e-06,
+      "loss": 0.0355,
+      "step": 1829
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.710000661062578e-06,
+      "loss": 0.0394,
+      "step": 1830
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.7003901560213283e-06,
+      "loss": 0.0366,
+      "step": 1831
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.690789288069965e-06,
+      "loss": 0.0117,
+      "step": 1832
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.6811980718958164e-06,
+      "loss": 0.021,
+      "step": 1833
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.671616522171445e-06,
+      "loss": 0.0198,
+      "step": 1834
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.6620446535546227e-06,
+      "loss": 0.0148,
+      "step": 1835
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.6524824806883097e-06,
+      "loss": 0.0227,
+      "step": 1836
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.6429300182006445e-06,
+      "loss": 0.0208,
+      "step": 1837
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.633387280704895e-06,
+      "loss": 0.0359,
+      "step": 1838
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.623854282799465e-06,
+      "loss": 0.0276,
+      "step": 1839
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.6143310390678544e-06,
+      "loss": 0.0261,
+      "step": 1840
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.604817564078643e-06,
+      "loss": 0.0309,
+      "step": 1841
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.595313872385463e-06,
+      "loss": 0.0507,
+      "step": 1842
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.5858199785269763e-06,
+      "loss": 0.0243,
+      "step": 1843
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.576335897026869e-06,
+      "loss": 0.0057,
+      "step": 1844
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.566861642393803e-06,
+      "loss": 0.0529,
+      "step": 1845
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.5573972291214154e-06,
+      "loss": 0.0198,
+      "step": 1846
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.5479426716882882e-06,
+      "loss": 0.0309,
+      "step": 1847
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.538497984557917e-06,
+      "loss": 0.0271,
+      "step": 1848
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.5290631821787124e-06,
+      "loss": 0.06,
+      "step": 1849
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.5196382789839477e-06,
+      "loss": 0.0416,
+      "step": 1850
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.5102232893917667e-06,
+      "loss": 0.0376,
+      "step": 1851
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.5008182278051384e-06,
+      "loss": 0.0384,
+      "step": 1852
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.491423108611842e-06,
+      "loss": 0.0321,
+      "step": 1853
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.4820379461844555e-06,
+      "loss": 0.0329,
+      "step": 1854
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.4726627548803205e-06,
+      "loss": 0.0307,
+      "step": 1855
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.463297549041529e-06,
+      "loss": 0.0232,
+      "step": 1856
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.4539423429948903e-06,
+      "loss": 0.0294,
+      "step": 1857
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.444597151051917e-06,
+      "loss": 0.032,
+      "step": 1858
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.4352619875088112e-06,
+      "loss": 0.0317,
+      "step": 1859
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.425936866646419e-06,
+      "loss": 0.0242,
+      "step": 1860
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.4166218027302366e-06,
+      "loss": 0.0267,
+      "step": 1861
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.4073168100103725e-06,
+      "loss": 0.0104,
+      "step": 1862
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.3980219027215187e-06,
+      "loss": 0.0175,
+      "step": 1863
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.3887370950829536e-06,
+      "loss": 0.0325,
+      "step": 1864
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.3794624012984913e-06,
+      "loss": 0.0204,
+      "step": 1865
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.3701978355564835e-06,
+      "loss": 0.0206,
+      "step": 1866
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.3609434120297823e-06,
+      "loss": 0.0143,
+      "step": 1867
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.3516991448757265e-06,
+      "loss": 0.0143,
+      "step": 1868
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.342465048236122e-06,
+      "loss": 0.0078,
+      "step": 1869
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.3332411362372063e-06,
+      "loss": 0.0139,
+      "step": 1870
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.3240274229896475e-06,
+      "loss": 0.0263,
+      "step": 1871
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.314823922588499e-06,
+      "loss": 0.0227,
+      "step": 1872
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.3056306491132062e-06,
+      "loss": 0.0125,
+      "step": 1873
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.2964476166275572e-06,
+      "loss": 0.0271,
+      "step": 1874
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.2872748391796736e-06,
+      "loss": 0.0192,
+      "step": 1875
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.2781123308019967e-06,
+      "loss": 0.0338,
+      "step": 1876
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.268960105511255e-06,
+      "loss": 0.0211,
+      "step": 1877
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.259818177308447e-06,
+      "loss": 0.0287,
+      "step": 1878
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.250686560178815e-06,
+      "loss": 0.0219,
+      "step": 1879
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.2415652680918262e-06,
+      "loss": 0.0261,
+      "step": 1880
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.2324543150011633e-06,
+      "loss": 0.0686,
+      "step": 1881
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.223353714844677e-06,
+      "loss": 0.0395,
+      "step": 1882
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.214263481544394e-06,
+      "loss": 0.0187,
+      "step": 1883
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.2051836290064773e-06,
+      "loss": 0.0202,
+      "step": 1884
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.196114171121205e-06,
+      "loss": 0.0155,
+      "step": 1885
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.1870551217629608e-06,
+      "loss": 0.0078,
+      "step": 1886
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.1780064947901957e-06,
+      "loss": 0.0181,
+      "step": 1887
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.168968304045431e-06,
+      "loss": 0.0229,
+      "step": 1888
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.159940563355207e-06,
+      "loss": 0.021,
+      "step": 1889
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.1509232865300878e-06,
+      "loss": 0.0099,
+      "step": 1890
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.1419164873646313e-06,
+      "loss": 0.029,
+      "step": 1891
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.1329201796373564e-06,
+      "loss": 0.0444,
+      "step": 1892
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.1239343771107444e-06,
+      "loss": 0.0275,
+      "step": 1893
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.1149590935311968e-06,
+      "loss": 0.0081,
+      "step": 1894
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.1059943426290228e-06,
+      "loss": 0.0302,
+      "step": 1895
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.097040138118431e-06,
+      "loss": 0.0281,
+      "step": 1896
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.08809649369748e-06,
+      "loss": 0.0283,
+      "step": 1897
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.0791634230480836e-06,
+      "loss": 0.0085,
+      "step": 1898
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.070240939835983e-06,
+      "loss": 0.0504,
+      "step": 1899
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.061329057710711e-06,
+      "loss": 0.0422,
+      "step": 1900
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.0524277903055942e-06,
+      "loss": 0.0334,
+      "step": 1901
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.043537151237712e-06,
+      "loss": 0.0309,
+      "step": 1902
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.0346571541078927e-06,
+      "loss": 0.0213,
+      "step": 1903
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.0257878125006767e-06,
+      "loss": 0.0289,
+      "step": 1904
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.0169291399843105e-06,
+      "loss": 0.0233,
+      "step": 1905
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.0080811501107187e-06,
+      "loss": 0.0415,
+      "step": 1906
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.9992438564154767e-06,
+      "loss": 0.0065,
+      "step": 1907
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.990417272417806e-06,
+      "loss": 0.0279,
+      "step": 1908
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.981601411620536e-06,
+      "loss": 0.0504,
+      "step": 1909
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.9727962875101e-06,
+      "loss": 0.0226,
+      "step": 1910
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.964001913556498e-06,
+      "loss": 0.0143,
+      "step": 1911
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.955218303213292e-06,
+      "loss": 0.0187,
+      "step": 1912
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.946445469917576e-06,
+      "loss": 0.0144,
+      "step": 1913
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.9376834270899523e-06,
+      "loss": 0.0279,
+      "step": 1914
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.0171,
+      "step": 1915
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.9201917664388634e-06,
+      "loss": 0.0241,
+      "step": 1916
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.9114621753739868e-06,
+      "loss": 0.014,
+      "step": 1917
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.9027434282943555e-06,
+      "loss": 0.0403,
+      "step": 1918
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.894035538537829e-06,
+      "loss": 0.0289,
+      "step": 1919
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.8853385194256677e-06,
+      "loss": 0.0117,
+      "step": 1920
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.8766523842625006e-06,
+      "loss": 0.0389,
+      "step": 1921
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.8679771463362973e-06,
+      "loss": 0.0091,
+      "step": 1922
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.8593128189183692e-06,
+      "loss": 0.0303,
+      "step": 1923
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.850659415263326e-06,
+      "loss": 0.0144,
+      "step": 1924
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.8420169486090765e-06,
+      "loss": 0.0292,
+      "step": 1925
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.833385432176787e-06,
+      "loss": 0.0051,
+      "step": 1926
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.8247648791708836e-06,
+      "loss": 0.0117,
+      "step": 1927
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.8161553027790167e-06,
+      "loss": 0.0287,
+      "step": 1928
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.8075567161720376e-06,
+      "loss": 0.0068,
+      "step": 1929
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.798969132503997e-06,
+      "loss": 0.0081,
+      "step": 1930
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.790392564912101e-06,
+      "loss": 0.0226,
+      "step": 1931
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.781827026516719e-06,
+      "loss": 0.0154,
+      "step": 1932
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.7732725304213314e-06,
+      "loss": 0.0139,
+      "step": 1933
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.7647290897125368e-06,
+      "loss": 0.0356,
+      "step": 1934
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.7561967174600234e-06,
+      "loss": 0.0347,
+      "step": 1935
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.747675426716534e-06,
+      "loss": 0.096,
+      "step": 1936
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.739165230517874e-06,
+      "loss": 0.0369,
+      "step": 1937
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.730666141882866e-06,
+      "loss": 0.0125,
+      "step": 1938
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.7221781738133425e-06,
+      "loss": 0.0377,
+      "step": 1939
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.713701339294129e-06,
+      "loss": 0.0359,
+      "step": 1940
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.705235651293011e-06,
+      "loss": 0.0252,
+      "step": 1941
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.6967811227607297e-06,
+      "loss": 0.0278,
+      "step": 1942
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.688337766630953e-06,
+      "loss": 0.0099,
+      "step": 1943
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.6799055958202515e-06,
+      "loss": 0.006,
+      "step": 1944
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.6714846232280932e-06,
+      "loss": 0.0288,
+      "step": 1945
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.6630748617368053e-06,
+      "loss": 0.0134,
+      "step": 1946
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.654676324211576e-06,
+      "loss": 0.0255,
+      "step": 1947
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.64628902350041e-06,
+      "loss": 0.0129,
+      "step": 1948
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.6379129724341347e-06,
+      "loss": 0.0105,
+      "step": 1949
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.6295481838263628e-06,
+      "loss": 0.0674,
+      "step": 1950
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.6211946704734714e-06,
+      "loss": 0.0083,
+      "step": 1951
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.612852445154601e-06,
+      "loss": 0.0273,
+      "step": 1952
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.604521520631614e-06,
+      "loss": 0.0257,
+      "step": 1953
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.5962019096490876e-06,
+      "loss": 0.0242,
+      "step": 1954
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.587893624934292e-06,
+      "loss": 0.0203,
+      "step": 1955
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.579596679197173e-06,
+      "loss": 0.0277,
+      "step": 1956
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.57131108513033e-06,
+      "loss": 0.0239,
+      "step": 1957
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.5630368554089935e-06,
+      "loss": 0.0345,
+      "step": 1958
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.5547740026910053e-06,
+      "loss": 0.0242,
+      "step": 1959
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.5465225396168134e-06,
+      "loss": 0.0114,
+      "step": 1960
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.538282478809432e-06,
+      "loss": 0.0204,
+      "step": 1961
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.5300538328744408e-06,
+      "loss": 0.0268,
+      "step": 1962
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.5218366143999484e-06,
+      "loss": 0.0042,
+      "step": 1963
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.5136308359565887e-06,
+      "loss": 0.0255,
+      "step": 1964
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.505436510097494e-06,
+      "loss": 0.0186,
+      "step": 1965
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.49725364935827e-06,
+      "loss": 0.0201,
+      "step": 1966
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4890822662569935e-06,
+      "loss": 0.0156,
+      "step": 1967
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.480922373294171e-06,
+      "loss": 0.0217,
+      "step": 1968
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.472773982952744e-06,
+      "loss": 0.0232,
+      "step": 1969
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.464637107698046e-06,
+      "loss": 0.0187,
+      "step": 1970
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4565117599778022e-06,
+      "loss": 0.0348,
+      "step": 1971
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4483979522221025e-06,
+      "loss": 0.0249,
+      "step": 1972
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.440295696843378e-06,
+      "loss": 0.0193,
+      "step": 1973
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4322050062363945e-06,
+      "loss": 0.0106,
+      "step": 1974
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4241258927782197e-06,
+      "loss": 0.0283,
+      "step": 1975
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4160583688282104e-06,
+      "loss": 0.0412,
+      "step": 1976
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4080024467280005e-06,
+      "loss": 0.0188,
+      "step": 1977
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3999581388014692e-06,
+      "loss": 0.0295,
+      "step": 1978
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3919254573547347e-06,
+      "loss": 0.0354,
+      "step": 1979
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.3839044146761236e-06,
+      "loss": 0.033,
+      "step": 1980
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3758950230361533e-06,
+      "loss": 0.0226,
+      "step": 1981
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.3678972946875324e-06,
+      "loss": 0.0222,
+      "step": 1982
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.3599112418651117e-06,
+      "loss": 0.014,
+      "step": 1983
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.351936876785891e-06,
+      "loss": 0.0194,
+      "step": 1984
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.3439742116489827e-06,
+      "loss": 0.024,
+      "step": 1985
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.336023258635607e-06,
+      "loss": 0.0121,
+      "step": 1986
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.328084029909067e-06,
+      "loss": 0.0082,
+      "step": 1987
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.3201565376147228e-06,
+      "loss": 0.0131,
+      "step": 1988
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.3122407938799907e-06,
+      "loss": 0.0352,
+      "step": 1989
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.304336810814305e-06,
+      "loss": 0.0507,
+      "step": 1990
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.296444600509109e-06,
+      "loss": 0.0109,
+      "step": 1991
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.288564175037844e-06,
+      "loss": 0.0379,
+      "step": 1992
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.2806955464559165e-06,
+      "loss": 0.0254,
+      "step": 1993
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.2728387268006924e-06,
+      "loss": 0.071,
+      "step": 1994
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.2649937280914614e-06,
+      "loss": 0.0175,
+      "step": 1995
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.2571605623294412e-06,
+      "loss": 0.0061,
+      "step": 1996
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.24933924149774e-06,
+      "loss": 0.0088,
+      "step": 1997
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.2415297775613478e-06,
+      "loss": 0.0098,
+      "step": 1998
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.2337321824671153e-06,
+      "loss": 0.0232,
+      "step": 1999
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.2259464681437404e-06,
+      "loss": 0.0307,
+      "step": 2000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.218172646501746e-06,
+      "loss": 0.0114,
+      "step": 2001
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.210410729433454e-06,
+      "loss": 0.0168,
+      "step": 2002
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.2026607288129788e-06,
+      "loss": 0.0097,
+      "step": 2003
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.19492265649621e-06,
+      "loss": 0.0121,
+      "step": 2004
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.1871965243207804e-06,
+      "loss": 0.0093,
+      "step": 2005
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.1794823441060663e-06,
+      "loss": 0.0113,
+      "step": 2006
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.17178012765315e-06,
+      "loss": 0.0129,
+      "step": 2007
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.164089886744821e-06,
+      "loss": 0.0093,
+      "step": 2008
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.1564116331455478e-06,
+      "loss": 0.0117,
+      "step": 2009
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.1487453786014513e-06,
+      "loss": 0.0079,
+      "step": 2010
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.1410911348403075e-06,
+      "loss": 0.0179,
+      "step": 2011
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.133448913571513e-06,
+      "loss": 0.0248,
+      "step": 2012
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.12581872648607e-06,
+      "loss": 0.0112,
+      "step": 2013
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.118200585256576e-06,
+      "loss": 0.0475,
+      "step": 2014
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.1105945015371985e-06,
+      "loss": 0.0086,
+      "step": 2015
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.1030004869636655e-06,
+      "loss": 0.1111,
+      "step": 2016
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.0954185531532322e-06,
+      "loss": 0.0215,
+      "step": 2017
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.087848711704673e-06,
+      "loss": 0.0092,
+      "step": 2018
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.0802909741982746e-06,
+      "loss": 0.0501,
+      "step": 2019
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.072745352195794e-06,
+      "loss": 0.0712,
+      "step": 2020
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.0652118572404623e-06,
+      "loss": 0.0427,
+      "step": 2021
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.057690500856958e-06,
+      "loss": 0.0196,
+      "step": 2022
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.0501812945513854e-06,
+      "loss": 0.0254,
+      "step": 2023
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.04268424981127e-06,
+      "loss": 0.0151,
+      "step": 2024
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.0351993781055192e-06,
+      "loss": 0.017,
+      "step": 2025
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.027726690884435e-06,
+      "loss": 0.0116,
+      "step": 2026
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.0202661995796647e-06,
+      "loss": 0.0476,
+      "step": 2027
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.0128179156042092e-06,
+      "loss": 0.0047,
+      "step": 2028
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.005381850352387e-06,
+      "loss": 0.0054,
+      "step": 2029
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.997958015199829e-06,
+      "loss": 0.0312,
+      "step": 2030
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.9905464215034586e-06,
+      "loss": 0.033,
+      "step": 2031
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.983147080601464e-06,
+      "loss": 0.0188,
+      "step": 2032
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.9757600038132987e-06,
+      "loss": 0.0652,
+      "step": 2033
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.96838520243965e-06,
+      "loss": 0.0308,
+      "step": 2034
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.9610226877624217e-06,
+      "loss": 0.0102,
+      "step": 2035
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.9536724710447295e-06,
+      "loss": 0.0587,
+      "step": 2036
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.946334563530874e-06,
+      "loss": 0.0229,
+      "step": 2037
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.939008976446326e-06,
+      "loss": 0.0093,
+      "step": 2038
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.931695720997704e-06,
+      "loss": 0.0232,
+      "step": 2039
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.9243948083727614e-06,
+      "loss": 0.0122,
+      "step": 2040
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.917106249740379e-06,
+      "loss": 0.006,
+      "step": 2041
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.0273,
+      "step": 2042
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.9025662390342646e-06,
+      "loss": 0.0418,
+      "step": 2043
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.8953148092037243e-06,
+      "loss": 0.019,
+      "step": 2044
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.8880757778520742e-06,
+      "loss": 0.0222,
+      "step": 2045
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.8808491560535292e-06,
+      "loss": 0.0164,
+      "step": 2046
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.8736349548633103e-06,
+      "loss": 0.0423,
+      "step": 2047
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.8664331853176454e-06,
+      "loss": 0.0237,
+      "step": 2048
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.859243858433738e-06,
+      "loss": 0.0054,
+      "step": 2049
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.8520669852097573e-06,
+      "loss": 0.0383,
+      "step": 2050
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.8449025766248263e-06,
+      "loss": 0.0181,
+      "step": 2051
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.8377506436389957e-06,
+      "loss": 0.0333,
+      "step": 2052
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.8306111971932361e-06,
+      "loss": 0.0074,
+      "step": 2053
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.8234842482094094e-06,
+      "loss": 0.0303,
+      "step": 2054
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.816369807590258e-06,
+      "loss": 0.023,
+      "step": 2055
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.809267886219399e-06,
+      "loss": 0.0207,
+      "step": 2056
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.8021784949612875e-06,
+      "loss": 0.0397,
+      "step": 2057
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.7951016446612147e-06,
+      "loss": 0.03,
+      "step": 2058
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.7880373461452893e-06,
+      "loss": 0.015,
+      "step": 2059
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.7809856102204148e-06,
+      "loss": 0.0171,
+      "step": 2060
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.7739464476742773e-06,
+      "loss": 0.0305,
+      "step": 2061
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.7669198692753242e-06,
+      "loss": 0.0285,
+      "step": 2062
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.7599058857727614e-06,
+      "loss": 0.0168,
+      "step": 2063
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.7529045078965167e-06,
+      "loss": 0.0298,
+      "step": 2064
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.7459157463572396e-06,
+      "loss": 0.0324,
+      "step": 2065
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.7389396118462831e-06,
+      "loss": 0.0072,
+      "step": 2066
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.731976115035673e-06,
+      "loss": 0.0306,
+      "step": 2067
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.725025266578112e-06,
+      "loss": 0.0421,
+      "step": 2068
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.7180870771069457e-06,
+      "loss": 0.0145,
+      "step": 2069
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.7111615572361628e-06,
+      "loss": 0.0212,
+      "step": 2070
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.7042487175603605e-06,
+      "loss": 0.0177,
+      "step": 2071
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.6973485686547419e-06,
+      "loss": 0.0184,
+      "step": 2072
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.6904611210750988e-06,
+      "loss": 0.0155,
+      "step": 2073
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.6835863853577894e-06,
+      "loss": 0.0176,
+      "step": 2074
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.6767243720197302e-06,
+      "loss": 0.0185,
+      "step": 2075
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.6698750915583673e-06,
+      "loss": 0.0126,
+      "step": 2076
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.6630385544516692e-06,
+      "loss": 0.0465,
+      "step": 2077
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.6562147711581188e-06,
+      "loss": 0.0201,
+      "step": 2078
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.6494037521166772e-06,
+      "loss": 0.0268,
+      "step": 2079
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.642605507746785e-06,
+      "loss": 0.0292,
+      "step": 2080
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.6358200484483434e-06,
+      "loss": 0.0127,
+      "step": 2081
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.629047384601684e-06,
+      "loss": 0.0136,
+      "step": 2082
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.6222875265675764e-06,
+      "loss": 0.0268,
+      "step": 2083
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.615540484687188e-06,
+      "loss": 0.0406,
+      "step": 2084
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.6088062692820939e-06,
+      "loss": 0.0293,
+      "step": 2085
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.602084890654233e-06,
+      "loss": 0.0143,
+      "step": 2086
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.5953763590859162e-06,
+      "loss": 0.0228,
+      "step": 2087
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.5886806848398018e-06,
+      "loss": 0.0465,
+      "step": 2088
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.5819978781588697e-06,
+      "loss": 0.0249,
+      "step": 2089
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.5753279492664264e-06,
+      "loss": 0.0235,
+      "step": 2090
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.5686709083660678e-06,
+      "loss": 0.0222,
+      "step": 2091
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.5620267656416842e-06,
+      "loss": 0.0158,
+      "step": 2092
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.5553955312574253e-06,
+      "loss": 0.0087,
+      "step": 2093
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.5487772153576962e-06,
+      "loss": 0.0263,
+      "step": 2094
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.5421718280671427e-06,
+      "loss": 0.0344,
+      "step": 2095
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.5355793794906305e-06,
+      "loss": 0.0141,
+      "step": 2096
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.5289998797132333e-06,
+      "loss": 0.0154,
+      "step": 2097
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.5224333388002132e-06,
+      "loss": 0.0108,
+      "step": 2098
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.515879766797005e-06,
+      "loss": 0.0185,
+      "step": 2099
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.509339173729214e-06,
+      "loss": 0.0142,
+      "step": 2100
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.5028115696025792e-06,
+      "loss": 0.0303,
+      "step": 2101
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.4962969644029757e-06,
+      "loss": 0.0102,
+      "step": 2102
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.4897953680963939e-06,
+      "loss": 0.0232,
+      "step": 2103
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.4833067906289168e-06,
+      "loss": 0.0138,
+      "step": 2104
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.4768312419267205e-06,
+      "loss": 0.0564,
+      "step": 2105
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.470368731896038e-06,
+      "loss": 0.0209,
+      "step": 2106
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.4639192704231697e-06,
+      "loss": 0.0397,
+      "step": 2107
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.4574828673744413e-06,
+      "loss": 0.035,
+      "step": 2108
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.4510595325962097e-06,
+      "loss": 0.0381,
+      "step": 2109
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.4446492759148411e-06,
+      "loss": 0.0262,
+      "step": 2110
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.4382521071366873e-06,
+      "loss": 0.0346,
+      "step": 2111
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.4318680360480874e-06,
+      "loss": 0.0168,
+      "step": 2112
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4254970724153372e-06,
+      "loss": 0.0242,
+      "step": 2113
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4191392259846804e-06,
+      "loss": 0.0489,
+      "step": 2114
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4127945064823023e-06,
+      "loss": 0.0141,
+      "step": 2115
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.4064629236142957e-06,
+      "loss": 0.0145,
+      "step": 2116
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.4001444870666647e-06,
+      "loss": 0.0152,
+      "step": 2117
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.393839206505304e-06,
+      "loss": 0.0162,
+      "step": 2118
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.3875470915759715e-06,
+      "loss": 0.0286,
+      "step": 2119
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.381268151904298e-06,
+      "loss": 0.0209,
+      "step": 2120
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.375002397095746e-06,
+      "loss": 0.0194,
+      "step": 2121
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.368749836735621e-06,
+      "loss": 0.0433,
+      "step": 2122
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.36251048038903e-06,
+      "loss": 0.0197,
+      "step": 2123
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.3562843376008916e-06,
+      "loss": 0.0301,
+      "step": 2124
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.3500714178959084e-06,
+      "loss": 0.014,
+      "step": 2125
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.3438717307785477e-06,
+      "loss": 0.0087,
+      "step": 2126
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.3376852857330424e-06,
+      "loss": 0.0135,
+      "step": 2127
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.3315120922233604e-06,
+      "loss": 0.0098,
+      "step": 2128
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.325352159693204e-06,
+      "loss": 0.0243,
+      "step": 2129
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.319205497565983e-06,
+      "loss": 0.0121,
+      "step": 2130
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.313072115244811e-06,
+      "loss": 0.009,
+      "step": 2131
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.3069520221124853e-06,
+      "loss": 0.0487,
+      "step": 2132
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.300845227531471e-06,
+      "loss": 0.0416,
+      "step": 2133
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.2947517408438925e-06,
+      "loss": 0.0286,
+      "step": 2134
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.2886715713715126e-06,
+      "loss": 0.0227,
+      "step": 2135
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.2826047284157217e-06,
+      "loss": 0.0171,
+      "step": 2136
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.2765512212575293e-06,
+      "loss": 0.0166,
+      "step": 2137
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.2705110591575343e-06,
+      "loss": 0.0317,
+      "step": 2138
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.2644842513559286e-06,
+      "loss": 0.0259,
+      "step": 2139
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.2584708070724738e-06,
+      "loss": 0.04,
+      "step": 2140
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.2524707355064824e-06,
+      "loss": 0.0246,
+      "step": 2141
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.2464840458368155e-06,
+      "loss": 0.0504,
+      "step": 2142
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.240510747221858e-06,
+      "loss": 0.0167,
+      "step": 2143
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.234550848799515e-06,
+      "loss": 0.0323,
+      "step": 2144
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.2286043596871843e-06,
+      "loss": 0.0105,
+      "step": 2145
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.2226712889817571e-06,
+      "loss": 0.0075,
+      "step": 2146
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.2167516457595973e-06,
+      "loss": 0.0124,
+      "step": 2147
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.2108454390765213e-06,
+      "loss": 0.0559,
+      "step": 2148
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.2049526779677966e-06,
+      "loss": 0.029,
+      "step": 2149
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.1990733714481185e-06,
+      "loss": 0.0373,
+      "step": 2150
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.1932075285115984e-06,
+      "loss": 0.0459,
+      "step": 2151
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.1873551581317543e-06,
+      "loss": 0.0289,
+      "step": 2152
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.1815162692614946e-06,
+      "loss": 0.0083,
+      "step": 2153
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.1756908708331028e-06,
+      "loss": 0.0431,
+      "step": 2154
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.1698789717582203e-06,
+      "loss": 0.0192,
+      "step": 2155
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.1640805809278455e-06,
+      "loss": 0.0248,
+      "step": 2156
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.1582957072123058e-06,
+      "loss": 0.0227,
+      "step": 2157
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.152524359461249e-06,
+      "loss": 0.0427,
+      "step": 2158
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.1467665465036405e-06,
+      "loss": 0.0139,
+      "step": 2159
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.1410222771477276e-06,
+      "loss": 0.0118,
+      "step": 2160
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.1352915601810476e-06,
+      "loss": 0.0116,
+      "step": 2161
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.1295744043704048e-06,
+      "loss": 0.0101,
+      "step": 2162
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.123870818461853e-06,
+      "loss": 0.0111,
+      "step": 2163
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.1181808111806924e-06,
+      "loss": 0.0291,
+      "step": 2164
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.1125043912314438e-06,
+      "loss": 0.0107,
+      "step": 2165
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.1068415672978516e-06,
+      "loss": 0.0122,
+      "step": 2166
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.1011923480428522e-06,
+      "loss": 0.0179,
+      "step": 2167
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.0955567421085733e-06,
+      "loss": 0.0091,
+      "step": 2168
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.0899347581163222e-06,
+      "loss": 0.0199,
+      "step": 2169
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.0843264046665558e-06,
+      "loss": 0.058,
+      "step": 2170
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.0787316903388912e-06,
+      "loss": 0.0119,
+      "step": 2171
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.073150623692073e-06,
+      "loss": 0.0182,
+      "step": 2172
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.0675832132639684e-06,
+      "loss": 0.02,
+      "step": 2173
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.0620294675715548e-06,
+      "loss": 0.0404,
+      "step": 2174
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.0564893951109068e-06,
+      "loss": 0.0167,
+      "step": 2175
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.0509630043571828e-06,
+      "loss": 0.0212,
+      "step": 2176
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.0454503037646059e-06,
+      "loss": 0.028,
+      "step": 2177
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.0399513017664575e-06,
+      "loss": 0.0219,
+      "step": 2178
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.0344660067750666e-06,
+      "loss": 0.0127,
+      "step": 2179
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.0289944271817898e-06,
+      "loss": 0.0428,
+      "step": 2180
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.0235365713570056e-06,
+      "loss": 0.0144,
+      "step": 2181
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.018092447650092e-06,
+      "loss": 0.0158,
+      "step": 2182
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.0126620643894258e-06,
+      "loss": 0.0218,
+      "step": 2183
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.0072454298823619e-06,
+      "loss": 0.0178,
+      "step": 2184
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.001842552415221e-06,
+      "loss": 0.0303,
+      "step": 2185
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 9.9645344025328e-07,
+      "loss": 0.0391,
+      "step": 2186
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 9.910781016407556e-07,
+      "loss": 0.0122,
+      "step": 2187
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 9.857165448007977e-07,
+      "loss": 0.0083,
+      "step": 2188
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 9.803687779354688e-07,
+      "loss": 0.0184,
+      "step": 2189
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 9.750348092257368e-07,
+      "loss": 0.0211,
+      "step": 2190
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 9.697146468314655e-07,
+      "loss": 0.027,
+      "step": 2191
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 9.644082988913894e-07,
+      "loss": 0.0287,
+      "step": 2192
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 9.591157735231206e-07,
+      "loss": 0.0181,
+      "step": 2193
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 9.538370788231154e-07,
+      "loss": 0.0479,
+      "step": 2194
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 9.48572222866676e-07,
+      "loss": 0.01,
+      "step": 2195
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 9.433212137079362e-07,
+      "loss": 0.0108,
+      "step": 2196
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 9.380840593798446e-07,
+      "loss": 0.0421,
+      "step": 2197
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 9.328607678941581e-07,
+      "loss": 0.0503,
+      "step": 2198
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 9.276513472414217e-07,
+      "loss": 0.0386,
+      "step": 2199
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 9.224558053909615e-07,
+      "loss": 0.0039,
+      "step": 2200
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 9.172741502908777e-07,
+      "loss": 0.0572,
+      "step": 2201
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.121063898680171e-07,
+      "loss": 0.0166,
+      "step": 2202
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.069525320279816e-07,
+      "loss": 0.0113,
+      "step": 2203
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.018125846550952e-07,
+      "loss": 0.0115,
+      "step": 2204
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.966865556124061e-07,
+      "loss": 0.0078,
+      "step": 2205
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.915744527416736e-07,
+      "loss": 0.0234,
+      "step": 2206
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.864762838633467e-07,
+      "loss": 0.0098,
+      "step": 2207
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 8.813920567765622e-07,
+      "loss": 0.0178,
+      "step": 2208
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 8.763217792591272e-07,
+      "loss": 0.0148,
+      "step": 2209
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 8.712654590675085e-07,
+      "loss": 0.0195,
+      "step": 2210
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 8.662231039368218e-07,
+      "loss": 0.0265,
+      "step": 2211
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 8.611947215808214e-07,
+      "loss": 0.0129,
+      "step": 2212
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 8.561803196918839e-07,
+      "loss": 0.0197,
+      "step": 2213
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 8.511799059409998e-07,
+      "loss": 0.0141,
+      "step": 2214
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 8.461934879777545e-07,
+      "loss": 0.0299,
+      "step": 2215
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 8.41221073430335e-07,
+      "loss": 0.0409,
+      "step": 2216
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 8.362626699054932e-07,
+      "loss": 0.0183,
+      "step": 2217
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 8.313182849885548e-07,
+      "loss": 0.0136,
+      "step": 2218
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 8.263879262433993e-07,
+      "loss": 0.0537,
+      "step": 2219
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 8.214716012124491e-07,
+      "loss": 0.0093,
+      "step": 2220
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 8.165693174166555e-07,
+      "loss": 0.0126,
+      "step": 2221
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 8.116810823554877e-07,
+      "loss": 0.0103,
+      "step": 2222
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 8.068069035069304e-07,
+      "loss": 0.009,
+      "step": 2223
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 8.019467883274579e-07,
+      "loss": 0.018,
+      "step": 2224
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.971007442520363e-07,
+      "loss": 0.0107,
+      "step": 2225
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.922687786941008e-07,
+      "loss": 0.0172,
+      "step": 2226
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.87450899045551e-07,
+      "loss": 0.0459,
+      "step": 2227
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.826471126767421e-07,
+      "loss": 0.0419,
+      "step": 2228
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.778574269364613e-07,
+      "loss": 0.0176,
+      "step": 2229
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.730818491519343e-07,
+      "loss": 0.0202,
+      "step": 2230
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.683203866287958e-07,
+      "loss": 0.0158,
+      "step": 2231
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.635730466510904e-07,
+      "loss": 0.0203,
+      "step": 2232
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.588398364812611e-07,
+      "loss": 0.0163,
+      "step": 2233
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.54120763360131e-07,
+      "loss": 0.008,
+      "step": 2234
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.494158345069014e-07,
+      "loss": 0.0283,
+      "step": 2235
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.4472505711913e-07,
+      "loss": 0.0117,
+      "step": 2236
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.400484383727269e-07,
+      "loss": 0.0189,
+      "step": 2237
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.35385985421948e-07,
+      "loss": 0.0337,
+      "step": 2238
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.307377053993714e-07,
+      "loss": 0.0103,
+      "step": 2239
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.261036054158965e-07,
+      "loss": 0.0273,
+      "step": 2240
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.214836925607338e-07,
+      "loss": 0.0145,
+      "step": 2241
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.168779739013821e-07,
+      "loss": 0.0117,
+      "step": 2242
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.122864564836351e-07,
+      "loss": 0.0253,
+      "step": 2243
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.077091473315534e-07,
+      "loss": 0.0529,
+      "step": 2244
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.031460534474699e-07,
+      "loss": 0.0156,
+      "step": 2245
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 6.985971818119653e-07,
+      "loss": 0.0216,
+      "step": 2246
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 6.94062539383863e-07,
+      "loss": 0.0234,
+      "step": 2247
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 6.895421331002217e-07,
+      "loss": 0.0285,
+      "step": 2248
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 6.85035969876322e-07,
+      "loss": 0.0062,
+      "step": 2249
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 6.805440566056542e-07,
+      "loss": 0.012,
+      "step": 2250
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 6.760664001599082e-07,
+      "loss": 0.0184,
+      "step": 2251
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 6.716030073889668e-07,
+      "loss": 0.008,
+      "step": 2252
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 6.671538851208903e-07,
+      "loss": 0.007,
+      "step": 2253
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 6.627190401619044e-07,
+      "loss": 0.0093,
+      "step": 2254
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 6.582984792964009e-07,
+      "loss": 0.0082,
+      "step": 2255
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 6.538922092869148e-07,
+      "loss": 0.0168,
+      "step": 2256
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 6.495002368741243e-07,
+      "loss": 0.017,
+      "step": 2257
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 6.451225687768293e-07,
+      "loss": 0.0136,
+      "step": 2258
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 6.40759211691947e-07,
+      "loss": 0.0151,
+      "step": 2259
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 6.364101722945082e-07,
+      "loss": 0.0347,
+      "step": 2260
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.320754572376342e-07,
+      "loss": 0.0385,
+      "step": 2261
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.27755073152535e-07,
+      "loss": 0.0461,
+      "step": 2262
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.234490266485016e-07,
+      "loss": 0.0064,
+      "step": 2263
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 6.191573243128812e-07,
+      "loss": 0.0515,
+      "step": 2264
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 6.148799727110911e-07,
+      "loss": 0.0109,
+      "step": 2265
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 6.106169783865801e-07,
+      "loss": 0.0157,
+      "step": 2266
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 6.063683478608473e-07,
+      "loss": 0.0133,
+      "step": 2267
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 6.021340876334092e-07,
+      "loss": 0.0187,
+      "step": 2268
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.979142041817998e-07,
+      "loss": 0.0072,
+      "step": 2269
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.937087039615619e-07,
+      "loss": 0.0191,
+      "step": 2270
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.895175934062347e-07,
+      "loss": 0.023,
+      "step": 2271
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.853408789273451e-07,
+      "loss": 0.0241,
+      "step": 2272
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.811785669143943e-07,
+      "loss": 0.0335,
+      "step": 2273
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.77030663734851e-07,
+      "loss": 0.036,
+      "step": 2274
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.72897175734145e-07,
+      "loss": 0.0182,
+      "step": 2275
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.687781092356481e-07,
+      "loss": 0.0269,
+      "step": 2276
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.646734705406742e-07,
+      "loss": 0.0368,
+      "step": 2277
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.605832659284682e-07,
+      "loss": 0.0205,
+      "step": 2278
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.56507501656186e-07,
+      "loss": 0.016,
+      "step": 2279
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.524461839589012e-07,
+      "loss": 0.0201,
+      "step": 2280
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.483993190495796e-07,
+      "loss": 0.0336,
+      "step": 2281
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.443669131190843e-07,
+      "loss": 0.0264,
+      "step": 2282
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.403489723361532e-07,
+      "loss": 0.0133,
+      "step": 2283
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.363455028473997e-07,
+      "loss": 0.0334,
+      "step": 2284
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.323565107772977e-07,
+      "loss": 0.0654,
+      "step": 2285
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.283820022281715e-07,
+      "loss": 0.0188,
+      "step": 2286
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.244219832801944e-07,
+      "loss": 0.0126,
+      "step": 2287
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.204764599913659e-07,
+      "loss": 0.0369,
+      "step": 2288
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.165454383975166e-07,
+      "loss": 0.0232,
+      "step": 2289
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.126289245122906e-07,
+      "loss": 0.012,
+      "step": 2290
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.087269243271342e-07,
+      "loss": 0.0227,
+      "step": 2291
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.048394438112969e-07,
+      "loss": 0.0327,
+      "step": 2292
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.009664889118137e-07,
+      "loss": 0.0199,
+      "step": 2293
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.971080655534988e-07,
+      "loss": 0.0108,
+      "step": 2294
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.932641796389348e-07,
+      "loss": 0.0054,
+      "step": 2295
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.0296,
+      "step": 2296
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.85620043640187e-07,
+      "loss": 0.014,
+      "step": 2297
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.818198052499368e-07,
+      "loss": 0.0116,
+      "step": 2298
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.780341276912893e-07,
+      "loss": 0.0322,
+      "step": 2299
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.7426301675554285e-07,
+      "loss": 0.0156,
+      "step": 2300
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.7050647821170856e-07,
+      "loss": 0.0316,
+      "step": 2301
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.6676451780650947e-07,
+      "loss": 0.0192,
+      "step": 2302
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.6303714126436306e-07,
+      "loss": 0.0265,
+      "step": 2303
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.593243542873804e-07,
+      "loss": 0.0215,
+      "step": 2304
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.5562616255534933e-07,
+      "loss": 0.0256,
+      "step": 2305
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.519425717257342e-07,
+      "loss": 0.0115,
+      "step": 2306
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.482735874336619e-07,
+      "loss": 0.0315,
+      "step": 2307
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.4461921529191045e-07,
+      "loss": 0.0172,
+      "step": 2308
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.409794608909102e-07,
+      "loss": 0.0177,
+      "step": 2309
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.3735432979872593e-07,
+      "loss": 0.01,
+      "step": 2310
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.3374382756105173e-07,
+      "loss": 0.0282,
+      "step": 2311
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.301479597012048e-07,
+      "loss": 0.0461,
+      "step": 2312
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.2656673172011055e-07,
+      "loss": 0.0151,
+      "step": 2313
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.230001490963043e-07,
+      "loss": 0.032,
+      "step": 2314
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.194482172859127e-07,
+      "loss": 0.0118,
+      "step": 2315
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.1591094172265256e-07,
+      "loss": 0.0091,
+      "step": 2316
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.1238832781781647e-07,
+      "loss": 0.0179,
+      "step": 2317
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.088803809602693e-07,
+      "loss": 0.0147,
+      "step": 2318
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.053871065164394e-07,
+      "loss": 0.0138,
+      "step": 2319
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.019085098303077e-07,
+      "loss": 0.0164,
+      "step": 2320
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.984445962234018e-07,
+      "loss": 0.0036,
+      "step": 2321
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.9499537099478955e-07,
+      "loss": 0.0232,
+      "step": 2322
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.915608394210646e-07,
+      "loss": 0.0139,
+      "step": 2323
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.8814100675634513e-07,
+      "loss": 0.0064,
+      "step": 2324
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.847358782322608e-07,
+      "loss": 0.0165,
+      "step": 2325
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 3.813454590579502e-07,
+      "loss": 0.0114,
+      "step": 2326
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.779697544200467e-07,
+      "loss": 0.0164,
+      "step": 2327
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.74608769482675e-07,
+      "loss": 0.0269,
+      "step": 2328
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.7126250938744335e-07,
+      "loss": 0.0146,
+      "step": 2329
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.679309792534291e-07,
+      "loss": 0.0449,
+      "step": 2330
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.646141841771811e-07,
+      "loss": 0.0213,
+      "step": 2331
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.613121292327049e-07,
+      "loss": 0.038,
+      "step": 2332
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.5802481947145196e-07,
+      "loss": 0.0265,
+      "step": 2333
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.547522599223263e-07,
+      "loss": 0.023,
+      "step": 2334
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.5149445559165886e-07,
+      "loss": 0.0198,
+      "step": 2335
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.482514114632107e-07,
+      "loss": 0.0095,
+      "step": 2336
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.450231324981657e-07,
+      "loss": 0.0145,
+      "step": 2337
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.418096236351165e-07,
+      "loss": 0.0477,
+      "step": 2338
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.3861088979006197e-07,
+      "loss": 0.0359,
+      "step": 2339
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.354269358563966e-07,
+      "loss": 0.0277,
+      "step": 2340
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.3225776670490874e-07,
+      "loss": 0.0107,
+      "step": 2341
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.2910338718376455e-07,
+      "loss": 0.008,
+      "step": 2342
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.2596380211850744e-07,
+      "loss": 0.022,
+      "step": 2343
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.2283901631205094e-07,
+      "loss": 0.0219,
+      "step": 2344
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.197290345446613e-07,
+      "loss": 0.0308,
+      "step": 2345
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.166338615739661e-07,
+      "loss": 0.0365,
+      "step": 2346
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.135535021349323e-07,
+      "loss": 0.0156,
+      "step": 2347
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.1048796093987033e-07,
+      "loss": 0.0147,
+      "step": 2348
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.074372426784156e-07,
+      "loss": 0.0246,
+      "step": 2349
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.044013520175337e-07,
+      "loss": 0.022,
+      "step": 2350
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.0138029360150533e-07,
+      "loss": 0.0104,
+      "step": 2351
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.983740720519179e-07,
+      "loss": 0.0029,
+      "step": 2352
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.95382691967665e-07,
+      "loss": 0.015,
+      "step": 2353
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.9240615792493377e-07,
+      "loss": 0.0247,
+      "step": 2354
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.894444744772007e-07,
+      "loss": 0.0372,
+      "step": 2355
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.86497646155226e-07,
+      "loss": 0.0056,
+      "step": 2356
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.8356567746704034e-07,
+      "loss": 0.0288,
+      "step": 2357
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.806485728979458e-07,
+      "loss": 0.0172,
+      "step": 2358
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.777463369105038e-07,
+      "loss": 0.0212,
+      "step": 2359
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.7485897394453067e-07,
+      "loss": 0.0171,
+      "step": 2360
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.7198648841709195e-07,
+      "loss": 0.0118,
+      "step": 2361
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.691288847224871e-07,
+      "loss": 0.0339,
+      "step": 2362
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.662861672322592e-07,
+      "loss": 0.0147,
+      "step": 2363
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.634583402951685e-07,
+      "loss": 0.0114,
+      "step": 2364
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.606454082372045e-07,
+      "loss": 0.013,
+      "step": 2365
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.578473753615651e-07,
+      "loss": 0.0182,
+      "step": 2366
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.5506424594865853e-07,
+      "loss": 0.0111,
+      "step": 2367
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.5229602425609123e-07,
+      "loss": 0.0228,
+      "step": 2368
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.4954271451866686e-07,
+      "loss": 0.0261,
+      "step": 2369
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.4680432094837394e-07,
+      "loss": 0.0278,
+      "step": 2370
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.4408084773438477e-07,
+      "loss": 0.0306,
+      "step": 2371
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.413722990430445e-07,
+      "loss": 0.0314,
+      "step": 2372
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.386786790178697e-07,
+      "loss": 0.017,
+      "step": 2373
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.359999917795375e-07,
+      "loss": 0.0441,
+      "step": 2374
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.3333624142587884e-07,
+      "loss": 0.0294,
+      "step": 2375
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.306874320318786e-07,
+      "loss": 0.0299,
+      "step": 2376
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.280535676496609e-07,
+      "loss": 0.0184,
+      "step": 2377
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.2543465230849158e-07,
+      "loss": 0.0159,
+      "step": 2378
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.2283069001476366e-07,
+      "loss": 0.0245,
+      "step": 2379
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.2024168475199615e-07,
+      "loss": 0.0423,
+      "step": 2380
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1766764048083088e-07,
+      "loss": 0.028,
+      "step": 2381
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1510856113901467e-07,
+      "loss": 0.0233,
+      "step": 2382
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1256445064140817e-07,
+      "loss": 0.0033,
+      "step": 2383
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1003531287996814e-07,
+      "loss": 0.0533,
+      "step": 2384
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.0752115172375076e-07,
+      "loss": 0.0125,
+      "step": 2385
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0502197101889608e-07,
+      "loss": 0.0174,
+      "step": 2386
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0253777458863034e-07,
+      "loss": 0.0127,
+      "step": 2387
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.000685662332569e-07,
+      "loss": 0.032,
+      "step": 2388
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.976143497301486e-07,
+      "loss": 0.0258,
+      "step": 2389
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.9517512883374667e-07,
+      "loss": 0.0187,
+      "step": 2390
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.9275090727555068e-07,
+      "loss": 0.028,
+      "step": 2391
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.9034168876411184e-07,
+      "loss": 0.0411,
+      "step": 2392
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8794747698503424e-07,
+      "loss": 0.026,
+      "step": 2393
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8556827560096247e-07,
+      "loss": 0.023,
+      "step": 2394
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8320408825157843e-07,
+      "loss": 0.0188,
+      "step": 2395
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8085491855359792e-07,
+      "loss": 0.0244,
+      "step": 2396
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.785207701007574e-07,
+      "loss": 0.0253,
+      "step": 2397
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.7620164646381943e-07,
+      "loss": 0.0054,
+      "step": 2398
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.738975511905594e-07,
+      "loss": 0.0462,
+      "step": 2399
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.7160848780576334e-07,
+      "loss": 0.0082,
+      "step": 2400
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.693344598112201e-07,
+      "loss": 0.0202,
+      "step": 2401
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.6707547068571805e-07,
+      "loss": 0.0069,
+      "step": 2402
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.6483152388504286e-07,
+      "loss": 0.0132,
+      "step": 2403
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.6260262284196083e-07,
+      "loss": 0.0186,
+      "step": 2404
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.6038877096622995e-07,
+      "loss": 0.0062,
+      "step": 2405
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.5818997164458005e-07,
+      "loss": 0.0304,
+      "step": 2406
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.56006228240716e-07,
+      "loss": 0.0209,
+      "step": 2407
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.5383754409530883e-07,
+      "loss": 0.0137,
+      "step": 2408
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.5168392252599473e-07,
+      "loss": 0.0156,
+      "step": 2409
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.495453668273672e-07,
+      "loss": 0.0274,
+      "step": 2410
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.4742188027096705e-07,
+      "loss": 0.0263,
+      "step": 2411
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.4531346610528908e-07,
+      "loss": 0.0229,
+      "step": 2412
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.4322012755576653e-07,
+      "loss": 0.0269,
+      "step": 2413
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.4114186782477e-07,
+      "loss": 0.0112,
+      "step": 2414
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.3907869009160525e-07,
+      "loss": 0.0169,
+      "step": 2415
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.3703059751250192e-07,
+      "loss": 0.0181,
+      "step": 2416
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.3499759322061824e-07,
+      "loss": 0.0455,
+      "step": 2417
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.3297968032602414e-07,
+      "loss": 0.0538,
+      "step": 2418
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.309768619157059e-07,
+      "loss": 0.0063,
+      "step": 2419
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.289891410535593e-07,
+      "loss": 0.014,
+      "step": 2420
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.2701652078038195e-07,
+      "loss": 0.0323,
+      "step": 2421
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.2505900411387328e-07,
+      "loss": 0.0194,
+      "step": 2422
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 0.0109,
+      "step": 2423
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.2118929355611542e-07,
+      "loss": 0.0533,
+      "step": 2424
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.192771055847197e-07,
+      "loss": 0.0241,
+      "step": 2425
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.1738003305968193e-07,
+      "loss": 0.0459,
+      "step": 2426
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.1549807888312947e-07,
+      "loss": 0.0166,
+      "step": 2427
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.136312459340605e-07,
+      "loss": 0.0136,
+      "step": 2428
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.1177953706833833e-07,
+      "loss": 0.0156,
+      "step": 2429
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.0994295511869257e-07,
+      "loss": 0.0223,
+      "step": 2430
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.0812150289471135e-07,
+      "loss": 0.0197,
+      "step": 2431
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.0631518318283574e-07,
+      "loss": 0.0314,
+      "step": 2432
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.0452399874635977e-07,
+      "loss": 0.0321,
+      "step": 2433
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.0274795232542045e-07,
+      "loss": 0.0161,
+      "step": 2434
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.0098704663699998e-07,
+      "loss": 0.0207,
+      "step": 2435
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 9.924128437491687e-08,
+      "loss": 0.0238,
+      "step": 2436
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 9.751066820982257e-08,
+      "loss": 0.0252,
+      "step": 2437
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 9.579520078919935e-08,
+      "loss": 0.0208,
+      "step": 2438
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 9.409488473735573e-08,
+      "loss": 0.0155,
+      "step": 2439
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 9.240972265541992e-08,
+      "loss": 0.0147,
+      "step": 2440
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 9.073971712133867e-08,
+      "loss": 0.0068,
+      "step": 2441
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 8.908487068987281e-08,
+      "loss": 0.0134,
+      "step": 2442
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 8.744518589259176e-08,
+      "loss": 0.024,
+      "step": 2443
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 8.582066523787236e-08,
+      "loss": 0.0191,
+      "step": 2444
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 8.421131121089221e-08,
+      "loss": 0.025,
+      "step": 2445
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 8.26171262736264e-08,
+      "loss": 0.0122,
+      "step": 2446
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 8.103811286484742e-08,
+      "loss": 0.0248,
+      "step": 2447
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 7.947427340011527e-08,
+      "loss": 0.0069,
+      "step": 2448
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 7.792561027177847e-08,
+      "loss": 0.0447,
+      "step": 2449
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 7.639212584897082e-08,
+      "loss": 0.0277,
+      "step": 2450
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 7.487382247760022e-08,
+      "loss": 0.0374,
+      "step": 2451
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 7.33707024803565e-08,
+      "loss": 0.0077,
+      "step": 2452
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 7.188276815669915e-08,
+      "loss": 0.048,
+      "step": 2453
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 7.04100217828596e-08,
+      "loss": 0.0308,
+      "step": 2454
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 6.895246561183011e-08,
+      "loss": 0.0379,
+      "step": 2455
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 6.751010187336926e-08,
+      "loss": 0.0075,
+      "step": 2456
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 6.608293277399314e-08,
+      "loss": 0.0098,
+      "step": 2457
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 6.467096049697196e-08,
+      "loss": 0.0151,
+      "step": 2458
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 6.327418720232903e-08,
+      "loss": 0.017,
+      "step": 2459
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 6.189261502683619e-08,
+      "loss": 0.0421,
+      "step": 2460
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 6.05262460840128e-08,
+      "loss": 0.0188,
+      "step": 2461
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.917508246411574e-08,
+      "loss": 0.0133,
+      "step": 2462
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.7839126234146006e-08,
+      "loss": 0.0146,
+      "step": 2463
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.651837943783656e-08,
+      "loss": 0.0114,
+      "step": 2464
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.521284409565675e-08,
+      "loss": 0.0509,
+      "step": 2465
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.39225222048001e-08,
+      "loss": 0.0265,
+      "step": 2466
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.2647415739193184e-08,
+      "loss": 0.0176,
+      "step": 2467
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.138752664948121e-08,
+      "loss": 0.034,
+      "step": 2468
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.0142856863033553e-08,
+      "loss": 0.0055,
+      "step": 2469
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.8913408283934874e-08,
+      "loss": 0.0133,
+      "step": 2470
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.769918279298402e-08,
+      "loss": 0.0412,
+      "step": 2471
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.650018224769404e-08,
+      "loss": 0.0292,
+      "step": 2472
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.5316408482286575e-08,
+      "loss": 0.0196,
+      "step": 2473
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.414786330768639e-08,
+      "loss": 0.018,
+      "step": 2474
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.2994548511525735e-08,
+      "loss": 0.0147,
+      "step": 2475
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.185646585813663e-08,
+      "loss": 0.0207,
+      "step": 2476
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.0733617088546395e-08,
+      "loss": 0.0049,
+      "step": 2477
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.962600392047988e-08,
+      "loss": 0.0378,
+      "step": 2478
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.8533628048356145e-08,
+      "loss": 0.0229,
+      "step": 2479
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.745649114328065e-08,
+      "loss": 0.0335,
+      "step": 2480
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.6394594853048634e-08,
+      "loss": 0.0404,
+      "step": 2481
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.534794080214066e-08,
+      "loss": 0.0183,
+      "step": 2482
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.431653059171814e-08,
+      "loss": 0.0217,
+      "step": 2483
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.3300365799625586e-08,
+      "loss": 0.0281,
+      "step": 2484
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.229944798038176e-08,
+      "loss": 0.04,
+      "step": 2485
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.131377866518293e-08,
+      "loss": 0.029,
+      "step": 2486
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.034335936189736e-08,
+      "loss": 0.0194,
+      "step": 2487
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.9388191555064227e-08,
+      "loss": 0.0067,
+      "step": 2488
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.844827670589134e-08,
+      "loss": 0.0191,
+      "step": 2489
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.7523616252252972e-08,
+      "loss": 0.0081,
+      "step": 2490
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.6614211608686492e-08,
+      "loss": 0.0122,
+      "step": 2491
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.572006416639128e-08,
+      "loss": 0.0177,
+      "step": 2492
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.4841175293226494e-08,
+      "loss": 0.0272,
+      "step": 2493
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.3977546333709966e-08,
+      "loss": 0.0139,
+      "step": 2494
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.312917860901154e-08,
+      "loss": 0.0232,
+      "step": 2495
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.229607341695861e-08,
+      "loss": 0.0058,
+      "step": 2496
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.147823203202948e-08,
+      "loss": 0.017,
+      "step": 2497
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.0675655705348908e-08,
+      "loss": 0.0286,
+      "step": 2498
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.9888345664692555e-08,
+      "loss": 0.0286,
+      "step": 2499
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.9116303114480316e-08,
+      "loss": 0.012,
+      "step": 2500
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.835952923577744e-08,
+      "loss": 0.0422,
+      "step": 2501
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.7618025186287858e-08,
+      "loss": 0.0226,
+      "step": 2502
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.6891792100360848e-08,
+      "loss": 0.038,
+      "step": 2503
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.618083108897994e-08,
+      "loss": 0.0348,
+      "step": 2504
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.5485143239767352e-08,
+      "loss": 0.0226,
+      "step": 2505
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.480472961698065e-08,
+      "loss": 0.011,
+      "step": 2506
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.4139591261511653e-08,
+      "loss": 0.0037,
+      "step": 2507
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.3489729190881983e-08,
+      "loss": 0.0119,
+      "step": 2508
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.28551443992464e-08,
+      "loss": 0.0219,
+      "step": 2509
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.2235837857387246e-08,
+      "loss": 0.0112,
+      "step": 2510
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.1631810512713338e-08,
+      "loss": 0.0488,
+      "step": 2511
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.104306328926219e-08,
+      "loss": 0.0716,
+      "step": 2512
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.046959708769335e-08,
+      "loss": 0.0314,
+      "step": 2513
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 9.911412785290619e-09,
+      "loss": 0.0219,
+      "step": 2514
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 9.368511235958722e-09,
+      "loss": 0.012,
+      "step": 2515
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 8.840893270225525e-09,
+      "loss": 0.0266,
+      "step": 2516
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 8.328559695236493e-09,
+      "loss": 0.0142,
+      "step": 2517
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 7.831511294753569e-09,
+      "loss": 0.02,
+      "step": 2518
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 7.349748829158509e-09,
+      "loss": 0.0186,
+      "step": 2519
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.883273035447335e-09,
+      "loss": 0.0113,
+      "step": 2520
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 6.432084627230328e-09,
+      "loss": 0.0186,
+      "step": 2521
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.996184294732033e-09,
+      "loss": 0.023,
+      "step": 2522
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.575572704789034e-09,
+      "loss": 0.022,
+      "step": 2523
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.17025050084885e-09,
+      "loss": 0.0516,
+      "step": 2524
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.7802183029710406e-09,
+      "loss": 0.0277,
+      "step": 2525
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.405476707822765e-09,
+      "loss": 0.0289,
+      "step": 2526
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.046026288681004e-09,
+      "loss": 0.0471,
+      "step": 2527
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.701867595430342e-09,
+      "loss": 0.0074,
+      "step": 2528
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.37300115455963e-09,
+      "loss": 0.0071,
+      "step": 2529
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.0594274691686522e-09,
+      "loss": 0.0156,
+      "step": 2530
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.7611470189581326e-09,
+      "loss": 0.0224,
+      "step": 2531
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.4781602602363954e-09,
+      "loss": 0.0076,
+      "step": 2532
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.2104676259127047e-09,
+      "loss": 0.0303,
+      "step": 2533
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.9580695255017048e-09,
+      "loss": 0.0439,
+      "step": 2534
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.7209663451200897e-09,
+      "loss": 0.0255,
+      "step": 2535
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.4991584474854937e-09,
+      "loss": 0.0344,
+      "step": 2536
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.29264617191871e-09,
+      "loss": 0.0157,
+      "step": 2537
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.1014298343403617e-09,
+      "loss": 0.0086,
+      "step": 2538
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 9.25509727269791e-10,
+      "loss": 0.0134,
+      "step": 2539
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 7.648861198306101e-10,
+      "loss": 0.0179,
+      "step": 2540
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 6.1955925774293e-10,
+      "loss": 0.039,
+      "step": 2541
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.895293633255804e-10,
+      "loss": 0.0118,
+      "step": 2542
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 3.74796635497221e-10,
+      "loss": 0.0176,
+      "step": 2543
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.753612497763403e-10,
+      "loss": 0.016,
+      "step": 2544
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.9122335827681527e-10,
+      "loss": 0.0313,
+      "step": 2545
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.2238308971235234e-10,
+      "loss": 0.0274,
+      "step": 2546
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 6.884054939537698e-11,
+      "loss": 0.0271,
+      "step": 2547
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.059581923370303e-11,
+      "loss": 0.0196,
+      "step": 2548
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 7.648957733863427e-12,
+      "loss": 0.0504,
+      "step": 2549
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0,
+      "loss": 0.0361,
+      "step": 2550
+    },
+    {
+      "epoch": 10.0,
+      "step": 2550,
+      "total_flos": 6.043691515650048e+16,
+      "train_loss": 0.0615128264268495,
+      "train_runtime": 4671.4208,
+      "train_samples_per_second": 4.363,
+      "train_steps_per_second": 0.546
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2550,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 100,
+  "total_flos": 6.043691515650048e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}