diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.1605415860735009,
+  "eval_steps": 500,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-06,
+      "loss": 1.2989,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.157651144475182e-06,
+      "loss": 1.2283,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.004758654352406e-06,
+      "loss": 1.2452,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.315302288950362e-06,
+      "loss": 1.2448,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.331838906210157e-06,
+      "loss": 1.257,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0162409798827585e-05,
+      "loss": 1.2302,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0864647482583226e-05,
+      "loss": 1.212,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1472953433425543e-05,
+      "loss": 1.2297,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.200951730870481e-05,
+      "loss": 1.2718,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.248949005068534e-05,
+      "loss": 1.25,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2923678209823691e-05,
+      "loss": 1.284,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3320060943302767e-05,
+      "loss": 1.2596,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3684697711049637e-05,
+      "loss": 1.212,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4022298627058407e-05,
+      "loss": 1.1908,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4336597560562562e-05,
+      "loss": 1.2259,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4630604577900725e-05,
+      "loss": 1.2106,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4906781718673912e-05,
+      "loss": 1.2671,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.516716845317999e-05,
+      "loss": 1.2532,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5413473174472748e-05,
+      "loss": 1.2591,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.564714119516052e-05,
+      "loss": 1.2781,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5869406136935633e-05,
+      "loss": 1.2656,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6081329354298873e-05,
+      "loss": 1.2559,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6283830587647817e-05,
+      "loss": 1.2312,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.647771208777795e-05,
+      "loss": 1.2369,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6663677812420316e-05,
+      "loss": 1.243,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.684234885552482e-05,
+      "loss": 1.2245,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7014275963057212e-05,
+      "loss": 1.2699,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.717994977153359e-05,
+      "loss": 1.2328,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7339809249103263e-05,
+      "loss": 1.1813,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7494248705037744e-05,
+      "loss": 1.2576,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.764362364944784e-05,
+      "loss": 1.2681,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.778825572237591e-05,
+      "loss": 1.2397,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7928436864176096e-05,
+      "loss": 1.2427,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8064432863149094e-05,
+      "loss": 1.2252,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8196486388793384e-05,
+      "loss": 1.2142,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.832481959765517e-05,
+      "loss": 1.2665,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.844963638206834e-05,
+      "loss": 1.2361,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8571124318947933e-05,
+      "loss": 1.173,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8689456365402042e-05,
+      "loss": 1.2225,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8804792339635704e-05,
+      "loss": 1.2681,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8917280218967596e-05,
+      "loss": 1.2564,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9027057281410815e-05,
+      "loss": 1.2393,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.913425111291042e-05,
+      "loss": 1.2745,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9238980498774055e-05,
+      "loss": 1.2593,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9341356214914966e-05,
+      "loss": 1.2387,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9441481732123002e-05,
+      "loss": 1.2835,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9539453844588976e-05,
+      "loss": 1.2224,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.963536323225313e-05,
+      "loss": 1.2948,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9729294965166455e-05,
+      "loss": 1.2309,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.98213289568955e-05,
+      "loss": 1.2903,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9911540373026317e-05,
+      "loss": 1.2801,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 1.2455,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 1.2622,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996483001172335e-05,
+      "loss": 1.2051,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992966002344665e-05,
+      "loss": 1.2206,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989449003517e-05,
+      "loss": 1.2498,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985932004689332e-05,
+      "loss": 1.2692,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982415005861666e-05,
+      "loss": 1.2869,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978898007034e-05,
+      "loss": 1.2664,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975381008206333e-05,
+      "loss": 1.2224,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971864009378663e-05,
+      "loss": 1.3077,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9968347010550997e-05,
+      "loss": 1.2306,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996483001172333e-05,
+      "loss": 1.2494,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9961313012895664e-05,
+      "loss": 1.2461,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957796014067997e-05,
+      "loss": 1.2963,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995427901524033e-05,
+      "loss": 1.2357,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9950762016412664e-05,
+      "loss": 1.2395,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9947245017584994e-05,
+      "loss": 1.1918,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943728018757328e-05,
+      "loss": 1.2161,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994021101992966e-05,
+      "loss": 1.2473,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9936694021101995e-05,
+      "loss": 1.2574,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.993317702227433e-05,
+      "loss": 1.2367,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9929660023446662e-05,
+      "loss": 1.2481,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9926143024618992e-05,
+      "loss": 1.3223,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9922626025791326e-05,
+      "loss": 1.259,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.991910902696366e-05,
+      "loss": 1.2232,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9915592028135993e-05,
+      "loss": 1.2023,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9912075029308326e-05,
+      "loss": 1.2357,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.990855803048066e-05,
+      "loss": 1.2542,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9905041031652993e-05,
+      "loss": 1.2384,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9901524032825323e-05,
+      "loss": 1.2711,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9898007033997657e-05,
+      "loss": 1.222,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.989449003516999e-05,
+      "loss": 1.2772,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9890973036342324e-05,
+      "loss": 1.28,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9887456037514657e-05,
+      "loss": 1.2511,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988393903868699e-05,
+      "loss": 1.2441,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988042203985932e-05,
+      "loss": 1.1894,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9876905041031655e-05,
+      "loss": 1.2739,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9873388042203988e-05,
+      "loss": 1.2734,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.986987104337632e-05,
+      "loss": 1.2593,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9866354044548652e-05,
+      "loss": 1.222,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9862837045720985e-05,
+      "loss": 1.2288,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.985932004689332e-05,
+      "loss": 1.2263,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9855803048065652e-05,
+      "loss": 1.2753,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9852286049237986e-05,
+      "loss": 1.2921,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9848769050410316e-05,
+      "loss": 1.1903,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.984525205158265e-05,
+      "loss": 1.2407,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9841735052754983e-05,
+      "loss": 1.2721,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9838218053927317e-05,
+      "loss": 1.2731,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.983470105509965e-05,
+      "loss": 1.2303,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9831184056271984e-05,
+      "loss": 1.2578,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9827667057444317e-05,
+      "loss": 1.2874,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9824150058616647e-05,
+      "loss": 1.2545,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.982063305978898e-05,
+      "loss": 1.2261,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9817116060961314e-05,
+      "loss": 1.2481,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9813599062133648e-05,
+      "loss": 1.2261,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.981008206330598e-05,
+      "loss": 1.2741,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9806565064478315e-05,
+      "loss": 1.2026,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9803048065650645e-05,
+      "loss": 1.2126,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.979953106682298e-05,
+      "loss": 1.2648,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9796014067995312e-05,
+      "loss": 1.2309,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9792497069167646e-05,
+      "loss": 1.2655,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.978898007033998e-05,
+      "loss": 1.2746,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9785463071512313e-05,
+      "loss": 1.2686,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9781946072684643e-05,
+      "loss": 1.2459,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9778429073856976e-05,
+      "loss": 1.2854,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.977491207502931e-05,
+      "loss": 1.2334,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9771395076201643e-05,
+      "loss": 1.2291,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9767878077373977e-05,
+      "loss": 1.2283,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.976436107854631e-05,
+      "loss": 1.2449,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9760844079718644e-05,
+      "loss": 1.2809,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9757327080890974e-05,
+      "loss": 1.1972,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9753810082063308e-05,
+      "loss": 1.272,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.975029308323564e-05,
+      "loss": 1.2032,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.974677608440797e-05,
+      "loss": 1.1844,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9743259085580305e-05,
+      "loss": 1.2546,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.973974208675264e-05,
+      "loss": 1.2803,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9736225087924972e-05,
+      "loss": 1.2484,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9732708089097305e-05,
+      "loss": 1.2881,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.972919109026964e-05,
+      "loss": 1.25,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.972567409144197e-05,
+      "loss": 1.2427,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9722157092614303e-05,
+      "loss": 1.2256,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9718640093786636e-05,
+      "loss": 1.2608,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.971512309495897e-05,
+      "loss": 1.2447,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9711606096131303e-05,
+      "loss": 1.2333,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9708089097303637e-05,
+      "loss": 1.2404,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.970457209847597e-05,
+      "loss": 1.2485,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.97010550996483e-05,
+      "loss": 1.281,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9697538100820634e-05,
+      "loss": 1.2533,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9694021101992967e-05,
+      "loss": 1.2072,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.96905041031653e-05,
+      "loss": 1.2717,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9686987104337634e-05,
+      "loss": 1.3016,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9683470105509968e-05,
+      "loss": 1.2499,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9679953106682298e-05,
+      "loss": 1.2427,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.967643610785463e-05,
+      "loss": 1.2591,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9672919109026965e-05,
+      "loss": 1.255,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.96694021101993e-05,
+      "loss": 1.2626,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9665885111371632e-05,
+      "loss": 1.2348,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9662368112543966e-05,
+      "loss": 1.2486,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9658851113716296e-05,
+      "loss": 1.2569,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.965533411488863e-05,
+      "loss": 1.2986,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9651817116060963e-05,
+      "loss": 1.2388,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9648300117233296e-05,
+      "loss": 1.2505,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.964478311840563e-05,
+      "loss": 1.2587,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9641266119577964e-05,
+      "loss": 1.2578,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9637749120750297e-05,
+      "loss": 1.1769,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9634232121922627e-05,
+      "loss": 1.236,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.963071512309496e-05,
+      "loss": 1.2021,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9627198124267294e-05,
+      "loss": 1.2279,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9623681125439624e-05,
+      "loss": 1.2757,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9620164126611958e-05,
+      "loss": 1.2061,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.961664712778429e-05,
+      "loss": 1.2574,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9613130128956625e-05,
+      "loss": 1.2492,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.960961313012896e-05,
+      "loss": 1.2149,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9606096131301292e-05,
+      "loss": 1.2302,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9602579132473622e-05,
+      "loss": 1.217,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9599062133645956e-05,
+      "loss": 1.2454,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.959554513481829e-05,
+      "loss": 1.2599,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9592028135990623e-05,
+      "loss": 1.2616,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9588511137162956e-05,
+      "loss": 1.213,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.958499413833529e-05,
+      "loss": 1.2617,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9581477139507623e-05,
+      "loss": 1.2285,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9577960140679953e-05,
+      "loss": 1.2745,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9574443141852287e-05,
+      "loss": 1.2792,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.957092614302462e-05,
+      "loss": 1.2086,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9567409144196954e-05,
+      "loss": 1.2539,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9563892145369288e-05,
+      "loss": 1.2103,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.956037514654162e-05,
+      "loss": 1.203,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.955685814771395e-05,
+      "loss": 1.2195,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9553341148886285e-05,
+      "loss": 1.2237,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9549824150058618e-05,
+      "loss": 1.2178,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9546307151230952e-05,
+      "loss": 1.2866,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9542790152403285e-05,
+      "loss": 1.254,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.953927315357562e-05,
+      "loss": 1.184,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.953575615474795e-05,
+      "loss": 1.2311,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9532239155920282e-05,
+      "loss": 1.2549,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9528722157092616e-05,
+      "loss": 1.2641,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.952520515826495e-05,
+      "loss": 1.2265,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9521688159437283e-05,
+      "loss": 1.2519,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9518171160609617e-05,
+      "loss": 1.2031,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.951465416178195e-05,
+      "loss": 1.2373,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.951113716295428e-05,
+      "loss": 1.2621,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9507620164126614e-05,
+      "loss": 1.2545,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9504103165298947e-05,
+      "loss": 1.2649,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9500586166471277e-05,
+      "loss": 1.2677,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.949706916764361e-05,
+      "loss": 1.2643,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9493552168815944e-05,
+      "loss": 1.2076,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9490035169988278e-05,
+      "loss": 1.261,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.948651817116061e-05,
+      "loss": 1.276,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9483001172332945e-05,
+      "loss": 1.2272,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9479484173505275e-05,
+      "loss": 1.2521,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.947596717467761e-05,
+      "loss": 1.3199,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9472450175849942e-05,
+      "loss": 1.234,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9468933177022276e-05,
+      "loss": 1.2335,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.946541617819461e-05,
+      "loss": 1.2568,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9461899179366943e-05,
+      "loss": 1.2296,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9458382180539276e-05,
+      "loss": 1.243,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9454865181711606e-05,
+      "loss": 1.2448,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.945134818288394e-05,
+      "loss": 1.2355,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9447831184056273e-05,
+      "loss": 1.2777,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9444314185228607e-05,
+      "loss": 1.2366,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.944079718640094e-05,
+      "loss": 1.169,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9437280187573274e-05,
+      "loss": 1.1827,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9433763188745604e-05,
+      "loss": 1.2845,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9430246189917938e-05,
+      "loss": 1.2342,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.942672919109027e-05,
+      "loss": 1.1907,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9423212192262605e-05,
+      "loss": 1.2877,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9419695193434938e-05,
+      "loss": 1.258,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9416178194607272e-05,
+      "loss": 1.2525,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9412661195779602e-05,
+      "loss": 1.2251,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9409144196951935e-05,
+      "loss": 1.2556,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.940562719812427e-05,
+      "loss": 1.2242,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9402110199296603e-05,
+      "loss": 1.2639,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9398593200468936e-05,
+      "loss": 1.258,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.939507620164127e-05,
+      "loss": 1.2326,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9391559202813603e-05,
+      "loss": 1.256,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9388042203985933e-05,
+      "loss": 1.2565,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9384525205158267e-05,
+      "loss": 1.2284,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.93810082063306e-05,
+      "loss": 1.2535,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.937749120750293e-05,
+      "loss": 1.2684,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9373974208675264e-05,
+      "loss": 1.2659,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9370457209847597e-05,
+      "loss": 1.3303,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.936694021101993e-05,
+      "loss": 1.2513,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9363423212192265e-05,
+      "loss": 1.2688,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9359906213364598e-05,
+      "loss": 1.2292,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9356389214536928e-05,
+      "loss": 1.2336,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9352872215709262e-05,
+      "loss": 1.2747,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9349355216881595e-05,
+      "loss": 1.2719,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.934583821805393e-05,
+      "loss": 1.2507,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9342321219226262e-05,
+      "loss": 1.2515,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9338804220398596e-05,
+      "loss": 1.251,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.933528722157093e-05,
+      "loss": 1.1902,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.933177022274326e-05,
+      "loss": 1.2442,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9328253223915593e-05,
+      "loss": 1.2722,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9324736225087927e-05,
+      "loss": 1.2312,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.932121922626026e-05,
+      "loss": 1.2382,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9317702227432594e-05,
+      "loss": 1.2218,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9314185228604927e-05,
+      "loss": 1.2638,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9310668229777257e-05,
+      "loss": 1.2155,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.930715123094959e-05,
+      "loss": 1.2615,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9303634232121924e-05,
+      "loss": 1.194,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9300117233294258e-05,
+      "loss": 1.2642,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.929660023446659e-05,
+      "loss": 1.2044,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9293083235638925e-05,
+      "loss": 1.2729,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9289566236811255e-05,
+      "loss": 1.2696,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.928604923798359e-05,
+      "loss": 1.2028,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9282532239155922e-05,
+      "loss": 1.2785,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9279015240328256e-05,
+      "loss": 1.2503,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.927549824150059e-05,
+      "loss": 1.2359,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9271981242672923e-05,
+      "loss": 1.2321,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9268464243845256e-05,
+      "loss": 1.2836,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9264947245017586e-05,
+      "loss": 1.3116,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.926143024618992e-05,
+      "loss": 1.3002,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9257913247362253e-05,
+      "loss": 1.2434,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9254396248534583e-05,
+      "loss": 1.2622,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9250879249706917e-05,
+      "loss": 1.2534,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.924736225087925e-05,
+      "loss": 1.2206,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9243845252051584e-05,
+      "loss": 1.2562,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9240328253223918e-05,
+      "loss": 1.2212,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.923681125439625e-05,
+      "loss": 1.2322,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.923329425556858e-05,
+      "loss": 1.2502,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9229777256740915e-05,
+      "loss": 1.2332,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9226260257913248e-05,
+      "loss": 1.2911,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9222743259085582e-05,
+      "loss": 1.2281,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9219226260257915e-05,
+      "loss": 1.1978,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.921570926143025e-05,
+      "loss": 1.1962,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9212192262602582e-05,
+      "loss": 1.3173,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9208675263774912e-05,
+      "loss": 1.2807,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9205158264947246e-05,
+      "loss": 1.2426,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.920164126611958e-05,
+      "loss": 1.2626,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9198124267291913e-05,
+      "loss": 1.1852,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9194607268464247e-05,
+      "loss": 1.2201,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.919109026963658e-05,
+      "loss": 1.2632,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.918757327080891e-05,
+      "loss": 1.2871,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9184056271981244e-05,
+      "loss": 1.2534,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9180539273153577e-05,
+      "loss": 1.254,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.917702227432591e-05,
+      "loss": 1.1965,
+      "step": 287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9173505275498244e-05,
+      "loss": 1.2502,
+      "step": 288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9169988276670578e-05,
+      "loss": 1.301,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9166471277842908e-05,
+      "loss": 1.2426,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.916295427901524e-05,
+      "loss": 1.2193,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9159437280187575e-05,
+      "loss": 1.2496,
+      "step": 292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.915592028135991e-05,
+      "loss": 1.274,
+      "step": 293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.915240328253224e-05,
+      "loss": 1.2228,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9148886283704576e-05,
+      "loss": 1.215,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.914536928487691e-05,
+      "loss": 1.2546,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.914185228604924e-05,
+      "loss": 1.266,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9138335287221573e-05,
+      "loss": 1.2338,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9134818288393906e-05,
+      "loss": 1.3236,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9131301289566236e-05,
+      "loss": 1.2147,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.912778429073857e-05,
+      "loss": 1.2386,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9124267291910904e-05,
+      "loss": 1.2463,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9120750293083237e-05,
+      "loss": 1.1392,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.911723329425557e-05,
+      "loss": 1.2836,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9113716295427904e-05,
+      "loss": 1.2874,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9110199296600234e-05,
+      "loss": 1.2657,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9106682297772568e-05,
+      "loss": 1.2173,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.91031652989449e-05,
+      "loss": 1.1987,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9099648300117235e-05,
+      "loss": 1.2584,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.909613130128957e-05,
+      "loss": 1.2424,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9092614302461902e-05,
+      "loss": 1.2306,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9089097303634235e-05,
+      "loss": 1.2586,
+      "step": 312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9085580304806566e-05,
+      "loss": 1.2662,
+      "step": 313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.90820633059789e-05,
+      "loss": 1.2534,
+      "step": 314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9078546307151233e-05,
+      "loss": 1.2206,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9075029308323566e-05,
+      "loss": 1.2373,
+      "step": 316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.90715123094959e-05,
+      "loss": 1.2332,
+      "step": 317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9067995310668233e-05,
+      "loss": 1.2517,
+      "step": 318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9064478311840563e-05,
+      "loss": 1.3123,
+      "step": 319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9060961313012897e-05,
+      "loss": 1.2091,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.905744431418523e-05,
+      "loss": 1.2734,
+      "step": 321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9053927315357564e-05,
+      "loss": 1.2511,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9050410316529897e-05,
+      "loss": 1.2356,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.904689331770223e-05,
+      "loss": 1.2232,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.904337631887456e-05,
+      "loss": 1.2511,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9039859320046895e-05,
+      "loss": 1.2447,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9036342321219228e-05,
+      "loss": 1.1925,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9032825322391558e-05,
+      "loss": 1.2303,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9029308323563892e-05,
+      "loss": 1.2047,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.902579132473623e-05,
+      "loss": 1.2533,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9022274325908562e-05,
+      "loss": 1.298,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9018757327080892e-05,
+      "loss": 1.2282,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9015240328253226e-05,
+      "loss": 1.2214,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.901172332942556e-05,
+      "loss": 1.2398,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.900820633059789e-05,
+      "loss": 1.2506,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9004689331770223e-05,
+      "loss": 1.2901,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9001172332942557e-05,
+      "loss": 1.2468,
+      "step": 337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.899765533411489e-05,
+      "loss": 1.2553,
+      "step": 338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8994138335287224e-05,
+      "loss": 1.194,
+      "step": 339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8990621336459557e-05,
+      "loss": 1.2191,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8987104337631887e-05,
+      "loss": 1.2174,
+      "step": 341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.898358733880422e-05,
+      "loss": 1.2463,
+      "step": 342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8980070339976554e-05,
+      "loss": 1.2878,
+      "step": 343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8976553341148888e-05,
+      "loss": 1.2729,
+      "step": 344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.897303634232122e-05,
+      "loss": 1.2497,
+      "step": 345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8969519343493555e-05,
+      "loss": 1.2301,
+      "step": 346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.896600234466589e-05,
+      "loss": 1.2085,
+      "step": 347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.896248534583822e-05,
+      "loss": 1.2658,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8958968347010552e-05,
+      "loss": 1.2373,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8955451348182886e-05,
+      "loss": 1.2445,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.895193434935522e-05,
+      "loss": 1.2832,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8948417350527553e-05,
+      "loss": 1.2441,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8944900351699886e-05,
+      "loss": 1.2838,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8941383352872216e-05,
+      "loss": 1.2053,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.893786635404455e-05,
+      "loss": 1.2471,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8934349355216883e-05,
+      "loss": 1.2457,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8930832356389217e-05,
+      "loss": 1.2298,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.892731535756155e-05,
+      "loss": 1.2036,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8923798358733884e-05,
+      "loss": 1.2505,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8920281359906214e-05,
+      "loss": 1.2387,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8916764361078548e-05,
+      "loss": 1.2571,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.891324736225088e-05,
+      "loss": 1.2161,
+      "step": 362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.890973036342321e-05,
+      "loss": 1.2129,
+      "step": 363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8906213364595545e-05,
+      "loss": 1.2685,
+      "step": 364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.890269636576788e-05,
+      "loss": 1.2546,
+      "step": 365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8899179366940215e-05,
+      "loss": 1.2081,
+      "step": 366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8895662368112545e-05,
+      "loss": 1.2629,
+      "step": 367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.889214536928488e-05,
+      "loss": 1.2167,
+      "step": 368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8888628370457212e-05,
+      "loss": 1.3007,
+      "step": 369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8885111371629543e-05,
+      "loss": 1.2934,
+      "step": 370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8881594372801876e-05,
+      "loss": 1.2195,
+      "step": 371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.887807737397421e-05,
+      "loss": 1.2124,
+      "step": 372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8874560375146543e-05,
+      "loss": 1.1977,
+      "step": 373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8871043376318877e-05,
+      "loss": 1.2349,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.886752637749121e-05,
+      "loss": 1.1921,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.886400937866354e-05,
+      "loss": 1.2403,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8860492379835874e-05,
+      "loss": 1.1933,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8856975381008207e-05,
+      "loss": 1.216,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885345838218054e-05,
+      "loss": 1.232,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8849941383352874e-05,
+      "loss": 1.2421,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8846424384525208e-05,
+      "loss": 1.2422,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884290738569754e-05,
+      "loss": 1.266,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.883939038686987e-05,
+      "loss": 1.2018,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8835873388042205e-05,
+      "loss": 1.2345,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.883235638921454e-05,
+      "loss": 1.263,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8828839390386872e-05,
+      "loss": 1.2526,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8825322391559206e-05,
+      "loss": 1.2809,
+      "step": 387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.882180539273154e-05,
+      "loss": 1.2513,
+      "step": 388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.881828839390387e-05,
+      "loss": 1.2032,
+      "step": 389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8814771395076203e-05,
+      "loss": 1.2062,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8811254396248536e-05,
+      "loss": 1.2834,
+      "step": 391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.880773739742087e-05,
+      "loss": 1.2314,
+      "step": 392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8804220398593203e-05,
+      "loss": 1.2579,
+      "step": 393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8800703399765537e-05,
+      "loss": 1.2137,
+      "step": 394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8797186400937867e-05,
+      "loss": 1.2784,
+      "step": 395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.87936694021102e-05,
+      "loss": 1.2697,
+      "step": 396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8790152403282534e-05,
+      "loss": 1.2865,
+      "step": 397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8786635404454864e-05,
+      "loss": 1.2826,
+      "step": 398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8783118405627198e-05,
+      "loss": 1.2356,
+      "step": 399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.877960140679953e-05,
+      "loss": 1.2545,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8776084407971868e-05,
+      "loss": 1.2572,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.87725674091442e-05,
+      "loss": 1.233,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8769050410316532e-05,
+      "loss": 1.2959,
+      "step": 403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8765533411488865e-05,
+      "loss": 1.2132,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8762016412661196e-05,
+      "loss": 1.2736,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.875849941383353e-05,
+      "loss": 1.2537,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8754982415005863e-05,
+      "loss": 1.2195,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8751465416178196e-05,
+      "loss": 1.2707,
+      "step": 408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.874794841735053e-05,
+      "loss": 1.2505,
+      "step": 409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8744431418522863e-05,
+      "loss": 1.2184,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8740914419695193e-05,
+      "loss": 1.2849,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8737397420867527e-05,
+      "loss": 1.2546,
+      "step": 412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873388042203986e-05,
+      "loss": 1.2703,
+      "step": 413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8730363423212194e-05,
+      "loss": 1.2413,
+      "step": 414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8726846424384527e-05,
+      "loss": 1.2217,
+      "step": 415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872332942555686e-05,
+      "loss": 1.2984,
+      "step": 416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871981242672919e-05,
+      "loss": 1.2561,
+      "step": 417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8716295427901525e-05,
+      "loss": 1.1561,
+      "step": 418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8712778429073858e-05,
+      "loss": 1.1898,
+      "step": 419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870926143024619e-05,
+      "loss": 1.2337,
+      "step": 420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8705744431418525e-05,
+      "loss": 1.2401,
+      "step": 421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870222743259086e-05,
+      "loss": 1.2386,
+      "step": 422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8698710433763192e-05,
+      "loss": 1.2284,
+      "step": 423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8695193434935522e-05,
+      "loss": 1.2402,
+      "step": 424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8691676436107856e-05,
+      "loss": 1.2609,
+      "step": 425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.868815943728019e-05,
+      "loss": 1.242,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8684642438452523e-05,
+      "loss": 1.2398,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8681125439624856e-05,
+      "loss": 1.2528,
+      "step": 428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867760844079719e-05,
+      "loss": 1.2041,
+      "step": 429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867409144196952e-05,
+      "loss": 1.2623,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8670574443141854e-05,
+      "loss": 1.1978,
+      "step": 431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8667057444314187e-05,
+      "loss": 1.2116,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8663540445486517e-05,
+      "loss": 1.2481,
+      "step": 433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866002344665885e-05,
+      "loss": 1.2083,
+      "step": 434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8656506447831184e-05,
+      "loss": 1.2342,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865298944900352e-05,
+      "loss": 1.2854,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864947245017585e-05,
+      "loss": 1.299,
+      "step": 437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8645955451348185e-05,
+      "loss": 1.2212,
+      "step": 438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864243845252052e-05,
+      "loss": 1.2349,
+      "step": 439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863892145369285e-05,
+      "loss": 1.2428,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8635404454865182e-05,
+      "loss": 1.2691,
+      "step": 441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8631887456037516e-05,
+      "loss": 1.2134,
+      "step": 442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862837045720985e-05,
+      "loss": 1.2127,
+      "step": 443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8624853458382183e-05,
+      "loss": 1.2266,
+      "step": 444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8621336459554516e-05,
+      "loss": 1.1664,
+      "step": 445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8617819460726846e-05,
+      "loss": 1.2076,
+      "step": 446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861430246189918e-05,
+      "loss": 1.2489,
+      "step": 447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8610785463071513e-05,
+      "loss": 1.2366,
+      "step": 448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8607268464243847e-05,
+      "loss": 1.2396,
+      "step": 449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.860375146541618e-05,
+      "loss": 1.2524,
+      "step": 450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8600234466588514e-05,
+      "loss": 1.2876,
+      "step": 451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8596717467760844e-05,
+      "loss": 1.2603,
+      "step": 452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8593200468933178e-05,
+      "loss": 1.2268,
+      "step": 453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858968347010551e-05,
+      "loss": 1.2598,
+      "step": 454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8586166471277845e-05,
+      "loss": 1.2346,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8582649472450178e-05,
+      "loss": 1.2151,
+      "step": 456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8579132473622512e-05,
+      "loss": 1.2147,
+      "step": 457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8575615474794845e-05,
+      "loss": 1.239,
+      "step": 458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8572098475967175e-05,
+      "loss": 1.2722,
+      "step": 459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856858147713951e-05,
+      "loss": 1.258,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8565064478311842e-05,
+      "loss": 1.1988,
+      "step": 461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8561547479484176e-05,
+      "loss": 1.2802,
+      "step": 462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.855803048065651e-05,
+      "loss": 1.2147,
+      "step": 463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8554513481828843e-05,
+      "loss": 1.2312,
+      "step": 464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8550996483001173e-05,
+      "loss": 1.2385,
+      "step": 465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8547479484173507e-05,
+      "loss": 1.2842,
+      "step": 466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854396248534584e-05,
+      "loss": 1.2587,
+      "step": 467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854044548651817e-05,
+      "loss": 1.2441,
+      "step": 468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8536928487690504e-05,
+      "loss": 1.2699,
+      "step": 469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8533411488862837e-05,
+      "loss": 1.2533,
+      "step": 470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8529894490035174e-05,
+      "loss": 1.2665,
+      "step": 471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8526377491207504e-05,
+      "loss": 1.2384,
+      "step": 472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8522860492379838e-05,
+      "loss": 1.233,
+      "step": 473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851934349355217e-05,
+      "loss": 1.2217,
+      "step": 474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.85158264947245e-05,
+      "loss": 1.2426,
+      "step": 475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8512309495896835e-05,
+      "loss": 1.2631,
+      "step": 476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850879249706917e-05,
+      "loss": 1.2111,
+      "step": 477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8505275498241502e-05,
+      "loss": 1.245,
+      "step": 478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8501758499413836e-05,
+      "loss": 1.2829,
+      "step": 479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.849824150058617e-05,
+      "loss": 1.2734,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84947245017585e-05,
+      "loss": 1.2196,
+      "step": 481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8491207502930833e-05,
+      "loss": 1.222,
+      "step": 482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8487690504103166e-05,
+      "loss": 1.2464,
+      "step": 483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84841735052755e-05,
+      "loss": 1.211,
+      "step": 484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8480656506447833e-05,
+      "loss": 1.2223,
+      "step": 485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8477139507620167e-05,
+      "loss": 1.2354,
+      "step": 486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8473622508792497e-05,
+      "loss": 1.2661,
+      "step": 487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.847010550996483e-05,
+      "loss": 1.296,
+      "step": 488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8466588511137164e-05,
+      "loss": 1.2618,
+      "step": 489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8463071512309498e-05,
+      "loss": 1.2463,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.845955451348183e-05,
+      "loss": 1.2325,
+      "step": 491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8456037514654165e-05,
+      "loss": 1.296,
+      "step": 492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8452520515826498e-05,
+      "loss": 1.2115,
+      "step": 493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844900351699883e-05,
+      "loss": 1.2071,
+      "step": 494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8445486518171162e-05,
+      "loss": 1.208,
+      "step": 495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8441969519343495e-05,
+      "loss": 1.2644,
+      "step": 496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.843845252051583e-05,
+      "loss": 1.2288,
+      "step": 497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8434935521688162e-05,
+      "loss": 1.3065,
+      "step": 498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8431418522860496e-05,
+      "loss": 1.2427,
+      "step": 499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8427901524032826e-05,
+      "loss": 1.2271,
+      "step": 500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842438452520516e-05,
+      "loss": 1.2964,
+      "step": 501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8420867526377493e-05,
+      "loss": 1.2289,
+      "step": 502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8417350527549823e-05,
+      "loss": 1.2448,
+      "step": 503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8413833528722157e-05,
+      "loss": 1.2549,
+      "step": 504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.841031652989449e-05,
+      "loss": 1.2112,
+      "step": 505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8406799531066824e-05,
+      "loss": 1.2396,
+      "step": 506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8403282532239157e-05,
+      "loss": 1.2185,
+      "step": 507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.839976553341149e-05,
+      "loss": 1.275,
+      "step": 508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8396248534583824e-05,
+      "loss": 1.2335,
+      "step": 509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8392731535756155e-05,
+      "loss": 1.2339,
+      "step": 510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8389214536928488e-05,
+      "loss": 1.2888,
+      "step": 511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.838569753810082e-05,
+      "loss": 1.2569,
+      "step": 512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8382180539273155e-05,
+      "loss": 1.241,
+      "step": 513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.837866354044549e-05,
+      "loss": 1.205,
+      "step": 514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8375146541617822e-05,
+      "loss": 1.2415,
+      "step": 515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8371629542790152e-05,
+      "loss": 1.2244,
+      "step": 516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8368112543962486e-05,
+      "loss": 1.2431,
+      "step": 517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.836459554513482e-05,
+      "loss": 1.2255,
+      "step": 518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8361078546307153e-05,
+      "loss": 1.2609,
+      "step": 519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8357561547479486e-05,
+      "loss": 1.2486,
+      "step": 520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.835404454865182e-05,
+      "loss": 1.221,
+      "step": 521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.835052754982415e-05,
+      "loss": 1.2345,
+      "step": 522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8347010550996484e-05,
+      "loss": 1.2269,
+      "step": 523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8343493552168817e-05,
+      "loss": 1.315,
+      "step": 524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.833997655334115e-05,
+      "loss": 1.2436,
+      "step": 525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8336459554513484e-05,
+      "loss": 1.2165,
+      "step": 526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8332942555685818e-05,
+      "loss": 1.1952,
+      "step": 527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.832942555685815e-05,
+      "loss": 1.2762,
+      "step": 528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.832590855803048e-05,
+      "loss": 1.2362,
+      "step": 529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322391559202815e-05,
+      "loss": 1.2126,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831887456037515e-05,
+      "loss": 1.2915,
+      "step": 531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315357561547482e-05,
+      "loss": 1.2324,
+      "step": 532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311840562719816e-05,
+      "loss": 1.2356,
+      "step": 533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830832356389215e-05,
+      "loss": 1.29,
+      "step": 534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830480656506448e-05,
+      "loss": 1.2675,
+      "step": 535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301289566236813e-05,
+      "loss": 1.2509,
+      "step": 536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297772567409146e-05,
+      "loss": 1.1942,
+      "step": 537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294255568581476e-05,
+      "loss": 1.2482,
+      "step": 538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829073856975381e-05,
+      "loss": 1.222,
+      "step": 539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287221570926143e-05,
+      "loss": 1.1825,
+      "step": 540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283704572098477e-05,
+      "loss": 1.2132,
+      "step": 541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828018757327081e-05,
+      "loss": 1.2065,
+      "step": 542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276670574443144e-05,
+      "loss": 1.2542,
+      "step": 543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273153575615477e-05,
+      "loss": 1.2368,
+      "step": 544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269636576787808e-05,
+      "loss": 1.2427,
+      "step": 545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826611957796014e-05,
+      "loss": 1.2293,
+      "step": 546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262602579132475e-05,
+      "loss": 1.2208,
+      "step": 547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259085580304808e-05,
+      "loss": 1.2099,
+      "step": 548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8255568581477142e-05,
+      "loss": 1.2062,
+      "step": 549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8252051582649475e-05,
+      "loss": 1.2628,
+      "step": 550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8248534583821805e-05,
+      "loss": 1.2173,
+      "step": 551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.824501758499414e-05,
+      "loss": 1.2376,
+      "step": 552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8241500586166472e-05,
+      "loss": 1.2251,
+      "step": 553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8237983587338806e-05,
+      "loss": 1.2173,
+      "step": 554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.823446658851114e-05,
+      "loss": 1.2122,
+      "step": 555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230949589683473e-05,
+      "loss": 1.2551,
+      "step": 556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227432590855803e-05,
+      "loss": 1.1602,
+      "step": 557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223915592028137e-05,
+      "loss": 1.2016,
+      "step": 558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822039859320047e-05,
+      "loss": 1.2688,
+      "step": 559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8216881594372804e-05,
+      "loss": 1.2415,
+      "step": 560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213364595545137e-05,
+      "loss": 1.2547,
+      "step": 561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820984759671747e-05,
+      "loss": 1.1908,
+      "step": 562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206330597889804e-05,
+      "loss": 1.2692,
+      "step": 563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202813599062134e-05,
+      "loss": 1.1732,
+      "step": 564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199296600234468e-05,
+      "loss": 1.1626,
+      "step": 565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81957796014068e-05,
+      "loss": 1.2377,
+      "step": 566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192262602579135e-05,
+      "loss": 1.2539,
+      "step": 567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818874560375147e-05,
+      "loss": 1.2281,
+      "step": 568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185228604923802e-05,
+      "loss": 1.2752,
+      "step": 569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181711606096132e-05,
+      "loss": 1.2484,
+      "step": 570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178194607268466e-05,
+      "loss": 1.2656,
+      "step": 571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81746776084408e-05,
+      "loss": 1.2171,
+      "step": 572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817116060961313e-05,
+      "loss": 1.2076,
+      "step": 573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167643610785463e-05,
+      "loss": 1.1949,
+      "step": 574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8164126611957796e-05,
+      "loss": 1.2109,
+      "step": 575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816060961313013e-05,
+      "loss": 1.2777,
+      "step": 576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157092614302463e-05,
+      "loss": 1.257,
+      "step": 577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153575615474797e-05,
+      "loss": 1.2477,
+      "step": 578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815005861664713e-05,
+      "loss": 1.3005,
+      "step": 579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814654161781946e-05,
+      "loss": 1.2617,
+      "step": 580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143024618991794e-05,
+      "loss": 1.192,
+      "step": 581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8139507620164128e-05,
+      "loss": 1.2363,
+      "step": 582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.813599062133646e-05,
+      "loss": 1.2189,
+      "step": 583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8132473622508795e-05,
+      "loss": 1.2423,
+      "step": 584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8128956623681128e-05,
+      "loss": 1.2451,
+      "step": 585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.812543962485346e-05,
+      "loss": 1.2502,
+      "step": 586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8121922626025792e-05,
+      "loss": 1.2545,
+      "step": 587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8118405627198125e-05,
+      "loss": 1.2605,
+      "step": 588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.811488862837046e-05,
+      "loss": 1.2476,
+      "step": 589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8111371629542793e-05,
+      "loss": 1.2166,
+      "step": 590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8107854630715126e-05,
+      "loss": 1.2685,
+      "step": 591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8104337631887456e-05,
+      "loss": 1.2278,
+      "step": 592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.810082063305979e-05,
+      "loss": 1.2384,
+      "step": 593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8097303634232123e-05,
+      "loss": 1.2454,
+      "step": 594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8093786635404457e-05,
+      "loss": 1.2411,
+      "step": 595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.809026963657679e-05,
+      "loss": 1.2019,
+      "step": 596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8086752637749124e-05,
+      "loss": 1.2121,
+      "step": 597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8083235638921457e-05,
+      "loss": 1.2337,
+      "step": 598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8079718640093787e-05,
+      "loss": 1.2402,
+      "step": 599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807620164126612e-05,
+      "loss": 1.2291,
+      "step": 600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8072684642438455e-05,
+      "loss": 1.192,
+      "step": 601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069167643610788e-05,
+      "loss": 1.2837,
+      "step": 602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806565064478312e-05,
+      "loss": 1.2488,
+      "step": 603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062133645955455e-05,
+      "loss": 1.2531,
+      "step": 604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058616647127785e-05,
+      "loss": 1.2352,
+      "step": 605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805509964830012e-05,
+      "loss": 1.2934,
+      "step": 606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051582649472452e-05,
+      "loss": 1.2085,
+      "step": 607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8048065650644782e-05,
+      "loss": 1.1819,
+      "step": 608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8044548651817116e-05,
+      "loss": 1.2495,
+      "step": 609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.804103165298945e-05,
+      "loss": 1.2448,
+      "step": 610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8037514654161783e-05,
+      "loss": 1.2192,
+      "step": 611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8033997655334116e-05,
+      "loss": 1.1966,
+      "step": 612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.803048065650645e-05,
+      "loss": 1.2223,
+      "step": 613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8026963657678784e-05,
+      "loss": 1.268,
+      "step": 614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8023446658851114e-05,
+      "loss": 1.2605,
+      "step": 615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8019929660023447e-05,
+      "loss": 1.2399,
+      "step": 616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.801641266119578e-05,
+      "loss": 1.2502,
+      "step": 617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8012895662368114e-05,
+      "loss": 1.2445,
+      "step": 618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8009378663540448e-05,
+      "loss": 1.2567,
+      "step": 619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.800586166471278e-05,
+      "loss": 1.2741,
+      "step": 620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.800234466588511e-05,
+      "loss": 1.1864,
+      "step": 621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7998827667057445e-05,
+      "loss": 1.2727,
+      "step": 622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.799531066822978e-05,
+      "loss": 1.2955,
+      "step": 623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7991793669402112e-05,
+      "loss": 1.2457,
+      "step": 624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7988276670574446e-05,
+      "loss": 1.2304,
+      "step": 625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.798475967174678e-05,
+      "loss": 1.1873,
+      "step": 626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.798124267291911e-05,
+      "loss": 1.2198,
+      "step": 627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7977725674091443e-05,
+      "loss": 1.2332,
+      "step": 628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7974208675263776e-05,
+      "loss": 1.2748,
+      "step": 629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.797069167643611e-05,
+      "loss": 1.2596,
+      "step": 630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7967174677608443e-05,
+      "loss": 1.1932,
+      "step": 631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7963657678780777e-05,
+      "loss": 1.2337,
+      "step": 632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.796014067995311e-05,
+      "loss": 1.2026,
+      "step": 633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.795662368112544e-05,
+      "loss": 1.2139,
+      "step": 634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7953106682297774e-05,
+      "loss": 1.212,
+      "step": 635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7949589683470108e-05,
+      "loss": 1.211,
+      "step": 636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.794607268464244e-05,
+      "loss": 1.1995,
+      "step": 637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7942555685814775e-05,
+      "loss": 1.2814,
+      "step": 638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7939038686987108e-05,
+      "loss": 1.2527,
+      "step": 639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7935521688159438e-05,
+      "loss": 1.2638,
+      "step": 640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7932004689331772e-05,
+      "loss": 1.2279,
+      "step": 641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7928487690504105e-05,
+      "loss": 1.2442,
+      "step": 642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7924970691676435e-05,
+      "loss": 1.2455,
+      "step": 643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.792145369284877e-05,
+      "loss": 1.2153,
+      "step": 644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7917936694021102e-05,
+      "loss": 1.2211,
+      "step": 645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7914419695193436e-05,
+      "loss": 1.2745,
+      "step": 646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.791090269636577e-05,
+      "loss": 1.2477,
+      "step": 647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7907385697538103e-05,
+      "loss": 1.2115,
+      "step": 648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7903868698710437e-05,
+      "loss": 1.2344,
+      "step": 649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7900351699882767e-05,
+      "loss": 1.2249,
+      "step": 650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.78968347010551e-05,
+      "loss": 1.2709,
+      "step": 651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7893317702227434e-05,
+      "loss": 1.2584,
+      "step": 652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7889800703399767e-05,
+      "loss": 1.2109,
+      "step": 653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.78862837045721e-05,
+      "loss": 1.2237,
+      "step": 654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7882766705744434e-05,
+      "loss": 1.2189,
+      "step": 655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7879249706916764e-05,
+      "loss": 1.2217,
+      "step": 656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7875732708089098e-05,
+      "loss": 1.1956,
+      "step": 657
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.787221570926143e-05,
+      "loss": 1.2118,
+      "step": 658
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7868698710433765e-05,
+      "loss": 1.1965,
+      "step": 659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.78651817116061e-05,
+      "loss": 1.2348,
+      "step": 660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7861664712778432e-05,
+      "loss": 1.2818,
+      "step": 661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7858147713950762e-05,
+      "loss": 1.211,
+      "step": 662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7854630715123096e-05,
+      "loss": 1.186,
+      "step": 663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.785111371629543e-05,
+      "loss": 1.2011,
+      "step": 664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7847596717467763e-05,
+      "loss": 1.2441,
+      "step": 665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7844079718640096e-05,
+      "loss": 1.2067,
+      "step": 666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.784056271981243e-05,
+      "loss": 1.2504,
+      "step": 667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7837045720984763e-05,
+      "loss": 1.2821,
+      "step": 668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7833528722157094e-05,
+      "loss": 1.207,
+      "step": 669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7830011723329427e-05,
+      "loss": 1.2287,
+      "step": 670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.782649472450176e-05,
+      "loss": 1.2221,
+      "step": 671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7822977725674094e-05,
+      "loss": 1.2432,
+      "step": 672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7819460726846428e-05,
+      "loss": 1.2513,
+      "step": 673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.781594372801876e-05,
+      "loss": 1.2423,
+      "step": 674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.781242672919109e-05,
+      "loss": 1.2265,
+      "step": 675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7808909730363425e-05,
+      "loss": 1.1916,
+      "step": 676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.780539273153576e-05,
+      "loss": 1.2143,
+      "step": 677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.780187573270809e-05,
+      "loss": 1.2345,
+      "step": 678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7798358733880422e-05,
+      "loss": 1.2405,
+      "step": 679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7794841735052755e-05,
+      "loss": 1.24,
+      "step": 680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.779132473622509e-05,
+      "loss": 1.2473,
+      "step": 681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7787807737397423e-05,
+      "loss": 1.2167,
+      "step": 682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7784290738569756e-05,
+      "loss": 1.2578,
+      "step": 683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.778077373974209e-05,
+      "loss": 1.1861,
+      "step": 684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.777725674091442e-05,
+      "loss": 1.2148,
+      "step": 685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7773739742086753e-05,
+      "loss": 1.2427,
+      "step": 686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7770222743259087e-05,
+      "loss": 1.2382,
+      "step": 687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.776670574443142e-05,
+      "loss": 1.2037,
+      "step": 688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7763188745603754e-05,
+      "loss": 1.2089,
+      "step": 689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7759671746776087e-05,
+      "loss": 1.1841,
+      "step": 690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7756154747948417e-05,
+      "loss": 1.1952,
+      "step": 691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.775263774912075e-05,
+      "loss": 1.2935,
+      "step": 692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7749120750293085e-05,
+      "loss": 1.2569,
+      "step": 693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7745603751465418e-05,
+      "loss": 1.2097,
+      "step": 694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.774208675263775e-05,
+      "loss": 1.2585,
+      "step": 695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7738569753810085e-05,
+      "loss": 1.2303,
+      "step": 696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7735052754982415e-05,
+      "loss": 1.2454,
+      "step": 697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.773153575615475e-05,
+      "loss": 1.2296,
+      "step": 698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7728018757327082e-05,
+      "loss": 1.2827,
+      "step": 699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7724501758499416e-05,
+      "loss": 1.2717,
+      "step": 700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.772098475967175e-05,
+      "loss": 1.2351,
+      "step": 701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7717467760844083e-05,
+      "loss": 1.2394,
+      "step": 702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7713950762016416e-05,
+      "loss": 1.2127,
+      "step": 703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7710433763188747e-05,
+      "loss": 1.2397,
+      "step": 704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.770691676436108e-05,
+      "loss": 1.2192,
+      "step": 705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7703399765533414e-05,
+      "loss": 1.1646,
+      "step": 706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7699882766705747e-05,
+      "loss": 1.2378,
+      "step": 707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.769636576787808e-05,
+      "loss": 1.2601,
+      "step": 708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7692848769050414e-05,
+      "loss": 1.2738,
+      "step": 709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7689331770222744e-05,
+      "loss": 1.1934,
+      "step": 710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7685814771395078e-05,
+      "loss": 1.2263,
+      "step": 711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.768229777256741e-05,
+      "loss": 1.2583,
+      "step": 712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.767878077373974e-05,
+      "loss": 1.2254,
+      "step": 713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7675263774912075e-05,
+      "loss": 1.2206,
+      "step": 714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.767174677608441e-05,
+      "loss": 1.2699,
+      "step": 715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7668229777256742e-05,
+      "loss": 1.2783,
+      "step": 716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7664712778429076e-05,
+      "loss": 1.2326,
+      "step": 717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.766119577960141e-05,
+      "loss": 1.2422,
+      "step": 718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.765767878077374e-05,
+      "loss": 1.2516,
+      "step": 719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7654161781946073e-05,
+      "loss": 1.2351,
+      "step": 720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7650644783118406e-05,
+      "loss": 1.2112,
+      "step": 721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.764712778429074e-05,
+      "loss": 1.2395,
+      "step": 722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7643610785463073e-05,
+      "loss": 1.2236,
+      "step": 723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7640093786635407e-05,
+      "loss": 1.2018,
+      "step": 724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.763657678780774e-05,
+      "loss": 1.2212,
+      "step": 725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.763305978898007e-05,
+      "loss": 1.2594,
+      "step": 726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7629542790152404e-05,
+      "loss": 1.2411,
+      "step": 727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7626025791324738e-05,
+      "loss": 1.2374,
+      "step": 728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.762250879249707e-05,
+      "loss": 1.2619,
+      "step": 729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7618991793669405e-05,
+      "loss": 1.2418,
+      "step": 730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7615474794841738e-05,
+      "loss": 1.1731,
+      "step": 731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7611957796014068e-05,
+      "loss": 1.2345,
+      "step": 732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7608440797186402e-05,
+      "loss": 1.2609,
+      "step": 733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7604923798358735e-05,
+      "loss": 1.2324,
+      "step": 734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.760140679953107e-05,
+      "loss": 1.2019,
+      "step": 735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7597889800703402e-05,
+      "loss": 1.2101,
+      "step": 736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7594372801875736e-05,
+      "loss": 1.2622,
+      "step": 737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.759085580304807e-05,
+      "loss": 1.2468,
+      "step": 738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.75873388042204e-05,
+      "loss": 1.2087,
+      "step": 739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7583821805392733e-05,
+      "loss": 1.1822,
+      "step": 740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7580304806565067e-05,
+      "loss": 1.2544,
+      "step": 741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.75767878077374e-05,
+      "loss": 1.2456,
+      "step": 742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7573270808909734e-05,
+      "loss": 1.2101,
+      "step": 743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7569753810082067e-05,
+      "loss": 1.2187,
+      "step": 744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7566236811254397e-05,
+      "loss": 1.1712,
+      "step": 745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.756271981242673e-05,
+      "loss": 1.2286,
+      "step": 746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7559202813599064e-05,
+      "loss": 1.2493,
+      "step": 747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7555685814771394e-05,
+      "loss": 1.2653,
+      "step": 748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7552168815943728e-05,
+      "loss": 1.2754,
+      "step": 749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.754865181711606e-05,
+      "loss": 1.2786,
+      "step": 750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7545134818288395e-05,
+      "loss": 1.2543,
+      "step": 751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.754161781946073e-05,
+      "loss": 1.255,
+      "step": 752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7538100820633062e-05,
+      "loss": 1.2517,
+      "step": 753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7534583821805392e-05,
+      "loss": 1.2621,
+      "step": 754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7531066822977726e-05,
+      "loss": 1.2596,
+      "step": 755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.752754982415006e-05,
+      "loss": 1.2099,
+      "step": 756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7524032825322393e-05,
+      "loss": 1.2378,
+      "step": 757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7520515826494726e-05,
+      "loss": 1.2029,
+      "step": 758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.751699882766706e-05,
+      "loss": 1.2686,
+      "step": 759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7513481828839393e-05,
+      "loss": 1.1735,
+      "step": 760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7509964830011724e-05,
+      "loss": 1.2531,
+      "step": 761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7506447831184057e-05,
+      "loss": 1.2284,
+      "step": 762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.750293083235639e-05,
+      "loss": 1.2207,
+      "step": 763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7499413833528724e-05,
+      "loss": 1.2462,
+      "step": 764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7495896834701058e-05,
+      "loss": 1.2542,
+      "step": 765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.749237983587339e-05,
+      "loss": 1.1681,
+      "step": 766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.748886283704572e-05,
+      "loss": 1.2399,
+      "step": 767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7485345838218055e-05,
+      "loss": 1.2274,
+      "step": 768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.748182883939039e-05,
+      "loss": 1.1959,
+      "step": 769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7478311840562722e-05,
+      "loss": 1.2274,
+      "step": 770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7474794841735055e-05,
+      "loss": 1.2078,
+      "step": 771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.747127784290739e-05,
+      "loss": 1.1842,
+      "step": 772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7467760844079722e-05,
+      "loss": 1.1887,
+      "step": 773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7464243845252053e-05,
+      "loss": 1.2633,
+      "step": 774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7460726846424386e-05,
+      "loss": 1.2388,
+      "step": 775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.745720984759672e-05,
+      "loss": 1.2259,
+      "step": 776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7453692848769053e-05,
+      "loss": 1.2455,
+      "step": 777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7450175849941387e-05,
+      "loss": 1.2247,
+      "step": 778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.744665885111372e-05,
+      "loss": 1.2343,
+      "step": 779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.744314185228605e-05,
+      "loss": 1.1888,
+      "step": 780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7439624853458384e-05,
+      "loss": 1.2337,
+      "step": 781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7436107854630717e-05,
+      "loss": 1.2296,
+      "step": 782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7432590855803048e-05,
+      "loss": 1.1954,
+      "step": 783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.742907385697538e-05,
+      "loss": 1.2417,
+      "step": 784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7425556858147715e-05,
+      "loss": 1.2881,
+      "step": 785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7422039859320048e-05,
+      "loss": 1.2129,
+      "step": 786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.741852286049238e-05,
+      "loss": 1.2015,
+      "step": 787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7415005861664715e-05,
+      "loss": 1.1959,
+      "step": 788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7411488862837045e-05,
+      "loss": 1.2211,
+      "step": 789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.740797186400938e-05,
+      "loss": 1.2806,
+      "step": 790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7404454865181712e-05,
+      "loss": 1.1938,
+      "step": 791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7400937866354046e-05,
+      "loss": 1.2055,
+      "step": 792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.739742086752638e-05,
+      "loss": 1.2078,
+      "step": 793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7393903868698713e-05,
+      "loss": 1.2384,
+      "step": 794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7390386869871046e-05,
+      "loss": 1.2032,
+      "step": 795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7386869871043377e-05,
+      "loss": 1.2282,
+      "step": 796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.738335287221571e-05,
+      "loss": 1.2322,
+      "step": 797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7379835873388044e-05,
+      "loss": 1.2009,
+      "step": 798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7376318874560377e-05,
+      "loss": 1.2523,
+      "step": 799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.737280187573271e-05,
+      "loss": 1.2445,
+      "step": 800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7369284876905044e-05,
+      "loss": 1.2236,
+      "step": 801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7365767878077374e-05,
+      "loss": 1.2332,
+      "step": 802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7362250879249708e-05,
+      "loss": 1.232,
+      "step": 803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.735873388042204e-05,
+      "loss": 1.2675,
+      "step": 804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7355216881594375e-05,
+      "loss": 1.2252,
+      "step": 805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.735169988276671e-05,
+      "loss": 1.2194,
+      "step": 806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7348182883939042e-05,
+      "loss": 1.2355,
+      "step": 807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7344665885111372e-05,
+      "loss": 1.2025,
+      "step": 808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7341148886283706e-05,
+      "loss": 1.2816,
+      "step": 809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.733763188745604e-05,
+      "loss": 1.1988,
+      "step": 810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7334114888628373e-05,
+      "loss": 1.2066,
+      "step": 811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7330597889800706e-05,
+      "loss": 1.1953,
+      "step": 812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.732708089097304e-05,
+      "loss": 1.2049,
+      "step": 813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7323563892145373e-05,
+      "loss": 1.2298,
+      "step": 814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7320046893317703e-05,
+      "loss": 1.1559,
+      "step": 815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7316529894490037e-05,
+      "loss": 1.1855,
+      "step": 816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.731301289566237e-05,
+      "loss": 1.2667,
+      "step": 817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.73094958968347e-05,
+      "loss": 1.2067,
+      "step": 818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7305978898007034e-05,
+      "loss": 1.2441,
+      "step": 819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7302461899179368e-05,
+      "loss": 1.2182,
+      "step": 820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.72989449003517e-05,
+      "loss": 1.241,
+      "step": 821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7295427901524035e-05,
+      "loss": 1.214,
+      "step": 822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7291910902696368e-05,
+      "loss": 1.2641,
+      "step": 823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7288393903868698e-05,
+      "loss": 1.2176,
+      "step": 824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7284876905041032e-05,
+      "loss": 1.2682,
+      "step": 825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7281359906213365e-05,
+      "loss": 1.2761,
+      "step": 826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.72778429073857e-05,
+      "loss": 1.2357,
+      "step": 827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7274325908558032e-05,
+      "loss": 1.1662,
+      "step": 828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7270808909730366e-05,
+      "loss": 1.2373,
+      "step": 829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.72672919109027e-05,
+      "loss": 1.2073,
+      "step": 830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.726377491207503e-05,
+      "loss": 1.2382,
+      "step": 831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7260257913247363e-05,
+      "loss": 1.2274,
+      "step": 832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7256740914419697e-05,
+      "loss": 1.2585,
+      "step": 833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7253223915592027e-05,
+      "loss": 1.1934,
+      "step": 834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.724970691676436e-05,
+      "loss": 1.2315,
+      "step": 835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7246189917936694e-05,
+      "loss": 1.2209,
+      "step": 836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7242672919109027e-05,
+      "loss": 1.2044,
+      "step": 837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.723915592028136e-05,
+      "loss": 1.187,
+      "step": 838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7235638921453694e-05,
+      "loss": 1.177,
+      "step": 839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7232121922626025e-05,
+      "loss": 1.2788,
+      "step": 840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7228604923798358e-05,
+      "loss": 1.2469,
+      "step": 841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.722508792497069e-05,
+      "loss": 1.2583,
+      "step": 842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7221570926143025e-05,
+      "loss": 1.2793,
+      "step": 843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.721805392731536e-05,
+      "loss": 1.2584,
+      "step": 844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7214536928487692e-05,
+      "loss": 1.2586,
+      "step": 845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7211019929660026e-05,
+      "loss": 1.2582,
+      "step": 846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7207502930832356e-05,
+      "loss": 1.2139,
+      "step": 847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.720398593200469e-05,
+      "loss": 1.2106,
+      "step": 848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7200468933177023e-05,
+      "loss": 1.2584,
+      "step": 849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7196951934349356e-05,
+      "loss": 1.2691,
+      "step": 850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.719343493552169e-05,
+      "loss": 1.2351,
+      "step": 851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7189917936694023e-05,
+      "loss": 1.2683,
+      "step": 852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7186400937866354e-05,
+      "loss": 1.2158,
+      "step": 853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7182883939038687e-05,
+      "loss": 1.2656,
+      "step": 854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.717936694021102e-05,
+      "loss": 1.1984,
+      "step": 855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7175849941383354e-05,
+      "loss": 1.2199,
+      "step": 856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7172332942555688e-05,
+      "loss": 1.2354,
+      "step": 857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.716881594372802e-05,
+      "loss": 1.1896,
+      "step": 858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.716529894490035e-05,
+      "loss": 1.2445,
+      "step": 859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7161781946072685e-05,
+      "loss": 1.2579,
+      "step": 860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.715826494724502e-05,
+      "loss": 1.2018,
+      "step": 861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7154747948417352e-05,
+      "loss": 1.2197,
+      "step": 862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7151230949589685e-05,
+      "loss": 1.2314,
+      "step": 863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.714771395076202e-05,
+      "loss": 1.2651,
+      "step": 864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7144196951934352e-05,
+      "loss": 1.2571,
+      "step": 865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7140679953106683e-05,
+      "loss": 1.2455,
+      "step": 866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7137162954279016e-05,
+      "loss": 1.2112,
+      "step": 867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.713364595545135e-05,
+      "loss": 1.2315,
+      "step": 868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.713012895662368e-05,
+      "loss": 1.2485,
+      "step": 869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7126611957796013e-05,
+      "loss": 1.2644,
+      "step": 870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7123094958968347e-05,
+      "loss": 1.2575,
+      "step": 871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.711957796014068e-05,
+      "loss": 1.2346,
+      "step": 872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7116060961313014e-05,
+      "loss": 1.2422,
+      "step": 873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7112543962485347e-05,
+      "loss": 1.3049,
+      "step": 874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7109026963657678e-05,
+      "loss": 1.242,
+      "step": 875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.710550996483001e-05,
+      "loss": 1.2649,
+      "step": 876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7101992966002345e-05,
+      "loss": 1.1927,
+      "step": 877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7098475967174678e-05,
+      "loss": 1.2343,
+      "step": 878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.709495896834701e-05,
+      "loss": 1.2425,
+      "step": 879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7091441969519345e-05,
+      "loss": 1.2489,
+      "step": 880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.708792497069168e-05,
+      "loss": 1.2252,
+      "step": 881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.708440797186401e-05,
+      "loss": 1.2066,
+      "step": 882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7080890973036342e-05,
+      "loss": 1.2299,
+      "step": 883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7077373974208676e-05,
+      "loss": 1.1831,
+      "step": 884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.707385697538101e-05,
+      "loss": 1.1892,
+      "step": 885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7070339976553343e-05,
+      "loss": 1.193,
+      "step": 886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7066822977725676e-05,
+      "loss": 1.2375,
+      "step": 887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7063305978898007e-05,
+      "loss": 1.2058,
+      "step": 888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.705978898007034e-05,
+      "loss": 1.2276,
+      "step": 889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7056271981242674e-05,
+      "loss": 1.233,
+      "step": 890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7052754982415007e-05,
+      "loss": 1.2335,
+      "step": 891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.704923798358734e-05,
+      "loss": 1.2326,
+      "step": 892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7045720984759674e-05,
+      "loss": 1.2496,
+      "step": 893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7042203985932004e-05,
+      "loss": 1.2282,
+      "step": 894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7038686987104338e-05,
+      "loss": 1.2639,
+      "step": 895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.703516998827667e-05,
+      "loss": 1.2308,
+      "step": 896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7031652989449005e-05,
+      "loss": 1.2324,
+      "step": 897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.702813599062134e-05,
+      "loss": 1.254,
+      "step": 898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7024618991793672e-05,
+      "loss": 1.1891,
+      "step": 899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7021101992966005e-05,
+      "loss": 1.2352,
+      "step": 900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7017584994138336e-05,
+      "loss": 1.2331,
+      "step": 901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.701406799531067e-05,
+      "loss": 1.2237,
+      "step": 902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7010550996483003e-05,
+      "loss": 1.226,
+      "step": 903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7007033997655333e-05,
+      "loss": 1.2707,
+      "step": 904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7003516998827666e-05,
+      "loss": 1.2357,
+      "step": 905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7e-05,
+      "loss": 1.2435,
+      "step": 906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6996483001172333e-05,
+      "loss": 1.2302,
+      "step": 907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6992966002344667e-05,
+      "loss": 1.2818,
+      "step": 908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6989449003517e-05,
+      "loss": 1.2869,
+      "step": 909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.698593200468933e-05,
+      "loss": 1.2461,
+      "step": 910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6982415005861664e-05,
+      "loss": 1.2377,
+      "step": 911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6978898007033998e-05,
+      "loss": 1.2034,
+      "step": 912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.697538100820633e-05,
+      "loss": 1.2708,
+      "step": 913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6971864009378665e-05,
+      "loss": 1.234,
+      "step": 914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6968347010550998e-05,
+      "loss": 1.2279,
+      "step": 915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6964830011723332e-05,
+      "loss": 1.2383,
+      "step": 916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6961313012895662e-05,
+      "loss": 1.18,
+      "step": 917
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6957796014067995e-05,
+      "loss": 1.2041,
+      "step": 918
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.695427901524033e-05,
+      "loss": 1.2745,
+      "step": 919
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6950762016412662e-05,
+      "loss": 1.2055,
+      "step": 920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6947245017584996e-05,
+      "loss": 1.2346,
+      "step": 921
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.694372801875733e-05,
+      "loss": 1.221,
+      "step": 922
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.694021101992966e-05,
+      "loss": 1.2078,
+      "step": 923
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6936694021101993e-05,
+      "loss": 1.2249,
+      "step": 924
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6933177022274327e-05,
+      "loss": 1.2538,
+      "step": 925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692966002344666e-05,
+      "loss": 1.224,
+      "step": 926
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6926143024618994e-05,
+      "loss": 1.1912,
+      "step": 927
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6922626025791327e-05,
+      "loss": 1.2678,
+      "step": 928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6919109026963657e-05,
+      "loss": 1.2229,
+      "step": 929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.691559202813599e-05,
+      "loss": 1.2504,
+      "step": 930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6912075029308324e-05,
+      "loss": 1.1941,
+      "step": 931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6908558030480655e-05,
+      "loss": 1.2699,
+      "step": 932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.690504103165299e-05,
+      "loss": 1.1976,
+      "step": 933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6901524032825325e-05,
+      "loss": 1.2477,
+      "step": 934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689800703399766e-05,
+      "loss": 1.2239,
+      "step": 935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689449003516999e-05,
+      "loss": 1.2252,
+      "step": 936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6890973036342322e-05,
+      "loss": 1.2528,
+      "step": 937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6887456037514656e-05,
+      "loss": 1.2248,
+      "step": 938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6883939038686986e-05,
+      "loss": 1.1717,
+      "step": 939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.688042203985932e-05,
+      "loss": 1.2232,
+      "step": 940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6876905041031653e-05,
+      "loss": 1.2437,
+      "step": 941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6873388042203986e-05,
+      "loss": 1.2202,
+      "step": 942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.686987104337632e-05,
+      "loss": 1.2124,
+      "step": 943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6866354044548653e-05,
+      "loss": 1.1595,
+      "step": 944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6862837045720984e-05,
+      "loss": 1.2464,
+      "step": 945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6859320046893317e-05,
+      "loss": 1.2345,
+      "step": 946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.685580304806565e-05,
+      "loss": 1.2488,
+      "step": 947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6852286049237984e-05,
+      "loss": 1.198,
+      "step": 948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6848769050410318e-05,
+      "loss": 1.2446,
+      "step": 949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.684525205158265e-05,
+      "loss": 1.1555,
+      "step": 950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6841735052754985e-05,
+      "loss": 1.2188,
+      "step": 951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6838218053927315e-05,
+      "loss": 1.2191,
+      "step": 952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.683470105509965e-05,
+      "loss": 1.258,
+      "step": 953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6831184056271982e-05,
+      "loss": 1.2403,
+      "step": 954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6827667057444315e-05,
+      "loss": 1.1876,
+      "step": 955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.682415005861665e-05,
+      "loss": 1.2353,
+      "step": 956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6820633059788983e-05,
+      "loss": 1.24,
+      "step": 957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6817116060961313e-05,
+      "loss": 1.22,
+      "step": 958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6813599062133646e-05,
+      "loss": 1.1702,
+      "step": 959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.681008206330598e-05,
+      "loss": 1.2506,
+      "step": 960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6806565064478313e-05,
+      "loss": 1.2703,
+      "step": 961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6803048065650647e-05,
+      "loss": 1.2122,
+      "step": 962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.679953106682298e-05,
+      "loss": 1.2401,
+      "step": 963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.679601406799531e-05,
+      "loss": 1.2534,
+      "step": 964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6792497069167644e-05,
+      "loss": 1.1836,
+      "step": 965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6788980070339977e-05,
+      "loss": 1.2292,
+      "step": 966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6785463071512308e-05,
+      "loss": 1.2939,
+      "step": 967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6781946072684644e-05,
+      "loss": 1.2228,
+      "step": 968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6778429073856978e-05,
+      "loss": 1.2332,
+      "step": 969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.677491207502931e-05,
+      "loss": 1.2071,
+      "step": 970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.677139507620164e-05,
+      "loss": 1.2094,
+      "step": 971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6767878077373975e-05,
+      "loss": 1.2221,
+      "step": 972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.676436107854631e-05,
+      "loss": 1.2475,
+      "step": 973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.676084407971864e-05,
+      "loss": 1.1795,
+      "step": 974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6757327080890972e-05,
+      "loss": 1.2568,
+      "step": 975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6753810082063306e-05,
+      "loss": 1.258,
+      "step": 976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.675029308323564e-05,
+      "loss": 1.2449,
+      "step": 977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6746776084407973e-05,
+      "loss": 1.2751,
+      "step": 978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6743259085580306e-05,
+      "loss": 1.2129,
+      "step": 979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6739742086752637e-05,
+      "loss": 1.2767,
+      "step": 980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.673622508792497e-05,
+      "loss": 1.2405,
+      "step": 981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6732708089097304e-05,
+      "loss": 1.2269,
+      "step": 982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6729191090269637e-05,
+      "loss": 1.2185,
+      "step": 983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.672567409144197e-05,
+      "loss": 1.213,
+      "step": 984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6722157092614304e-05,
+      "loss": 1.241,
+      "step": 985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6718640093786638e-05,
+      "loss": 1.192,
+      "step": 986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6715123094958968e-05,
+      "loss": 1.1963,
+      "step": 987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.67116060961313e-05,
+      "loss": 1.2643,
+      "step": 988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6708089097303635e-05,
+      "loss": 1.1732,
+      "step": 989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.670457209847597e-05,
+      "loss": 1.2412,
+      "step": 990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6701055099648302e-05,
+      "loss": 1.2445,
+      "step": 991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6697538100820636e-05,
+      "loss": 1.2235,
+      "step": 992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6694021101992966e-05,
+      "loss": 1.2661,
+      "step": 993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.66905041031653e-05,
+      "loss": 1.2322,
+      "step": 994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6686987104337633e-05,
+      "loss": 1.2718,
+      "step": 995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6683470105509966e-05,
+      "loss": 1.266,
+      "step": 996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.66799531066823e-05,
+      "loss": 1.2123,
+      "step": 997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6676436107854633e-05,
+      "loss": 1.2409,
+      "step": 998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6672919109026963e-05,
+      "loss": 1.2367,
+      "step": 999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6669402110199297e-05,
+      "loss": 1.1988,
+      "step": 1000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.666588511137163e-05,
+      "loss": 1.2741,
+      "step": 1001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.666236811254396e-05,
+      "loss": 1.2287,
+      "step": 1002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6658851113716298e-05,
+      "loss": 1.2512,
+      "step": 1003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.665533411488863e-05,
+      "loss": 1.2263,
+      "step": 1004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6651817116060965e-05,
+      "loss": 1.2249,
+      "step": 1005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6648300117233295e-05,
+      "loss": 1.2559,
+      "step": 1006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6644783118405628e-05,
+      "loss": 1.2055,
+      "step": 1007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6641266119577962e-05,
+      "loss": 1.262,
+      "step": 1008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6637749120750292e-05,
+      "loss": 1.2156,
+      "step": 1009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6634232121922625e-05,
+      "loss": 1.2881,
+      "step": 1010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.663071512309496e-05,
+      "loss": 1.2044,
+      "step": 1011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6627198124267292e-05,
+      "loss": 1.2553,
+      "step": 1012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6623681125439626e-05,
+      "loss": 1.1956,
+      "step": 1013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.662016412661196e-05,
+      "loss": 1.2455,
+      "step": 1014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.661664712778429e-05,
+      "loss": 1.2303,
+      "step": 1015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6613130128956623e-05,
+      "loss": 1.1659,
+      "step": 1016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6609613130128957e-05,
+      "loss": 1.2483,
+      "step": 1017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.660609613130129e-05,
+      "loss": 1.2604,
+      "step": 1018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6602579132473624e-05,
+      "loss": 1.1778,
+      "step": 1019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6599062133645957e-05,
+      "loss": 1.2196,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6595545134818287e-05,
+      "loss": 1.2368,
+      "step": 1021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.659202813599062e-05,
+      "loss": 1.2605,
+      "step": 1022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6588511137162954e-05,
+      "loss": 1.2068,
+      "step": 1023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6584994138335288e-05,
+      "loss": 1.2491,
+      "step": 1024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.658147713950762e-05,
+      "loss": 1.2746,
+      "step": 1025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6577960140679955e-05,
+      "loss": 1.244,
+      "step": 1026
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.657444314185229e-05,
+      "loss": 1.2401,
+      "step": 1027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.657092614302462e-05,
+      "loss": 1.2352,
+      "step": 1028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6567409144196952e-05,
+      "loss": 1.2439,
+      "step": 1029
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6563892145369286e-05,
+      "loss": 1.2281,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.656037514654162e-05,
+      "loss": 1.2215,
+      "step": 1031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6556858147713953e-05,
+      "loss": 1.198,
+      "step": 1032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6553341148886286e-05,
+      "loss": 1.2122,
+      "step": 1033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6549824150058616e-05,
+      "loss": 1.2016,
+      "step": 1034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.654630715123095e-05,
+      "loss": 1.232,
+      "step": 1035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6542790152403283e-05,
+      "loss": 1.2464,
+      "step": 1036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6539273153575614e-05,
+      "loss": 1.2095,
+      "step": 1037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6535756154747947e-05,
+      "loss": 1.2248,
+      "step": 1038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6532239155920284e-05,
+      "loss": 1.2665,
+      "step": 1039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6528722157092618e-05,
+      "loss": 1.2255,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6525205158264948e-05,
+      "loss": 1.2349,
+      "step": 1041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.652168815943728e-05,
+      "loss": 1.1908,
+      "step": 1042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6518171160609615e-05,
+      "loss": 1.2259,
+      "step": 1043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6514654161781945e-05,
+      "loss": 1.2231,
+      "step": 1044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.651113716295428e-05,
+      "loss": 1.2542,
+      "step": 1045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6507620164126612e-05,
+      "loss": 1.2049,
+      "step": 1046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6504103165298945e-05,
+      "loss": 1.2247,
+      "step": 1047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.650058616647128e-05,
+      "loss": 1.216,
+      "step": 1048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6497069167643613e-05,
+      "loss": 1.2658,
+      "step": 1049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6493552168815943e-05,
+      "loss": 1.2034,
+      "step": 1050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6490035169988276e-05,
+      "loss": 1.2629,
+      "step": 1051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.648651817116061e-05,
+      "loss": 1.2326,
+      "step": 1052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6483001172332943e-05,
+      "loss": 1.2205,
+      "step": 1053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6479484173505277e-05,
+      "loss": 1.2312,
+      "step": 1054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.647596717467761e-05,
+      "loss": 1.2595,
+      "step": 1055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.647245017584994e-05,
+      "loss": 1.2246,
+      "step": 1056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6468933177022274e-05,
+      "loss": 1.222,
+      "step": 1057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6465416178194607e-05,
+      "loss": 1.2485,
+      "step": 1058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.646189917936694e-05,
+      "loss": 1.2199,
+      "step": 1059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6458382180539275e-05,
+      "loss": 1.2202,
+      "step": 1060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6454865181711608e-05,
+      "loss": 1.2504,
+      "step": 1061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.645134818288394e-05,
+      "loss": 1.2358,
+      "step": 1062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6447831184056272e-05,
+      "loss": 1.2815,
+      "step": 1063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6444314185228605e-05,
+      "loss": 1.2333,
+      "step": 1064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.644079718640094e-05,
+      "loss": 1.2197,
+      "step": 1065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6437280187573272e-05,
+      "loss": 1.2426,
+      "step": 1066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6433763188745606e-05,
+      "loss": 1.1726,
+      "step": 1067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.643024618991794e-05,
+      "loss": 1.2121,
+      "step": 1068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.642672919109027e-05,
+      "loss": 1.2481,
+      "step": 1069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6423212192262603e-05,
+      "loss": 1.2059,
+      "step": 1070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6419695193434937e-05,
+      "loss": 1.2021,
+      "step": 1071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6416178194607267e-05,
+      "loss": 1.2403,
+      "step": 1072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.64126611957796e-05,
+      "loss": 1.2439,
+      "step": 1073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6409144196951937e-05,
+      "loss": 1.2033,
+      "step": 1074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.640562719812427e-05,
+      "loss": 1.2458,
+      "step": 1075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.64021101992966e-05,
+      "loss": 1.2772,
+      "step": 1076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6398593200468934e-05,
+      "loss": 1.245,
+      "step": 1077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6395076201641268e-05,
+      "loss": 1.2211,
+      "step": 1078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6391559202813598e-05,
+      "loss": 1.2307,
+      "step": 1079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.638804220398593e-05,
+      "loss": 1.2453,
+      "step": 1080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6384525205158265e-05,
+      "loss": 1.2306,
+      "step": 1081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.63810082063306e-05,
+      "loss": 1.2008,
+      "step": 1082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6377491207502932e-05,
+      "loss": 1.2114,
+      "step": 1083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6373974208675266e-05,
+      "loss": 1.2651,
+      "step": 1084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6370457209847596e-05,
+      "loss": 1.2106,
+      "step": 1085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.636694021101993e-05,
+      "loss": 1.1907,
+      "step": 1086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6363423212192263e-05,
+      "loss": 1.2141,
+      "step": 1087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6359906213364596e-05,
+      "loss": 1.2011,
+      "step": 1088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.635638921453693e-05,
+      "loss": 1.2121,
+      "step": 1089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6352872215709263e-05,
+      "loss": 1.2389,
+      "step": 1090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6349355216881593e-05,
+      "loss": 1.2557,
+      "step": 1091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6345838218053927e-05,
+      "loss": 1.2157,
+      "step": 1092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.634232121922626e-05,
+      "loss": 1.239,
+      "step": 1093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6338804220398594e-05,
+      "loss": 1.2122,
+      "step": 1094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6335287221570928e-05,
+      "loss": 1.1996,
+      "step": 1095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.633177022274326e-05,
+      "loss": 1.2346,
+      "step": 1096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6328253223915595e-05,
+      "loss": 1.2145,
+      "step": 1097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6324736225087925e-05,
+      "loss": 1.2174,
+      "step": 1098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6321219226260258e-05,
+      "loss": 1.2082,
+      "step": 1099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6317702227432592e-05,
+      "loss": 1.2199,
+      "step": 1100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6314185228604925e-05,
+      "loss": 1.2299,
+      "step": 1101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.631066822977726e-05,
+      "loss": 1.191,
+      "step": 1102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6307151230949592e-05,
+      "loss": 1.2279,
+      "step": 1103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6303634232121922e-05,
+      "loss": 1.1994,
+      "step": 1104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6300117233294256e-05,
+      "loss": 1.2622,
+      "step": 1105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.629660023446659e-05,
+      "loss": 1.2549,
+      "step": 1106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.629308323563892e-05,
+      "loss": 1.2179,
+      "step": 1107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6289566236811253e-05,
+      "loss": 1.2341,
+      "step": 1108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6286049237983587e-05,
+      "loss": 1.2697,
+      "step": 1109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.628253223915592e-05,
+      "loss": 1.2299,
+      "step": 1110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6279015240328254e-05,
+      "loss": 1.2508,
+      "step": 1111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6275498241500587e-05,
+      "loss": 1.2275,
+      "step": 1112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.627198124267292e-05,
+      "loss": 1.2379,
+      "step": 1113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.626846424384525e-05,
+      "loss": 1.2442,
+      "step": 1114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6264947245017584e-05,
+      "loss": 1.2146,
+      "step": 1115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6261430246189918e-05,
+      "loss": 1.2144,
+      "step": 1116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.625791324736225e-05,
+      "loss": 1.257,
+      "step": 1117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6254396248534585e-05,
+      "loss": 1.2365,
+      "step": 1118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.625087924970692e-05,
+      "loss": 1.2534,
+      "step": 1119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.624736225087925e-05,
+      "loss": 1.2337,
+      "step": 1120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6243845252051582e-05,
+      "loss": 1.1922,
+      "step": 1121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6240328253223916e-05,
+      "loss": 1.2206,
+      "step": 1122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.623681125439625e-05,
+      "loss": 1.2103,
+      "step": 1123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.6233294255568583e-05,
+      "loss": 1.2365,
+      "step": 1124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6229777256740916e-05,
+      "loss": 1.2553,
+      "step": 1125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6226260257913246e-05,
+      "loss": 1.2673,
+      "step": 1126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.622274325908558e-05,
+      "loss": 1.2005,
+      "step": 1127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6219226260257914e-05,
+      "loss": 1.2383,
+      "step": 1128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6215709261430247e-05,
+      "loss": 1.1619,
+      "step": 1129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.621219226260258e-05,
+      "loss": 1.2334,
+      "step": 1130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6208675263774914e-05,
+      "loss": 1.2069,
+      "step": 1131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6205158264947248e-05,
+      "loss": 1.2439,
+      "step": 1132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6201641266119578e-05,
+      "loss": 1.2191,
+      "step": 1133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.619812426729191e-05,
+      "loss": 1.2768,
+      "step": 1134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6194607268464245e-05,
+      "loss": 1.1972,
+      "step": 1135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.619109026963658e-05,
+      "loss": 1.1716,
+      "step": 1136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6187573270808912e-05,
+      "loss": 1.2209,
+      "step": 1137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6184056271981245e-05,
+      "loss": 1.2019,
+      "step": 1138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6180539273153576e-05,
+      "loss": 1.1963,
+      "step": 1139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.617702227432591e-05,
+      "loss": 1.2076,
+      "step": 1140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6173505275498243e-05,
+      "loss": 1.2119,
+      "step": 1141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6169988276670573e-05,
+      "loss": 1.2739,
+      "step": 1142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6166471277842906e-05,
+      "loss": 1.2496,
+      "step": 1143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.616295427901524e-05,
+      "loss": 1.2581,
+      "step": 1144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6159437280187573e-05,
+      "loss": 1.2636,
+      "step": 1145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6155920281359907e-05,
+      "loss": 1.2384,
+      "step": 1146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.615240328253224e-05,
+      "loss": 1.1999,
+      "step": 1147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6148886283704574e-05,
+      "loss": 1.2429,
+      "step": 1148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6145369284876904e-05,
+      "loss": 1.2405,
+      "step": 1149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6141852286049238e-05,
+      "loss": 1.216,
+      "step": 1150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.613833528722157e-05,
+      "loss": 1.1931,
+      "step": 1151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6134818288393905e-05,
+      "loss": 1.204,
+      "step": 1152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6131301289566238e-05,
+      "loss": 1.1752,
+      "step": 1153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.612778429073857e-05,
+      "loss": 1.1927,
+      "step": 1154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6124267291910902e-05,
+      "loss": 1.1986,
+      "step": 1155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6120750293083235e-05,
+      "loss": 1.2415,
+      "step": 1156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.611723329425557e-05,
+      "loss": 1.1907,
+      "step": 1157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6113716295427902e-05,
+      "loss": 1.203,
+      "step": 1158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6110199296600236e-05,
+      "loss": 1.2111,
+      "step": 1159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.610668229777257e-05,
+      "loss": 1.2394,
+      "step": 1160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.61031652989449e-05,
+      "loss": 1.2248,
+      "step": 1161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6099648300117233e-05,
+      "loss": 1.19,
+      "step": 1162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6096131301289567e-05,
+      "loss": 1.2122,
+      "step": 1163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.60926143024619e-05,
+      "loss": 1.2849,
+      "step": 1164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6089097303634234e-05,
+      "loss": 1.2292,
+      "step": 1165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6085580304806567e-05,
+      "loss": 1.2411,
+      "step": 1166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.60820633059789e-05,
+      "loss": 1.2494,
+      "step": 1167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.607854630715123e-05,
+      "loss": 1.2275,
+      "step": 1168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6075029308323564e-05,
+      "loss": 1.2151,
+      "step": 1169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6071512309495898e-05,
+      "loss": 1.2393,
+      "step": 1170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.606799531066823e-05,
+      "loss": 1.2184,
+      "step": 1171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6064478311840565e-05,
+      "loss": 1.2407,
+      "step": 1172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.60609613130129e-05,
+      "loss": 1.2247,
+      "step": 1173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.605744431418523e-05,
+      "loss": 1.1849,
+      "step": 1174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6053927315357562e-05,
+      "loss": 1.1702,
+      "step": 1175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6050410316529896e-05,
+      "loss": 1.217,
+      "step": 1176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6046893317702226e-05,
+      "loss": 1.2258,
+      "step": 1177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.604337631887456e-05,
+      "loss": 1.2378,
+      "step": 1178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6039859320046893e-05,
+      "loss": 1.2007,
+      "step": 1179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6036342321219226e-05,
+      "loss": 1.2444,
+      "step": 1180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.603282532239156e-05,
+      "loss": 1.1853,
+      "step": 1181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6029308323563893e-05,
+      "loss": 1.1744,
+      "step": 1182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6025791324736227e-05,
+      "loss": 1.2526,
+      "step": 1183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6022274325908557e-05,
+      "loss": 1.2007,
+      "step": 1184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.601875732708089e-05,
+      "loss": 1.1935,
+      "step": 1185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6015240328253224e-05,
+      "loss": 1.2495,
+      "step": 1186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6011723329425558e-05,
+      "loss": 1.2665,
+      "step": 1187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.600820633059789e-05,
+      "loss": 1.2183,
+      "step": 1188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6004689331770225e-05,
+      "loss": 1.2188,
+      "step": 1189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6001172332942555e-05,
+      "loss": 1.204,
+      "step": 1190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5997655334114888e-05,
+      "loss": 1.286,
+      "step": 1191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5994138335287222e-05,
+      "loss": 1.2708,
+      "step": 1192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5990621336459555e-05,
+      "loss": 1.2398,
+      "step": 1193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.598710433763189e-05,
+      "loss": 1.1917,
+      "step": 1194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5983587338804222e-05,
+      "loss": 1.2226,
+      "step": 1195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5980070339976553e-05,
+      "loss": 1.1732,
+      "step": 1196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5976553341148886e-05,
+      "loss": 1.2343,
+      "step": 1197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.597303634232122e-05,
+      "loss": 1.1882,
+      "step": 1198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5969519343493553e-05,
+      "loss": 1.2133,
+      "step": 1199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5966002344665887e-05,
+      "loss": 1.2564,
+      "step": 1200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.596248534583822e-05,
+      "loss": 1.2838,
+      "step": 1201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5958968347010554e-05,
+      "loss": 1.1787,
+      "step": 1202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5955451348182884e-05,
+      "loss": 1.2315,
+      "step": 1203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5951934349355217e-05,
+      "loss": 1.2288,
+      "step": 1204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.594841735052755e-05,
+      "loss": 1.22,
+      "step": 1205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5944900351699884e-05,
+      "loss": 1.2488,
+      "step": 1206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5941383352872218e-05,
+      "loss": 1.2061,
+      "step": 1207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.593786635404455e-05,
+      "loss": 1.2062,
+      "step": 1208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.593434935521688e-05,
+      "loss": 1.217,
+      "step": 1209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5930832356389215e-05,
+      "loss": 1.2538,
+      "step": 1210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.592731535756155e-05,
+      "loss": 1.2482,
+      "step": 1211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.592379835873388e-05,
+      "loss": 1.2026,
+      "step": 1212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5920281359906212e-05,
+      "loss": 1.221,
+      "step": 1213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5916764361078546e-05,
+      "loss": 1.2054,
+      "step": 1214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.591324736225088e-05,
+      "loss": 1.2032,
+      "step": 1215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5909730363423213e-05,
+      "loss": 1.2165,
+      "step": 1216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5906213364595546e-05,
+      "loss": 1.2594,
+      "step": 1217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.590269636576788e-05,
+      "loss": 1.2232,
+      "step": 1218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.589917936694021e-05,
+      "loss": 1.2237,
+      "step": 1219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5895662368112544e-05,
+      "loss": 1.1981,
+      "step": 1220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5892145369284877e-05,
+      "loss": 1.1558,
+      "step": 1221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.588862837045721e-05,
+      "loss": 1.2359,
+      "step": 1222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5885111371629544e-05,
+      "loss": 1.2455,
+      "step": 1223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5881594372801878e-05,
+      "loss": 1.2105,
+      "step": 1224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5878077373974208e-05,
+      "loss": 1.2657,
+      "step": 1225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.587456037514654e-05,
+      "loss": 1.2194,
+      "step": 1226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5871043376318875e-05,
+      "loss": 1.2768,
+      "step": 1227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.586752637749121e-05,
+      "loss": 1.2089,
+      "step": 1228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5864009378663542e-05,
+      "loss": 1.2833,
+      "step": 1229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5860492379835875e-05,
+      "loss": 1.2078,
+      "step": 1230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5856975381008206e-05,
+      "loss": 1.27,
+      "step": 1231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.585345838218054e-05,
+      "loss": 1.222,
+      "step": 1232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5849941383352873e-05,
+      "loss": 1.2686,
+      "step": 1233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5846424384525206e-05,
+      "loss": 1.2563,
+      "step": 1234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.584290738569754e-05,
+      "loss": 1.2188,
+      "step": 1235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5839390386869873e-05,
+      "loss": 1.2644,
+      "step": 1236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5835873388042207e-05,
+      "loss": 1.2422,
+      "step": 1237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5832356389214537e-05,
+      "loss": 1.2344,
+      "step": 1238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.582883939038687e-05,
+      "loss": 1.1985,
+      "step": 1239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5825322391559204e-05,
+      "loss": 1.2275,
+      "step": 1240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5821805392731537e-05,
+      "loss": 1.2306,
+      "step": 1241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.581828839390387e-05,
+      "loss": 1.1954,
+      "step": 1242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5814771395076204e-05,
+      "loss": 1.2313,
+      "step": 1243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5811254396248535e-05,
+      "loss": 1.255,
+      "step": 1244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5807737397420868e-05,
+      "loss": 1.1995,
+      "step": 1245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.58042203985932e-05,
+      "loss": 1.2066,
+      "step": 1246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5800703399765532e-05,
+      "loss": 1.2759,
+      "step": 1247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5797186400937865e-05,
+      "loss": 1.2552,
+      "step": 1248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.57936694021102e-05,
+      "loss": 1.2117,
+      "step": 1249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5790152403282532e-05,
+      "loss": 1.214,
+      "step": 1250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5786635404454866e-05,
+      "loss": 1.2498,
+      "step": 1251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.57831184056272e-05,
+      "loss": 1.221,
+      "step": 1252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5779601406799533e-05,
+      "loss": 1.1825,
+      "step": 1253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5776084407971863e-05,
+      "loss": 1.203,
+      "step": 1254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5772567409144197e-05,
+      "loss": 1.2366,
+      "step": 1255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.576905041031653e-05,
+      "loss": 1.1896,
+      "step": 1256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5765533411488864e-05,
+      "loss": 1.2454,
+      "step": 1257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5762016412661197e-05,
+      "loss": 1.2318,
+      "step": 1258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.575849941383353e-05,
+      "loss": 1.2906,
+      "step": 1259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.575498241500586e-05,
+      "loss": 1.2873,
+      "step": 1260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5751465416178194e-05,
+      "loss": 1.1845,
+      "step": 1261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5747948417350528e-05,
+      "loss": 1.2133,
+      "step": 1262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.574443141852286e-05,
+      "loss": 1.2232,
+      "step": 1263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5740914419695195e-05,
+      "loss": 1.2203,
+      "step": 1264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.573739742086753e-05,
+      "loss": 1.1917,
+      "step": 1265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.573388042203986e-05,
+      "loss": 1.2329,
+      "step": 1266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5730363423212192e-05,
+      "loss": 1.1737,
+      "step": 1267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5726846424384526e-05,
+      "loss": 1.2294,
+      "step": 1268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.572332942555686e-05,
+      "loss": 1.2758,
+      "step": 1269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5719812426729193e-05,
+      "loss": 1.2924,
+      "step": 1270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5716295427901526e-05,
+      "loss": 1.2148,
+      "step": 1271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.571277842907386e-05,
+      "loss": 1.2524,
+      "step": 1272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.570926143024619e-05,
+      "loss": 1.1779,
+      "step": 1273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5705744431418523e-05,
+      "loss": 1.1859,
+      "step": 1274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5702227432590857e-05,
+      "loss": 1.2316,
+      "step": 1275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.569871043376319e-05,
+      "loss": 1.2256,
+      "step": 1276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5695193434935524e-05,
+      "loss": 1.1993,
+      "step": 1277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5691676436107857e-05,
+      "loss": 1.2029,
+      "step": 1278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5688159437280188e-05,
+      "loss": 1.2189,
+      "step": 1279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.568464243845252e-05,
+      "loss": 1.3076,
+      "step": 1280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5681125439624855e-05,
+      "loss": 1.2262,
+      "step": 1281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5677608440797185e-05,
+      "loss": 1.2094,
+      "step": 1282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.567409144196952e-05,
+      "loss": 1.2041,
+      "step": 1283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5670574443141852e-05,
+      "loss": 1.2598,
+      "step": 1284
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5667057444314185e-05,
+      "loss": 1.2057,
+      "step": 1285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.566354044548652e-05,
+      "loss": 1.2753,
+      "step": 1286
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5660023446658852e-05,
+      "loss": 1.2381,
+      "step": 1287
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5656506447831183e-05,
+      "loss": 1.197,
+      "step": 1288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5652989449003516e-05,
+      "loss": 1.2315,
+      "step": 1289
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.564947245017585e-05,
+      "loss": 1.2037,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5645955451348183e-05,
+      "loss": 1.2277,
+      "step": 1291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5642438452520517e-05,
+      "loss": 1.2214,
+      "step": 1292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.563892145369285e-05,
+      "loss": 1.2337,
+      "step": 1293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5635404454865184e-05,
+      "loss": 1.1863,
+      "step": 1294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5631887456037514e-05,
+      "loss": 1.2409,
+      "step": 1295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5628370457209847e-05,
+      "loss": 1.2133,
+      "step": 1296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.562485345838218e-05,
+      "loss": 1.2646,
+      "step": 1297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5621336459554514e-05,
+      "loss": 1.2103,
+      "step": 1298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5617819460726848e-05,
+      "loss": 1.1925,
+      "step": 1299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.561430246189918e-05,
+      "loss": 1.173,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.561078546307151e-05,
+      "loss": 1.209,
+      "step": 1301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5607268464243845e-05,
+      "loss": 1.2408,
+      "step": 1302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.560375146541618e-05,
+      "loss": 1.2749,
+      "step": 1303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5600234466588512e-05,
+      "loss": 1.2443,
+      "step": 1304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5596717467760846e-05,
+      "loss": 1.2274,
+      "step": 1305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.559320046893318e-05,
+      "loss": 1.2118,
+      "step": 1306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5589683470105513e-05,
+      "loss": 1.2105,
+      "step": 1307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5586166471277843e-05,
+      "loss": 1.2724,
+      "step": 1308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5582649472450176e-05,
+      "loss": 1.2398,
+      "step": 1309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.557913247362251e-05,
+      "loss": 1.247,
+      "step": 1310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5575615474794843e-05,
+      "loss": 1.2091,
+      "step": 1311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5572098475967177e-05,
+      "loss": 1.1997,
+      "step": 1312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.556858147713951e-05,
+      "loss": 1.2428,
+      "step": 1313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.556506447831184e-05,
+      "loss": 1.1909,
+      "step": 1314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5561547479484174e-05,
+      "loss": 1.2309,
+      "step": 1315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5558030480656508e-05,
+      "loss": 1.2119,
+      "step": 1316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5554513481828838e-05,
+      "loss": 1.2483,
+      "step": 1317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.555099648300117e-05,
+      "loss": 1.2069,
+      "step": 1318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5547479484173505e-05,
+      "loss": 1.1998,
+      "step": 1319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.554396248534584e-05,
+      "loss": 1.2156,
+      "step": 1320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5540445486518172e-05,
+      "loss": 1.1994,
+      "step": 1321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5536928487690505e-05,
+      "loss": 1.2535,
+      "step": 1322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5533411488862836e-05,
+      "loss": 1.214,
+      "step": 1323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.552989449003517e-05,
+      "loss": 1.2357,
+      "step": 1324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5526377491207503e-05,
+      "loss": 1.1742,
+      "step": 1325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5522860492379836e-05,
+      "loss": 1.2288,
+      "step": 1326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.551934349355217e-05,
+      "loss": 1.2169,
+      "step": 1327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5515826494724503e-05,
+      "loss": 1.2325,
+      "step": 1328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5512309495896837e-05,
+      "loss": 1.239,
+      "step": 1329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5508792497069167e-05,
+      "loss": 1.2658,
+      "step": 1330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.55052754982415e-05,
+      "loss": 1.2246,
+      "step": 1331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5501758499413834e-05,
+      "loss": 1.2576,
+      "step": 1332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5498241500586167e-05,
+      "loss": 1.2253,
+      "step": 1333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.54947245017585e-05,
+      "loss": 1.2203,
+      "step": 1334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5491207502930834e-05,
+      "loss": 1.2381,
+      "step": 1335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5487690504103165e-05,
+      "loss": 1.2598,
+      "step": 1336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5484173505275498e-05,
+      "loss": 1.2289,
+      "step": 1337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.548065650644783e-05,
+      "loss": 1.2194,
+      "step": 1338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5477139507620165e-05,
+      "loss": 1.2315,
+      "step": 1339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.54736225087925e-05,
+      "loss": 1.2632,
+      "step": 1340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5470105509964832e-05,
+      "loss": 1.2073,
+      "step": 1341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5466588511137166e-05,
+      "loss": 1.2254,
+      "step": 1342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5463071512309496e-05,
+      "loss": 1.2053,
+      "step": 1343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.545955451348183e-05,
+      "loss": 1.2409,
+      "step": 1344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5456037514654163e-05,
+      "loss": 1.2681,
+      "step": 1345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5452520515826496e-05,
+      "loss": 1.253,
+      "step": 1346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.544900351699883e-05,
+      "loss": 1.271,
+      "step": 1347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5445486518171164e-05,
+      "loss": 1.2674,
+      "step": 1348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5441969519343494e-05,
+      "loss": 1.2031,
+      "step": 1349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5438452520515827e-05,
+      "loss": 1.2016,
+      "step": 1350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.543493552168816e-05,
+      "loss": 1.2699,
+      "step": 1351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.543141852286049e-05,
+      "loss": 1.2271,
+      "step": 1352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5427901524032824e-05,
+      "loss": 1.2153,
+      "step": 1353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5424384525205158e-05,
+      "loss": 1.1862,
+      "step": 1354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.542086752637749e-05,
+      "loss": 1.3138,
+      "step": 1355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5417350527549825e-05,
+      "loss": 1.1836,
+      "step": 1356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.541383352872216e-05,
+      "loss": 1.2527,
+      "step": 1357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.541031652989449e-05,
+      "loss": 1.2443,
+      "step": 1358
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5406799531066822e-05,
+      "loss": 1.2378,
+      "step": 1359
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5403282532239156e-05,
+      "loss": 1.2366,
+      "step": 1360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.539976553341149e-05,
+      "loss": 1.2085,
+      "step": 1361
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5396248534583823e-05,
+      "loss": 1.1944,
+      "step": 1362
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5392731535756156e-05,
+      "loss": 1.2371,
+      "step": 1363
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.538921453692849e-05,
+      "loss": 1.2625,
+      "step": 1364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.538569753810082e-05,
+      "loss": 1.2392,
+      "step": 1365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5382180539273153e-05,
+      "loss": 1.2381,
+      "step": 1366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5378663540445487e-05,
+      "loss": 1.1968,
+      "step": 1367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.537514654161782e-05,
+      "loss": 1.24,
+      "step": 1368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5371629542790154e-05,
+      "loss": 1.2073,
+      "step": 1369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5368112543962488e-05,
+      "loss": 1.2344,
+      "step": 1370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5364595545134818e-05,
+      "loss": 1.2355,
+      "step": 1371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.536107854630715e-05,
+      "loss": 1.2587,
+      "step": 1372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5357561547479485e-05,
+      "loss": 1.2676,
+      "step": 1373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5354044548651818e-05,
+      "loss": 1.2697,
+      "step": 1374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5350527549824152e-05,
+      "loss": 1.2134,
+      "step": 1375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5347010550996485e-05,
+      "loss": 1.1815,
+      "step": 1376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.534349355216882e-05,
+      "loss": 1.2664,
+      "step": 1377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.533997655334115e-05,
+      "loss": 1.2275,
+      "step": 1378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5336459554513482e-05,
+      "loss": 1.2512,
+      "step": 1379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5332942555685816e-05,
+      "loss": 1.2261,
+      "step": 1380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.532942555685815e-05,
+      "loss": 1.2325,
+      "step": 1381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5325908558030483e-05,
+      "loss": 1.2691,
+      "step": 1382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5322391559202817e-05,
+      "loss": 1.2446,
+      "step": 1383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5318874560375147e-05,
+      "loss": 1.2019,
+      "step": 1384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.531535756154748e-05,
+      "loss": 1.2174,
+      "step": 1385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5311840562719814e-05,
+      "loss": 1.217,
+      "step": 1386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5308323563892144e-05,
+      "loss": 1.2328,
+      "step": 1387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5304806565064477e-05,
+      "loss": 1.2367,
+      "step": 1388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.530128956623681e-05,
+      "loss": 1.2236,
+      "step": 1389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5297772567409144e-05,
+      "loss": 1.2103,
+      "step": 1390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5294255568581478e-05,
+      "loss": 1.2284,
+      "step": 1391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.529073856975381e-05,
+      "loss": 1.191,
+      "step": 1392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.528722157092614e-05,
+      "loss": 1.2285,
+      "step": 1393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5283704572098475e-05,
+      "loss": 1.24,
+      "step": 1394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.528018757327081e-05,
+      "loss": 1.2403,
+      "step": 1395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5276670574443142e-05,
+      "loss": 1.1818,
+      "step": 1396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5273153575615476e-05,
+      "loss": 1.2022,
+      "step": 1397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.526963657678781e-05,
+      "loss": 1.1895,
+      "step": 1398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5266119577960143e-05,
+      "loss": 1.2211,
+      "step": 1399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5262602579132473e-05,
+      "loss": 1.1824,
+      "step": 1400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5259085580304806e-05,
+      "loss": 1.2447,
+      "step": 1401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.525556858147714e-05,
+      "loss": 1.2318,
+      "step": 1402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5252051582649472e-05,
+      "loss": 1.2505,
+      "step": 1403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5248534583821805e-05,
+      "loss": 1.2797,
+      "step": 1404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5245017584994139e-05,
+      "loss": 1.2491,
+      "step": 1405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.524150058616647e-05,
+      "loss": 1.2297,
+      "step": 1406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5237983587338804e-05,
+      "loss": 1.2347,
+      "step": 1407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.5234466588511138e-05,
+      "loss": 1.2245,
+      "step": 1408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.523094958968347e-05,
+      "loss": 1.2277,
+      "step": 1409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5227432590855803e-05,
+      "loss": 1.1845,
+      "step": 1410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5223915592028137e-05,
+      "loss": 1.2624,
+      "step": 1411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5220398593200468e-05,
+      "loss": 1.1971,
+      "step": 1412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5216881594372802e-05,
+      "loss": 1.2711,
+      "step": 1413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5213364595545135e-05,
+      "loss": 1.2287,
+      "step": 1414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5209847596717469e-05,
+      "loss": 1.1874,
+      "step": 1415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.52063305978898e-05,
+      "loss": 1.1961,
+      "step": 1416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5202813599062134e-05,
+      "loss": 1.2311,
+      "step": 1417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5199296600234468e-05,
+      "loss": 1.228,
+      "step": 1418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5195779601406798e-05,
+      "loss": 1.2908,
+      "step": 1419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5192262602579132e-05,
+      "loss": 1.2002,
+      "step": 1420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5188745603751467e-05,
+      "loss": 1.2147,
+      "step": 1421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5185228604923797e-05,
+      "loss": 1.2251,
+      "step": 1422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.518171160609613e-05,
+      "loss": 1.2489,
+      "step": 1423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5178194607268464e-05,
+      "loss": 1.2737,
+      "step": 1424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5174677608440796e-05,
+      "loss": 1.2549,
+      "step": 1425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.517116060961313e-05,
+      "loss": 1.2153,
+      "step": 1426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5167643610785463e-05,
+      "loss": 1.1863,
+      "step": 1427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5164126611957795e-05,
+      "loss": 1.2295,
+      "step": 1428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5160609613130128e-05,
+      "loss": 1.2035,
+      "step": 1429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5157092614302462e-05,
+      "loss": 1.2322,
+      "step": 1430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5153575615474795e-05,
+      "loss": 1.2438,
+      "step": 1431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5150058616647127e-05,
+      "loss": 1.2477,
+      "step": 1432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.514654161781946e-05,
+      "loss": 1.2123,
+      "step": 1433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.5143024618991794e-05,
+      "loss": 1.2148,
+      "step": 1434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5139507620164126e-05,
+      "loss": 1.2466,
+      "step": 1435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.513599062133646e-05,
+      "loss": 1.2467,
+      "step": 1436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5132473622508793e-05,
+      "loss": 1.2274,
+      "step": 1437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5128956623681125e-05,
+      "loss": 1.1963,
+      "step": 1438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5125439624853458e-05,
+      "loss": 1.1986,
+      "step": 1439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5121922626025792e-05,
+      "loss": 1.2463,
+      "step": 1440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5118405627198124e-05,
+      "loss": 1.2602,
+      "step": 1441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5114888628370457e-05,
+      "loss": 1.1855,
+      "step": 1442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.511137162954279e-05,
+      "loss": 1.2492,
+      "step": 1443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5107854630715123e-05,
+      "loss": 1.2376,
+      "step": 1444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5104337631887456e-05,
+      "loss": 1.1955,
+      "step": 1445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.510082063305979e-05,
+      "loss": 1.2263,
+      "step": 1446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5097303634232121e-05,
+      "loss": 1.2599,
+      "step": 1447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5093786635404455e-05,
+      "loss": 1.2159,
+      "step": 1448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5090269636576789e-05,
+      "loss": 1.2194,
+      "step": 1449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5086752637749122e-05,
+      "loss": 1.2095,
+      "step": 1450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5083235638921454e-05,
+      "loss": 1.208,
+      "step": 1451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5079718640093787e-05,
+      "loss": 1.1836,
+      "step": 1452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5076201641266121e-05,
+      "loss": 1.2236,
+      "step": 1453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5072684642438451e-05,
+      "loss": 1.2317,
+      "step": 1454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5069167643610785e-05,
+      "loss": 1.2048,
+      "step": 1455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5065650644783118e-05,
+      "loss": 1.2183,
+      "step": 1456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.506213364595545e-05,
+      "loss": 1.1751,
+      "step": 1457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5058616647127783e-05,
+      "loss": 1.181,
+      "step": 1458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5055099648300117e-05,
+      "loss": 1.2032,
+      "step": 1459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5051582649472449e-05,
+      "loss": 1.2494,
+      "step": 1460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5048065650644782e-05,
+      "loss": 1.2559,
+      "step": 1461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5044548651817116e-05,
+      "loss": 1.1735,
+      "step": 1462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5041031652989448e-05,
+      "loss": 1.1952,
+      "step": 1463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5037514654161781e-05,
+      "loss": 1.2134,
+      "step": 1464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5033997655334115e-05,
+      "loss": 1.2108,
+      "step": 1465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5030480656506448e-05,
+      "loss": 1.1964,
+      "step": 1466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.502696365767878e-05,
+      "loss": 1.2221,
+      "step": 1467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5023446658851114e-05,
+      "loss": 1.2253,
+      "step": 1468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5019929660023447e-05,
+      "loss": 1.2106,
+      "step": 1469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5016412661195779e-05,
+      "loss": 1.2542,
+      "step": 1470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5012895662368112e-05,
+      "loss": 1.223,
+      "step": 1471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5009378663540446e-05,
+      "loss": 1.2443,
+      "step": 1472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5005861664712778e-05,
+      "loss": 1.2745,
+      "step": 1473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5002344665885111e-05,
+      "loss": 1.1359,
+      "step": 1474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4998827667057445e-05,
+      "loss": 1.1425,
+      "step": 1475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4995310668229777e-05,
+      "loss": 1.2442,
+      "step": 1476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.499179366940211e-05,
+      "loss": 1.2423,
+      "step": 1477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4988276670574444e-05,
+      "loss": 1.2195,
+      "step": 1478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4984759671746776e-05,
+      "loss": 1.2609,
+      "step": 1479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4981242672919109e-05,
+      "loss": 1.2114,
+      "step": 1480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4977725674091443e-05,
+      "loss": 1.2642,
+      "step": 1481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4974208675263774e-05,
+      "loss": 1.2269,
+      "step": 1482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4970691676436108e-05,
+      "loss": 1.2788,
+      "step": 1483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4967174677608442e-05,
+      "loss": 1.2489,
+      "step": 1484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4963657678780775e-05,
+      "loss": 1.2209,
+      "step": 1485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4960140679953107e-05,
+      "loss": 1.2547,
+      "step": 1486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.495662368112544e-05,
+      "loss": 1.2031,
+      "step": 1487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4953106682297774e-05,
+      "loss": 1.2622,
+      "step": 1488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4949589683470104e-05,
+      "loss": 1.1903,
+      "step": 1489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4946072684642438e-05,
+      "loss": 1.204,
+      "step": 1490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4942555685814771e-05,
+      "loss": 1.1957,
+      "step": 1491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4939038686987103e-05,
+      "loss": 1.2413,
+      "step": 1492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4935521688159436e-05,
+      "loss": 1.2303,
+      "step": 1493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.493200468933177e-05,
+      "loss": 1.2271,
+      "step": 1494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4928487690504102e-05,
+      "loss": 1.2209,
+      "step": 1495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4924970691676435e-05,
+      "loss": 1.2308,
+      "step": 1496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4921453692848769e-05,
+      "loss": 1.2298,
+      "step": 1497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.49179366940211e-05,
+      "loss": 1.2582,
+      "step": 1498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4914419695193434e-05,
+      "loss": 1.2682,
+      "step": 1499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4910902696365768e-05,
+      "loss": 1.2521,
+      "step": 1500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.49073856975381e-05,
+      "loss": 1.2158,
+      "step": 1501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4903868698710433e-05,
+      "loss": 1.2603,
+      "step": 1502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4900351699882767e-05,
+      "loss": 1.2172,
+      "step": 1503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.48968347010551e-05,
+      "loss": 1.2017,
+      "step": 1504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4893317702227432e-05,
+      "loss": 1.2523,
+      "step": 1505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4889800703399766e-05,
+      "loss": 1.2109,
+      "step": 1506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4886283704572099e-05,
+      "loss": 1.2383,
+      "step": 1507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4882766705744431e-05,
+      "loss": 1.2243,
+      "step": 1508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4879249706916764e-05,
+      "loss": 1.2396,
+      "step": 1509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4875732708089098e-05,
+      "loss": 1.196,
+      "step": 1510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.487221570926143e-05,
+      "loss": 1.2174,
+      "step": 1511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4868698710433763e-05,
+      "loss": 1.2221,
+      "step": 1512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4865181711606097e-05,
+      "loss": 1.2393,
+      "step": 1513
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4861664712778429e-05,
+      "loss": 1.2181,
+      "step": 1514
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4858147713950762e-05,
+      "loss": 1.2085,
+      "step": 1515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4854630715123096e-05,
+      "loss": 1.2426,
+      "step": 1516
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4851113716295428e-05,
+      "loss": 1.2302,
+      "step": 1517
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4847596717467761e-05,
+      "loss": 1.2276,
+      "step": 1518
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4844079718640095e-05,
+      "loss": 1.2425,
+      "step": 1519
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4840562719812428e-05,
+      "loss": 1.196,
+      "step": 1520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.483704572098476e-05,
+      "loss": 1.2464,
+      "step": 1521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4833528722157093e-05,
+      "loss": 1.2466,
+      "step": 1522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4830011723329427e-05,
+      "loss": 1.2306,
+      "step": 1523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4826494724501757e-05,
+      "loss": 1.2393,
+      "step": 1524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.482297772567409e-05,
+      "loss": 1.1882,
+      "step": 1525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4819460726846424e-05,
+      "loss": 1.2348,
+      "step": 1526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4815943728018756e-05,
+      "loss": 1.218,
+      "step": 1527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.481242672919109e-05,
+      "loss": 1.1784,
+      "step": 1528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4808909730363423e-05,
+      "loss": 1.2092,
+      "step": 1529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4805392731535755e-05,
+      "loss": 1.1527,
+      "step": 1530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4801875732708088e-05,
+      "loss": 1.2126,
+      "step": 1531
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4798358733880422e-05,
+      "loss": 1.1655,
+      "step": 1532
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4794841735052754e-05,
+      "loss": 1.287,
+      "step": 1533
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4791324736225087e-05,
+      "loss": 1.2112,
+      "step": 1534
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.478780773739742e-05,
+      "loss": 1.2487,
+      "step": 1535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4784290738569753e-05,
+      "loss": 1.1741,
+      "step": 1536
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4780773739742086e-05,
+      "loss": 1.2354,
+      "step": 1537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.477725674091442e-05,
+      "loss": 1.217,
+      "step": 1538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4773739742086753e-05,
+      "loss": 1.2007,
+      "step": 1539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4770222743259085e-05,
+      "loss": 1.2256,
+      "step": 1540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4766705744431419e-05,
+      "loss": 1.2159,
+      "step": 1541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4763188745603752e-05,
+      "loss": 1.2273,
+      "step": 1542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4759671746776084e-05,
+      "loss": 1.2301,
+      "step": 1543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4756154747948417e-05,
+      "loss": 1.2175,
+      "step": 1544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4752637749120751e-05,
+      "loss": 1.1568,
+      "step": 1545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4749120750293083e-05,
+      "loss": 1.1938,
+      "step": 1546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4745603751465416e-05,
+      "loss": 1.2032,
+      "step": 1547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.474208675263775e-05,
+      "loss": 1.2681,
+      "step": 1548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4738569753810082e-05,
+      "loss": 1.2212,
+      "step": 1549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4735052754982415e-05,
+      "loss": 1.2402,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4731535756154749e-05,
+      "loss": 1.2071,
+      "step": 1551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.472801875732708e-05,
+      "loss": 1.2314,
+      "step": 1552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4724501758499414e-05,
+      "loss": 1.205,
+      "step": 1553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4720984759671748e-05,
+      "loss": 1.2514,
+      "step": 1554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4717467760844081e-05,
+      "loss": 1.2572,
+      "step": 1555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4713950762016413e-05,
+      "loss": 1.2557,
+      "step": 1556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4710433763188746e-05,
+      "loss": 1.2648,
+      "step": 1557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.470691676436108e-05,
+      "loss": 1.2312,
+      "step": 1558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.470339976553341e-05,
+      "loss": 1.219,
+      "step": 1559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4699882766705744e-05,
+      "loss": 1.2048,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4696365767878077e-05,
+      "loss": 1.1931,
+      "step": 1561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4692848769050409e-05,
+      "loss": 1.162,
+      "step": 1562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4689331770222743e-05,
+      "loss": 1.2345,
+      "step": 1563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4685814771395076e-05,
+      "loss": 1.2298,
+      "step": 1564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4682297772567408e-05,
+      "loss": 1.2038,
+      "step": 1565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4678780773739741e-05,
+      "loss": 1.2182,
+      "step": 1566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4675263774912075e-05,
+      "loss": 1.2359,
+      "step": 1567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4671746776084407e-05,
+      "loss": 1.2097,
+      "step": 1568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.466822977725674e-05,
+      "loss": 1.2182,
+      "step": 1569
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4664712778429074e-05,
+      "loss": 1.1856,
+      "step": 1570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4661195779601406e-05,
+      "loss": 1.2012,
+      "step": 1571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.465767878077374e-05,
+      "loss": 1.2458,
+      "step": 1572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4654161781946073e-05,
+      "loss": 1.2226,
+      "step": 1573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4650644783118406e-05,
+      "loss": 1.2772,
+      "step": 1574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4647127784290738e-05,
+      "loss": 1.2493,
+      "step": 1575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4643610785463072e-05,
+      "loss": 1.2456,
+      "step": 1576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4640093786635405e-05,
+      "loss": 1.2404,
+      "step": 1577
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4636576787807737e-05,
+      "loss": 1.2283,
+      "step": 1578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.463305978898007e-05,
+      "loss": 1.1942,
+      "step": 1579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4629542790152404e-05,
+      "loss": 1.2799,
+      "step": 1580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4626025791324736e-05,
+      "loss": 1.2666,
+      "step": 1581
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.462250879249707e-05,
+      "loss": 1.1934,
+      "step": 1582
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4618991793669403e-05,
+      "loss": 1.2007,
+      "step": 1583
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4615474794841735e-05,
+      "loss": 1.2099,
+      "step": 1584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4611957796014068e-05,
+      "loss": 1.2401,
+      "step": 1585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4608440797186402e-05,
+      "loss": 1.2031,
+      "step": 1586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4604923798358734e-05,
+      "loss": 1.2036,
+      "step": 1587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4601406799531067e-05,
+      "loss": 1.2584,
+      "step": 1588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.45978898007034e-05,
+      "loss": 1.2336,
+      "step": 1589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.459437280187573e-05,
+      "loss": 1.2249,
+      "step": 1590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4590855803048064e-05,
+      "loss": 1.2335,
+      "step": 1591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.45873388042204e-05,
+      "loss": 1.2163,
+      "step": 1592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4583821805392733e-05,
+      "loss": 1.248,
+      "step": 1593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4580304806565063e-05,
+      "loss": 1.2276,
+      "step": 1594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4576787807737397e-05,
+      "loss": 1.2554,
+      "step": 1595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.457327080890973e-05,
+      "loss": 1.1845,
+      "step": 1596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4569753810082062e-05,
+      "loss": 1.1922,
+      "step": 1597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4566236811254396e-05,
+      "loss": 1.2067,
+      "step": 1598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4562719812426729e-05,
+      "loss": 1.2052,
+      "step": 1599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4559202813599061e-05,
+      "loss": 1.2,
+      "step": 1600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4555685814771394e-05,
+      "loss": 1.2136,
+      "step": 1601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4552168815943728e-05,
+      "loss": 1.2218,
+      "step": 1602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.454865181711606e-05,
+      "loss": 1.215,
+      "step": 1603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4545134818288393e-05,
+      "loss": 1.2456,
+      "step": 1604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4541617819460727e-05,
+      "loss": 1.2409,
+      "step": 1605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4538100820633059e-05,
+      "loss": 1.1815,
+      "step": 1606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4534583821805392e-05,
+      "loss": 1.2119,
+      "step": 1607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4531066822977726e-05,
+      "loss": 1.1752,
+      "step": 1608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.452754982415006e-05,
+      "loss": 1.2051,
+      "step": 1609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4524032825322391e-05,
+      "loss": 1.2217,
+      "step": 1610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4520515826494725e-05,
+      "loss": 1.1874,
+      "step": 1611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4516998827667058e-05,
+      "loss": 1.2521,
+      "step": 1612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.451348182883939e-05,
+      "loss": 1.1921,
+      "step": 1613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4509964830011723e-05,
+      "loss": 1.2577,
+      "step": 1614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.4506447831184057e-05,
+      "loss": 1.2428,
+      "step": 1615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4502930832356389e-05,
+      "loss": 1.2073,
+      "step": 1616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4499413833528722e-05,
+      "loss": 1.2197,
+      "step": 1617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4495896834701056e-05,
+      "loss": 1.1998,
+      "step": 1618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4492379835873388e-05,
+      "loss": 1.2459,
+      "step": 1619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4488862837045721e-05,
+      "loss": 1.2402,
+      "step": 1620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4485345838218055e-05,
+      "loss": 1.288,
+      "step": 1621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4481828839390387e-05,
+      "loss": 1.2008,
+      "step": 1622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.447831184056272e-05,
+      "loss": 1.2618,
+      "step": 1623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4474794841735054e-05,
+      "loss": 1.2085,
+      "step": 1624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4471277842907384e-05,
+      "loss": 1.2337,
+      "step": 1625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4467760844079717e-05,
+      "loss": 1.2472,
+      "step": 1626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4464243845252053e-05,
+      "loss": 1.2596,
+      "step": 1627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4460726846424386e-05,
+      "loss": 1.2621,
+      "step": 1628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4457209847596716e-05,
+      "loss": 1.261,
+      "step": 1629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.445369284876905e-05,
+      "loss": 1.2138,
+      "step": 1630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4450175849941383e-05,
+      "loss": 1.1868,
+      "step": 1631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4446658851113715e-05,
+      "loss": 1.2306,
+      "step": 1632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4443141852286049e-05,
+      "loss": 1.2151,
+      "step": 1633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4439624853458382e-05,
+      "loss": 1.2464,
+      "step": 1634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4436107854630714e-05,
+      "loss": 1.2345,
+      "step": 1635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4432590855803047e-05,
+      "loss": 1.2387,
+      "step": 1636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4429073856975381e-05,
+      "loss": 1.237,
+      "step": 1637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4425556858147713e-05,
+      "loss": 1.206,
+      "step": 1638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4422039859320046e-05,
+      "loss": 1.2088,
+      "step": 1639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.441852286049238e-05,
+      "loss": 1.1782,
+      "step": 1640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4415005861664712e-05,
+      "loss": 1.1844,
+      "step": 1641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4411488862837045e-05,
+      "loss": 1.1998,
+      "step": 1642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4407971864009379e-05,
+      "loss": 1.2079,
+      "step": 1643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4404454865181712e-05,
+      "loss": 1.2166,
+      "step": 1644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4400937866354044e-05,
+      "loss": 1.2129,
+      "step": 1645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4397420867526378e-05,
+      "loss": 1.2511,
+      "step": 1646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4393903868698711e-05,
+      "loss": 1.1897,
+      "step": 1647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4390386869871043e-05,
+      "loss": 1.2149,
+      "step": 1648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4386869871043376e-05,
+      "loss": 1.2512,
+      "step": 1649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.438335287221571e-05,
+      "loss": 1.186,
+      "step": 1650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4379835873388042e-05,
+      "loss": 1.2112,
+      "step": 1651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4376318874560375e-05,
+      "loss": 1.2464,
+      "step": 1652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4372801875732709e-05,
+      "loss": 1.2232,
+      "step": 1653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.436928487690504e-05,
+      "loss": 1.1947,
+      "step": 1654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4365767878077374e-05,
+      "loss": 1.2503,
+      "step": 1655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4362250879249708e-05,
+      "loss": 1.207,
+      "step": 1656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.435873388042204e-05,
+      "loss": 1.2326,
+      "step": 1657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4355216881594373e-05,
+      "loss": 1.192,
+      "step": 1658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4351699882766707e-05,
+      "loss": 1.2446,
+      "step": 1659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4348182883939037e-05,
+      "loss": 1.2152,
+      "step": 1660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.434466588511137e-05,
+      "loss": 1.2232,
+      "step": 1661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4341148886283704e-05,
+      "loss": 1.2329,
+      "step": 1662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4337631887456039e-05,
+      "loss": 1.2114,
+      "step": 1663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.433411488862837e-05,
+      "loss": 1.227,
+      "step": 1664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4330597889800703e-05,
+      "loss": 1.1745,
+      "step": 1665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4327080890973036e-05,
+      "loss": 1.2053,
+      "step": 1666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4323563892145368e-05,
+      "loss": 1.1685,
+      "step": 1667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4320046893317702e-05,
+      "loss": 1.1735,
+      "step": 1668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4316529894490035e-05,
+      "loss": 1.2156,
+      "step": 1669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4313012895662367e-05,
+      "loss": 1.2673,
+      "step": 1670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.43094958968347e-05,
+      "loss": 1.2102,
+      "step": 1671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4305978898007034e-05,
+      "loss": 1.2208,
+      "step": 1672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4302461899179366e-05,
+      "loss": 1.1901,
+      "step": 1673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.42989449003517e-05,
+      "loss": 1.2153,
+      "step": 1674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4295427901524033e-05,
+      "loss": 1.1849,
+      "step": 1675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4291910902696365e-05,
+      "loss": 1.2409,
+      "step": 1676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4288393903868698e-05,
+      "loss": 1.2249,
+      "step": 1677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4284876905041032e-05,
+      "loss": 1.183,
+      "step": 1678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4281359906213364e-05,
+      "loss": 1.2748,
+      "step": 1679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4277842907385697e-05,
+      "loss": 1.2006,
+      "step": 1680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.427432590855803e-05,
+      "loss": 1.2231,
+      "step": 1681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4270808909730364e-05,
+      "loss": 1.1905,
+      "step": 1682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4267291910902696e-05,
+      "loss": 1.1927,
+      "step": 1683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.426377491207503e-05,
+      "loss": 1.1946,
+      "step": 1684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4260257913247363e-05,
+      "loss": 1.1925,
+      "step": 1685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4256740914419695e-05,
+      "loss": 1.2325,
+      "step": 1686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4253223915592028e-05,
+      "loss": 1.2257,
+      "step": 1687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4249706916764362e-05,
+      "loss": 1.2307,
+      "step": 1688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4246189917936694e-05,
+      "loss": 1.1727,
+      "step": 1689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4242672919109027e-05,
+      "loss": 1.2376,
+      "step": 1690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.423915592028136e-05,
+      "loss": 1.2065,
+      "step": 1691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4235638921453693e-05,
+      "loss": 1.2148,
+      "step": 1692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4232121922626026e-05,
+      "loss": 1.2278,
+      "step": 1693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.422860492379836e-05,
+      "loss": 1.1572,
+      "step": 1694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.422508792497069e-05,
+      "loss": 1.2114,
+      "step": 1695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4221570926143023e-05,
+      "loss": 1.2622,
+      "step": 1696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4218053927315357e-05,
+      "loss": 1.1979,
+      "step": 1697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4214536928487692e-05,
+      "loss": 1.1278,
+      "step": 1698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4211019929660022e-05,
+      "loss": 1.2357,
+      "step": 1699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4207502930832356e-05,
+      "loss": 1.223,
+      "step": 1700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.420398593200469e-05,
+      "loss": 1.2319,
+      "step": 1701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4200468933177021e-05,
+      "loss": 1.2102,
+      "step": 1702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4196951934349355e-05,
+      "loss": 1.2009,
+      "step": 1703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4193434935521688e-05,
+      "loss": 1.2568,
+      "step": 1704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.418991793669402e-05,
+      "loss": 1.1936,
+      "step": 1705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4186400937866353e-05,
+      "loss": 1.2279,
+      "step": 1706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4182883939038687e-05,
+      "loss": 1.1693,
+      "step": 1707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4179366940211019e-05,
+      "loss": 1.2268,
+      "step": 1708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4175849941383352e-05,
+      "loss": 1.1787,
+      "step": 1709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4172332942555686e-05,
+      "loss": 1.2104,
+      "step": 1710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4168815943728018e-05,
+      "loss": 1.2374,
+      "step": 1711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4165298944900351e-05,
+      "loss": 1.215,
+      "step": 1712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4161781946072685e-05,
+      "loss": 1.2079,
+      "step": 1713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4158264947245017e-05,
+      "loss": 1.1771,
+      "step": 1714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.415474794841735e-05,
+      "loss": 1.1782,
+      "step": 1715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4151230949589684e-05,
+      "loss": 1.2516,
+      "step": 1716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4147713950762017e-05,
+      "loss": 1.196,
+      "step": 1717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4144196951934349e-05,
+      "loss": 1.2479,
+      "step": 1718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4140679953106683e-05,
+      "loss": 1.2179,
+      "step": 1719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4137162954279016e-05,
+      "loss": 1.1923,
+      "step": 1720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4133645955451348e-05,
+      "loss": 1.2614,
+      "step": 1721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4130128956623681e-05,
+      "loss": 1.2342,
+      "step": 1722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4126611957796015e-05,
+      "loss": 1.2796,
+      "step": 1723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4123094958968347e-05,
+      "loss": 1.1987,
+      "step": 1724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.411957796014068e-05,
+      "loss": 1.2612,
+      "step": 1725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4116060961313014e-05,
+      "loss": 1.1774,
+      "step": 1726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4112543962485346e-05,
+      "loss": 1.2257,
+      "step": 1727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.410902696365768e-05,
+      "loss": 1.2239,
+      "step": 1728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4105509964830013e-05,
+      "loss": 1.1513,
+      "step": 1729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4101992966002343e-05,
+      "loss": 1.1968,
+      "step": 1730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4098475967174676e-05,
+      "loss": 1.2387,
+      "step": 1731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.409495896834701e-05,
+      "loss": 1.2198,
+      "step": 1732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4091441969519345e-05,
+      "loss": 1.2015,
+      "step": 1733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4087924970691675e-05,
+      "loss": 1.2393,
+      "step": 1734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4084407971864009e-05,
+      "loss": 1.2193,
+      "step": 1735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4080890973036342e-05,
+      "loss": 1.1713,
+      "step": 1736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4077373974208674e-05,
+      "loss": 1.1787,
+      "step": 1737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4073856975381008e-05,
+      "loss": 1.2528,
+      "step": 1738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4070339976553341e-05,
+      "loss": 1.2745,
+      "step": 1739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4066822977725673e-05,
+      "loss": 1.251,
+      "step": 1740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4063305978898007e-05,
+      "loss": 1.2044,
+      "step": 1741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.405978898007034e-05,
+      "loss": 1.2197,
+      "step": 1742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4056271981242672e-05,
+      "loss": 1.2156,
+      "step": 1743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4052754982415005e-05,
+      "loss": 1.2523,
+      "step": 1744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4049237983587339e-05,
+      "loss": 1.1816,
+      "step": 1745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.404572098475967e-05,
+      "loss": 1.2435,
+      "step": 1746
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4042203985932004e-05,
+      "loss": 1.1767,
+      "step": 1747
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4038686987104338e-05,
+      "loss": 1.2407,
+      "step": 1748
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.403516998827667e-05,
+      "loss": 1.1905,
+      "step": 1749
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4031652989449003e-05,
+      "loss": 1.2543,
+      "step": 1750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4028135990621337e-05,
+      "loss": 1.2262,
+      "step": 1751
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.402461899179367e-05,
+      "loss": 1.2329,
+      "step": 1752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4021101992966002e-05,
+      "loss": 1.2319,
+      "step": 1753
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4017584994138336e-05,
+      "loss": 1.2059,
+      "step": 1754
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4014067995310669e-05,
+      "loss": 1.2335,
+      "step": 1755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4010550996483001e-05,
+      "loss": 1.2007,
+      "step": 1756
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4007033997655334e-05,
+      "loss": 1.2309,
+      "step": 1757
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4003516998827668e-05,
+      "loss": 1.1705,
+      "step": 1758
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4e-05,
+      "loss": 1.2111,
+      "step": 1759
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3996483001172333e-05,
+      "loss": 1.2106,
+      "step": 1760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3992966002344667e-05,
+      "loss": 1.1813,
+      "step": 1761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3989449003516999e-05,
+      "loss": 1.2188,
+      "step": 1762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3985932004689332e-05,
+      "loss": 1.2214,
+      "step": 1763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3982415005861666e-05,
+      "loss": 1.2332,
+      "step": 1764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3978898007033996e-05,
+      "loss": 1.2036,
+      "step": 1765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.397538100820633e-05,
+      "loss": 1.1711,
+      "step": 1766
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3971864009378663e-05,
+      "loss": 1.24,
+      "step": 1767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3968347010550996e-05,
+      "loss": 1.2587,
+      "step": 1768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3964830011723328e-05,
+      "loss": 1.2186,
+      "step": 1769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3961313012895662e-05,
+      "loss": 1.2233,
+      "step": 1770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3957796014067995e-05,
+      "loss": 1.2248,
+      "step": 1771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3954279015240327e-05,
+      "loss": 1.2431,
+      "step": 1772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.395076201641266e-05,
+      "loss": 1.2578,
+      "step": 1773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3947245017584994e-05,
+      "loss": 1.2221,
+      "step": 1774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3943728018757326e-05,
+      "loss": 1.2141,
+      "step": 1775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.394021101992966e-05,
+      "loss": 1.2598,
+      "step": 1776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3936694021101993e-05,
+      "loss": 1.2312,
+      "step": 1777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3933177022274325e-05,
+      "loss": 1.2151,
+      "step": 1778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3929660023446658e-05,
+      "loss": 1.1931,
+      "step": 1779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3926143024618992e-05,
+      "loss": 1.2581,
+      "step": 1780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3922626025791324e-05,
+      "loss": 1.276,
+      "step": 1781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3919109026963657e-05,
+      "loss": 1.2633,
+      "step": 1782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.391559202813599e-05,
+      "loss": 1.1935,
+      "step": 1783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3912075029308323e-05,
+      "loss": 1.2106,
+      "step": 1784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3908558030480656e-05,
+      "loss": 1.1757,
+      "step": 1785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.390504103165299e-05,
+      "loss": 1.1917,
+      "step": 1786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3901524032825323e-05,
+      "loss": 1.2271,
+      "step": 1787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3898007033997655e-05,
+      "loss": 1.233,
+      "step": 1788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3894490035169989e-05,
+      "loss": 1.2188,
+      "step": 1789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3890973036342322e-05,
+      "loss": 1.2508,
+      "step": 1790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3887456037514654e-05,
+      "loss": 1.2102,
+      "step": 1791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3883939038686987e-05,
+      "loss": 1.2317,
+      "step": 1792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3880422039859321e-05,
+      "loss": 1.2171,
+      "step": 1793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3876905041031653e-05,
+      "loss": 1.1819,
+      "step": 1794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3873388042203986e-05,
+      "loss": 1.2283,
+      "step": 1795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.386987104337632e-05,
+      "loss": 1.2479,
+      "step": 1796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.386635404454865e-05,
+      "loss": 1.2424,
+      "step": 1797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3862837045720985e-05,
+      "loss": 1.2071,
+      "step": 1798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3859320046893319e-05,
+      "loss": 1.214,
+      "step": 1799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3855803048065649e-05,
+      "loss": 1.1818,
+      "step": 1800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3852286049237982e-05,
+      "loss": 1.2428,
+      "step": 1801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3848769050410316e-05,
+      "loss": 1.2008,
+      "step": 1802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3845252051582648e-05,
+      "loss": 1.2311,
+      "step": 1803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3841735052754981e-05,
+      "loss": 1.2253,
+      "step": 1804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3838218053927315e-05,
+      "loss": 1.2285,
+      "step": 1805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3834701055099648e-05,
+      "loss": 1.2213,
+      "step": 1806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.383118405627198e-05,
+      "loss": 1.1902,
+      "step": 1807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3827667057444314e-05,
+      "loss": 1.2142,
+      "step": 1808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3824150058616647e-05,
+      "loss": 1.2618,
+      "step": 1809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3820633059788979e-05,
+      "loss": 1.2074,
+      "step": 1810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3817116060961313e-05,
+      "loss": 1.2397,
+      "step": 1811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3813599062133646e-05,
+      "loss": 1.1755,
+      "step": 1812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3810082063305978e-05,
+      "loss": 1.2054,
+      "step": 1813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3806565064478311e-05,
+      "loss": 1.2412,
+      "step": 1814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3803048065650645e-05,
+      "loss": 1.2179,
+      "step": 1815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3799531066822977e-05,
+      "loss": 1.2361,
+      "step": 1816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.379601406799531e-05,
+      "loss": 1.2581,
+      "step": 1817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3792497069167644e-05,
+      "loss": 1.2315,
+      "step": 1818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3788980070339976e-05,
+      "loss": 1.2574,
+      "step": 1819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.378546307151231e-05,
+      "loss": 1.1946,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3781946072684643e-05,
+      "loss": 1.2552,
+      "step": 1821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3778429073856976e-05,
+      "loss": 1.191,
+      "step": 1822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3774912075029308e-05,
+      "loss": 1.1998,
+      "step": 1823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3771395076201642e-05,
+      "loss": 1.2514,
+      "step": 1824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3767878077373975e-05,
+      "loss": 1.2301,
+      "step": 1825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3764361078546307e-05,
+      "loss": 1.2621,
+      "step": 1826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.376084407971864e-05,
+      "loss": 1.1842,
+      "step": 1827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3757327080890974e-05,
+      "loss": 1.2081,
+      "step": 1828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3753810082063306e-05,
+      "loss": 1.2372,
+      "step": 1829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.375029308323564e-05,
+      "loss": 1.2102,
+      "step": 1830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3746776084407973e-05,
+      "loss": 1.169,
+      "step": 1831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3743259085580303e-05,
+      "loss": 1.2192,
+      "step": 1832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3739742086752638e-05,
+      "loss": 1.1861,
+      "step": 1833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3736225087924972e-05,
+      "loss": 1.1673,
+      "step": 1834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3732708089097302e-05,
+      "loss": 1.2335,
+      "step": 1835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3729191090269635e-05,
+      "loss": 1.2061,
+      "step": 1836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3725674091441969e-05,
+      "loss": 1.23,
+      "step": 1837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.37221570926143e-05,
+      "loss": 1.1918,
+      "step": 1838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3718640093786634e-05,
+      "loss": 1.1594,
+      "step": 1839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3715123094958968e-05,
+      "loss": 1.1697,
+      "step": 1840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3711606096131301e-05,
+      "loss": 1.2151,
+      "step": 1841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3708089097303633e-05,
+      "loss": 1.2151,
+      "step": 1842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3704572098475967e-05,
+      "loss": 1.2176,
+      "step": 1843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.37010550996483e-05,
+      "loss": 1.224,
+      "step": 1844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3697538100820632e-05,
+      "loss": 1.25,
+      "step": 1845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3694021101992966e-05,
+      "loss": 1.2788,
+      "step": 1846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3690504103165299e-05,
+      "loss": 1.2098,
+      "step": 1847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3686987104337631e-05,
+      "loss": 1.2382,
+      "step": 1848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3683470105509964e-05,
+      "loss": 1.2265,
+      "step": 1849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3679953106682298e-05,
+      "loss": 1.1955,
+      "step": 1850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.367643610785463e-05,
+      "loss": 1.2722,
+      "step": 1851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3672919109026963e-05,
+      "loss": 1.215,
+      "step": 1852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3669402110199297e-05,
+      "loss": 1.1967,
+      "step": 1853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3665885111371629e-05,
+      "loss": 1.2093,
+      "step": 1854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3662368112543962e-05,
+      "loss": 1.2117,
+      "step": 1855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3658851113716296e-05,
+      "loss": 1.1771,
+      "step": 1856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.365533411488863e-05,
+      "loss": 1.2698,
+      "step": 1857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3651817116060961e-05,
+      "loss": 1.2357,
+      "step": 1858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3648300117233295e-05,
+      "loss": 1.216,
+      "step": 1859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3644783118405628e-05,
+      "loss": 1.2684,
+      "step": 1860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.364126611957796e-05,
+      "loss": 1.2508,
+      "step": 1861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3637749120750294e-05,
+      "loss": 1.305,
+      "step": 1862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3634232121922627e-05,
+      "loss": 1.209,
+      "step": 1863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3630715123094959e-05,
+      "loss": 1.2133,
+      "step": 1864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3627198124267292e-05,
+      "loss": 1.2818,
+      "step": 1865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3623681125439626e-05,
+      "loss": 1.1934,
+      "step": 1866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3620164126611956e-05,
+      "loss": 1.2161,
+      "step": 1867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3616647127784291e-05,
+      "loss": 1.167,
+      "step": 1868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3613130128956625e-05,
+      "loss": 1.1878,
+      "step": 1869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3609613130128955e-05,
+      "loss": 1.1702,
+      "step": 1870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3606096131301288e-05,
+      "loss": 1.2245,
+      "step": 1871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3602579132473622e-05,
+      "loss": 1.2401,
+      "step": 1872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3599062133645954e-05,
+      "loss": 1.2622,
+      "step": 1873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3595545134818287e-05,
+      "loss": 1.1907,
+      "step": 1874
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3592028135990621e-05,
+      "loss": 1.2414,
+      "step": 1875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3588511137162954e-05,
+      "loss": 1.2365,
+      "step": 1876
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3584994138335286e-05,
+      "loss": 1.2104,
+      "step": 1877
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.358147713950762e-05,
+      "loss": 1.2096,
+      "step": 1878
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3577960140679953e-05,
+      "loss": 1.2114,
+      "step": 1879
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3574443141852285e-05,
+      "loss": 1.1521,
+      "step": 1880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3570926143024619e-05,
+      "loss": 1.2185,
+      "step": 1881
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3567409144196952e-05,
+      "loss": 1.2249,
+      "step": 1882
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3563892145369284e-05,
+      "loss": 1.2331,
+      "step": 1883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3560375146541617e-05,
+      "loss": 1.216,
+      "step": 1884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3556858147713951e-05,
+      "loss": 1.2455,
+      "step": 1885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3553341148886283e-05,
+      "loss": 1.2237,
+      "step": 1886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3549824150058616e-05,
+      "loss": 1.2425,
+      "step": 1887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.354630715123095e-05,
+      "loss": 1.1681,
+      "step": 1888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3542790152403282e-05,
+      "loss": 1.1844,
+      "step": 1889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3539273153575615e-05,
+      "loss": 1.2427,
+      "step": 1890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3535756154747949e-05,
+      "loss": 1.2855,
+      "step": 1891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.353223915592028e-05,
+      "loss": 1.2269,
+      "step": 1892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3528722157092614e-05,
+      "loss": 1.2174,
+      "step": 1893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3525205158264948e-05,
+      "loss": 1.1992,
+      "step": 1894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3521688159437281e-05,
+      "loss": 1.1968,
+      "step": 1895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3518171160609613e-05,
+      "loss": 1.2337,
+      "step": 1896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3514654161781947e-05,
+      "loss": 1.2451,
+      "step": 1897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.351113716295428e-05,
+      "loss": 1.2417,
+      "step": 1898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3507620164126612e-05,
+      "loss": 1.2036,
+      "step": 1899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3504103165298945e-05,
+      "loss": 1.2011,
+      "step": 1900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3500586166471279e-05,
+      "loss": 1.1984,
+      "step": 1901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3497069167643609e-05,
+      "loss": 1.2061,
+      "step": 1902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3493552168815943e-05,
+      "loss": 1.2303,
+      "step": 1903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3490035169988278e-05,
+      "loss": 1.179,
+      "step": 1904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3486518171160608e-05,
+      "loss": 1.2063,
+      "step": 1905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3483001172332941e-05,
+      "loss": 1.2309,
+      "step": 1906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3479484173505275e-05,
+      "loss": 1.2293,
+      "step": 1907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3475967174677607e-05,
+      "loss": 1.2014,
+      "step": 1908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.347245017584994e-05,
+      "loss": 1.2527,
+      "step": 1909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3468933177022274e-05,
+      "loss": 1.2166,
+      "step": 1910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3465416178194607e-05,
+      "loss": 1.1899,
+      "step": 1911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.346189917936694e-05,
+      "loss": 1.203,
+      "step": 1912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3458382180539273e-05,
+      "loss": 1.1705,
+      "step": 1913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3454865181711606e-05,
+      "loss": 1.2173,
+      "step": 1914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3451348182883938e-05,
+      "loss": 1.1736,
+      "step": 1915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3447831184056272e-05,
+      "loss": 1.1637,
+      "step": 1916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3444314185228605e-05,
+      "loss": 1.2593,
+      "step": 1917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3440797186400937e-05,
+      "loss": 1.2267,
+      "step": 1918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.343728018757327e-05,
+      "loss": 1.1689,
+      "step": 1919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3433763188745604e-05,
+      "loss": 1.204,
+      "step": 1920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3430246189917936e-05,
+      "loss": 1.187,
+      "step": 1921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.342672919109027e-05,
+      "loss": 1.2174,
+      "step": 1922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3423212192262603e-05,
+      "loss": 1.2043,
+      "step": 1923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3419695193434935e-05,
+      "loss": 1.2476,
+      "step": 1924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3416178194607268e-05,
+      "loss": 1.2235,
+      "step": 1925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3412661195779602e-05,
+      "loss": 1.2443,
+      "step": 1926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3409144196951934e-05,
+      "loss": 1.1968,
+      "step": 1927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3405627198124267e-05,
+      "loss": 1.2082,
+      "step": 1928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.34021101992966e-05,
+      "loss": 1.2211,
+      "step": 1929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3398593200468934e-05,
+      "loss": 1.2031,
+      "step": 1930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3395076201641266e-05,
+      "loss": 1.2112,
+      "step": 1931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.33915592028136e-05,
+      "loss": 1.1744,
+      "step": 1932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3388042203985933e-05,
+      "loss": 1.1835,
+      "step": 1933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3384525205158265e-05,
+      "loss": 1.2161,
+      "step": 1934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3381008206330598e-05,
+      "loss": 1.2502,
+      "step": 1935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3377491207502932e-05,
+      "loss": 1.2322,
+      "step": 1936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3373974208675262e-05,
+      "loss": 1.265,
+      "step": 1937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3370457209847596e-05,
+      "loss": 1.1892,
+      "step": 1938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.336694021101993e-05,
+      "loss": 1.2489,
+      "step": 1939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3363423212192261e-05,
+      "loss": 1.2299,
+      "step": 1940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3359906213364594e-05,
+      "loss": 1.1895,
+      "step": 1941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3356389214536928e-05,
+      "loss": 1.1915,
+      "step": 1942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.335287221570926e-05,
+      "loss": 1.2282,
+      "step": 1943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3349355216881593e-05,
+      "loss": 1.214,
+      "step": 1944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3345838218053927e-05,
+      "loss": 1.1991,
+      "step": 1945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.334232121922626e-05,
+      "loss": 1.2218,
+      "step": 1946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3338804220398592e-05,
+      "loss": 1.2125,
+      "step": 1947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3335287221570926e-05,
+      "loss": 1.1889,
+      "step": 1948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.333177022274326e-05,
+      "loss": 1.2195,
+      "step": 1949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3328253223915591e-05,
+      "loss": 1.1985,
+      "step": 1950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3324736225087925e-05,
+      "loss": 1.1861,
+      "step": 1951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3321219226260258e-05,
+      "loss": 1.2135,
+      "step": 1952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.331770222743259e-05,
+      "loss": 1.2549,
+      "step": 1953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3314185228604924e-05,
+      "loss": 1.1822,
+      "step": 1954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3310668229777257e-05,
+      "loss": 1.2019,
+      "step": 1955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3307151230949589e-05,
+      "loss": 1.202,
+      "step": 1956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3303634232121922e-05,
+      "loss": 1.224,
+      "step": 1957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3300117233294256e-05,
+      "loss": 1.1709,
+      "step": 1958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3296600234466588e-05,
+      "loss": 1.1869,
+      "step": 1959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3293083235638921e-05,
+      "loss": 1.2224,
+      "step": 1960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3289566236811255e-05,
+      "loss": 1.2009,
+      "step": 1961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3286049237983587e-05,
+      "loss": 1.1752,
+      "step": 1962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.328253223915592e-05,
+      "loss": 1.1565,
+      "step": 1963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3279015240328254e-05,
+      "loss": 1.2126,
+      "step": 1964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3275498241500587e-05,
+      "loss": 1.2623,
+      "step": 1965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3271981242672919e-05,
+      "loss": 1.2661,
+      "step": 1966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3268464243845253e-05,
+      "loss": 1.1885,
+      "step": 1967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3264947245017586e-05,
+      "loss": 1.2019,
+      "step": 1968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3261430246189918e-05,
+      "loss": 1.1774,
+      "step": 1969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3257913247362251e-05,
+      "loss": 1.2006,
+      "step": 1970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3254396248534585e-05,
+      "loss": 1.1864,
+      "step": 1971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3250879249706915e-05,
+      "loss": 1.1649,
+      "step": 1972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3247362250879249e-05,
+      "loss": 1.1896,
+      "step": 1973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3243845252051584e-05,
+      "loss": 1.2718,
+      "step": 1974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3240328253223914e-05,
+      "loss": 1.2216,
+      "step": 1975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3236811254396248e-05,
+      "loss": 1.2651,
+      "step": 1976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3233294255568581e-05,
+      "loss": 1.231,
+      "step": 1977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3229777256740913e-05,
+      "loss": 1.236,
+      "step": 1978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3226260257913246e-05,
+      "loss": 1.2362,
+      "step": 1979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.322274325908558e-05,
+      "loss": 1.2319,
+      "step": 1980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3219226260257912e-05,
+      "loss": 1.2182,
+      "step": 1981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3215709261430245e-05,
+      "loss": 1.1932,
+      "step": 1982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3212192262602579e-05,
+      "loss": 1.1597,
+      "step": 1983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3208675263774912e-05,
+      "loss": 1.2216,
+      "step": 1984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3205158264947244e-05,
+      "loss": 1.2597,
+      "step": 1985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3201641266119578e-05,
+      "loss": 1.2491,
+      "step": 1986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3198124267291911e-05,
+      "loss": 1.177,
+      "step": 1987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3194607268464243e-05,
+      "loss": 1.1986,
+      "step": 1988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3191090269636577e-05,
+      "loss": 1.2567,
+      "step": 1989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.318757327080891e-05,
+      "loss": 1.2589,
+      "step": 1990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3184056271981242e-05,
+      "loss": 1.1817,
+      "step": 1991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3180539273153575e-05,
+      "loss": 1.2126,
+      "step": 1992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3177022274325909e-05,
+      "loss": 1.2656,
+      "step": 1993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.317350527549824e-05,
+      "loss": 1.1892,
+      "step": 1994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3169988276670574e-05,
+      "loss": 1.2405,
+      "step": 1995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3166471277842908e-05,
+      "loss": 1.2038,
+      "step": 1996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.316295427901524e-05,
+      "loss": 1.2192,
+      "step": 1997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3159437280187573e-05,
+      "loss": 1.2051,
+      "step": 1998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3155920281359907e-05,
+      "loss": 1.255,
+      "step": 1999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.315240328253224e-05,
+      "loss": 1.2248,
+      "step": 2000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3148886283704572e-05,
+      "loss": 1.209,
+      "step": 2001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3145369284876906e-05,
+      "loss": 1.2337,
+      "step": 2002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3141852286049239e-05,
+      "loss": 1.1905,
+      "step": 2003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3138335287221571e-05,
+      "loss": 1.1893,
+      "step": 2004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3134818288393904e-05,
+      "loss": 1.2038,
+      "step": 2005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3131301289566238e-05,
+      "loss": 1.2288,
+      "step": 2006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3127784290738568e-05,
+      "loss": 1.2317,
+      "step": 2007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3124267291910902e-05,
+      "loss": 1.2391,
+      "step": 2008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3120750293083235e-05,
+      "loss": 1.1933,
+      "step": 2009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3117233294255567e-05,
+      "loss": 1.1992,
+      "step": 2010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.31137162954279e-05,
+      "loss": 1.2594,
+      "step": 2011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3110199296600234e-05,
+      "loss": 1.2282,
+      "step": 2012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3106682297772566e-05,
+      "loss": 1.1743,
+      "step": 2013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.31031652989449e-05,
+      "loss": 1.1926,
+      "step": 2014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3099648300117233e-05,
+      "loss": 1.2164,
+      "step": 2015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3096131301289565e-05,
+      "loss": 1.253,
+      "step": 2016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3092614302461898e-05,
+      "loss": 1.2754,
+      "step": 2017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3089097303634232e-05,
+      "loss": 1.2148,
+      "step": 2018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3085580304806565e-05,
+      "loss": 1.1891,
+      "step": 2019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3082063305978897e-05,
+      "loss": 1.2122,
+      "step": 2020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.307854630715123e-05,
+      "loss": 1.2298,
+      "step": 2021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3075029308323564e-05,
+      "loss": 1.2055,
+      "step": 2022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3071512309495896e-05,
+      "loss": 1.2023,
+      "step": 2023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.306799531066823e-05,
+      "loss": 1.185,
+      "step": 2024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3064478311840563e-05,
+      "loss": 1.2418,
+      "step": 2025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3060961313012895e-05,
+      "loss": 1.2047,
+      "step": 2026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3057444314185228e-05,
+      "loss": 1.2001,
+      "step": 2027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3053927315357562e-05,
+      "loss": 1.2232,
+      "step": 2028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3050410316529894e-05,
+      "loss": 1.2191,
+      "step": 2029
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3046893317702227e-05,
+      "loss": 1.2237,
+      "step": 2030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3043376318874561e-05,
+      "loss": 1.212,
+      "step": 2031
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3039859320046893e-05,
+      "loss": 1.2472,
+      "step": 2032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3036342321219226e-05,
+      "loss": 1.2226,
+      "step": 2033
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.303282532239156e-05,
+      "loss": 1.252,
+      "step": 2034
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3029308323563893e-05,
+      "loss": 1.2119,
+      "step": 2035
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3025791324736225e-05,
+      "loss": 1.2242,
+      "step": 2036
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3022274325908559e-05,
+      "loss": 1.1766,
+      "step": 2037
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3018757327080892e-05,
+      "loss": 1.1575,
+      "step": 2038
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3015240328253224e-05,
+      "loss": 1.2064,
+      "step": 2039
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3011723329425558e-05,
+      "loss": 1.1466,
+      "step": 2040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3008206330597891e-05,
+      "loss": 1.2172,
+      "step": 2041
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3004689331770221e-05,
+      "loss": 1.1389,
+      "step": 2042
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3001172332942555e-05,
+      "loss": 1.2635,
+      "step": 2043
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2997655334114888e-05,
+      "loss": 1.2286,
+      "step": 2044
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.299413833528722e-05,
+      "loss": 1.2112,
+      "step": 2045
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2990621336459554e-05,
+      "loss": 1.1804,
+      "step": 2046
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2987104337631887e-05,
+      "loss": 1.2334,
+      "step": 2047
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2983587338804219e-05,
+      "loss": 1.2006,
+      "step": 2048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2980070339976552e-05,
+      "loss": 1.1265,
+      "step": 2049
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2976553341148886e-05,
+      "loss": 1.2109,
+      "step": 2050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2973036342321218e-05,
+      "loss": 1.1794,
+      "step": 2051
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2969519343493551e-05,
+      "loss": 1.2103,
+      "step": 2052
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2966002344665885e-05,
+      "loss": 1.243,
+      "step": 2053
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2962485345838218e-05,
+      "loss": 1.2368,
+      "step": 2054
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.295896834701055e-05,
+      "loss": 1.24,
+      "step": 2055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2955451348182884e-05,
+      "loss": 1.2259,
+      "step": 2056
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2951934349355217e-05,
+      "loss": 1.2465,
+      "step": 2057
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2948417350527549e-05,
+      "loss": 1.2475,
+      "step": 2058
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2944900351699883e-05,
+      "loss": 1.1879,
+      "step": 2059
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2941383352872216e-05,
+      "loss": 1.2556,
+      "step": 2060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2937866354044548e-05,
+      "loss": 1.2405,
+      "step": 2061
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2934349355216881e-05,
+      "loss": 1.1979,
+      "step": 2062
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2930832356389215e-05,
+      "loss": 1.2097,
+      "step": 2063
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2927315357561547e-05,
+      "loss": 1.2662,
+      "step": 2064
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.292379835873388e-05,
+      "loss": 1.1876,
+      "step": 2065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2920281359906214e-05,
+      "loss": 1.2376,
+      "step": 2066
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2916764361078546e-05,
+      "loss": 1.1897,
+      "step": 2067
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.291324736225088e-05,
+      "loss": 1.1989,
+      "step": 2068
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2909730363423213e-05,
+      "loss": 1.1837,
+      "step": 2069
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2906213364595546e-05,
+      "loss": 1.2327,
+      "step": 2070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2902696365767878e-05,
+      "loss": 1.1924,
+      "step": 2071
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2899179366940212e-05,
+      "loss": 1.239,
+      "step": 2072
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2895662368112545e-05,
+      "loss": 1.187,
+      "step": 2073
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2892145369284877e-05,
+      "loss": 1.2766,
+      "step": 2074
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.288862837045721e-05,
+      "loss": 1.1857,
+      "step": 2075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2885111371629544e-05,
+      "loss": 1.2284,
+      "step": 2076
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2881594372801874e-05,
+      "loss": 1.2357,
+      "step": 2077
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2878077373974208e-05,
+      "loss": 1.2294,
+      "step": 2078
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2874560375146541e-05,
+      "loss": 1.2291,
+      "step": 2079
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2871043376318873e-05,
+      "loss": 1.2088,
+      "step": 2080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2867526377491207e-05,
+      "loss": 1.1814,
+      "step": 2081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.286400937866354e-05,
+      "loss": 1.193,
+      "step": 2082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2860492379835872e-05,
+      "loss": 1.2052,
+      "step": 2083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2856975381008205e-05,
+      "loss": 1.234,
+      "step": 2084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2853458382180539e-05,
+      "loss": 1.2265,
+      "step": 2085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.284994138335287e-05,
+      "loss": 1.2027,
+      "step": 2086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2846424384525204e-05,
+      "loss": 1.2111,
+      "step": 2087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2842907385697538e-05,
+      "loss": 1.2068,
+      "step": 2088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2839390386869871e-05,
+      "loss": 1.2512,
+      "step": 2089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2835873388042203e-05,
+      "loss": 1.2176,
+      "step": 2090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2832356389214537e-05,
+      "loss": 1.1487,
+      "step": 2091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.282883939038687e-05,
+      "loss": 1.2284,
+      "step": 2092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2825322391559202e-05,
+      "loss": 1.259,
+      "step": 2093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2821805392731536e-05,
+      "loss": 1.1901,
+      "step": 2094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.281828839390387e-05,
+      "loss": 1.1748,
+      "step": 2095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2814771395076201e-05,
+      "loss": 1.1885,
+      "step": 2096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2811254396248535e-05,
+      "loss": 1.22,
+      "step": 2097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2807737397420868e-05,
+      "loss": 1.1821,
+      "step": 2098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.28042203985932e-05,
+      "loss": 1.2135,
+      "step": 2099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2800703399765533e-05,
+      "loss": 1.1703,
+      "step": 2100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2797186400937867e-05,
+      "loss": 1.1776,
+      "step": 2101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2793669402110199e-05,
+      "loss": 1.2032,
+      "step": 2102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2790152403282532e-05,
+      "loss": 1.2185,
+      "step": 2103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2786635404454866e-05,
+      "loss": 1.2152,
+      "step": 2104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2783118405627198e-05,
+      "loss": 1.2413,
+      "step": 2105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2779601406799531e-05,
+      "loss": 1.2181,
+      "step": 2106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2776084407971865e-05,
+      "loss": 1.1901,
+      "step": 2107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2772567409144198e-05,
+      "loss": 1.2358,
+      "step": 2108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.276905041031653e-05,
+      "loss": 1.224,
+      "step": 2109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2765533411488864e-05,
+      "loss": 1.2041,
+      "step": 2110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2762016412661197e-05,
+      "loss": 1.1962,
+      "step": 2111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2758499413833527e-05,
+      "loss": 1.2136,
+      "step": 2112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.275498241500586e-05,
+      "loss": 1.235,
+      "step": 2113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2751465416178194e-05,
+      "loss": 1.2344,
+      "step": 2114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2747948417350526e-05,
+      "loss": 1.2255,
+      "step": 2115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.274443141852286e-05,
+      "loss": 1.2763,
+      "step": 2116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2740914419695193e-05,
+      "loss": 1.1784,
+      "step": 2117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2737397420867525e-05,
+      "loss": 1.2433,
+      "step": 2118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2733880422039858e-05,
+      "loss": 1.2533,
+      "step": 2119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2730363423212192e-05,
+      "loss": 1.2452,
+      "step": 2120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2726846424384524e-05,
+      "loss": 1.1999,
+      "step": 2121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2723329425556857e-05,
+      "loss": 1.2263,
+      "step": 2122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2719812426729191e-05,
+      "loss": 1.1704,
+      "step": 2123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2716295427901524e-05,
+      "loss": 1.1729,
+      "step": 2124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2712778429073856e-05,
+      "loss": 1.2284,
+      "step": 2125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.270926143024619e-05,
+      "loss": 1.2183,
+      "step": 2126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2705744431418523e-05,
+      "loss": 1.2646,
+      "step": 2127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2702227432590855e-05,
+      "loss": 1.2217,
+      "step": 2128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2698710433763189e-05,
+      "loss": 1.2428,
+      "step": 2129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2695193434935522e-05,
+      "loss": 1.1794,
+      "step": 2130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2691676436107854e-05,
+      "loss": 1.118,
+      "step": 2131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2688159437280188e-05,
+      "loss": 1.2201,
+      "step": 2132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2684642438452521e-05,
+      "loss": 1.2217,
+      "step": 2133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2681125439624853e-05,
+      "loss": 1.2575,
+      "step": 2134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2677608440797186e-05,
+      "loss": 1.2429,
+      "step": 2135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.267409144196952e-05,
+      "loss": 1.16,
+      "step": 2136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2670574443141852e-05,
+      "loss": 1.2594,
+      "step": 2137
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2667057444314185e-05,
+      "loss": 1.215,
+      "step": 2138
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2663540445486519e-05,
+      "loss": 1.1889,
+      "step": 2139
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.266002344665885e-05,
+      "loss": 1.1797,
+      "step": 2140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2656506447831184e-05,
+      "loss": 1.246,
+      "step": 2141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2652989449003518e-05,
+      "loss": 1.2205,
+      "step": 2142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2649472450175851e-05,
+      "loss": 1.2247,
+      "step": 2143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2645955451348181e-05,
+      "loss": 1.2386,
+      "step": 2144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2642438452520517e-05,
+      "loss": 1.1771,
+      "step": 2145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.263892145369285e-05,
+      "loss": 1.2271,
+      "step": 2146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.263540445486518e-05,
+      "loss": 1.2118,
+      "step": 2147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2631887456037514e-05,
+      "loss": 1.2132,
+      "step": 2148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2628370457209847e-05,
+      "loss": 1.1952,
+      "step": 2149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2624853458382179e-05,
+      "loss": 1.2036,
+      "step": 2150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2621336459554513e-05,
+      "loss": 1.1786,
+      "step": 2151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2617819460726846e-05,
+      "loss": 1.2196,
+      "step": 2152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2614302461899178e-05,
+      "loss": 1.2213,
+      "step": 2153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2610785463071512e-05,
+      "loss": 1.1862,
+      "step": 2154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2607268464243845e-05,
+      "loss": 1.2372,
+      "step": 2155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2603751465416177e-05,
+      "loss": 1.1867,
+      "step": 2156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.260023446658851e-05,
+      "loss": 1.2192,
+      "step": 2157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2596717467760844e-05,
+      "loss": 1.2797,
+      "step": 2158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2593200468933177e-05,
+      "loss": 1.1766,
+      "step": 2159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.258968347010551e-05,
+      "loss": 1.1666,
+      "step": 2160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2586166471277843e-05,
+      "loss": 1.2031,
+      "step": 2161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2582649472450176e-05,
+      "loss": 1.1801,
+      "step": 2162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2579132473622508e-05,
+      "loss": 1.2002,
+      "step": 2163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2575615474794842e-05,
+      "loss": 1.2804,
+      "step": 2164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2572098475967175e-05,
+      "loss": 1.1876,
+      "step": 2165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2568581477139507e-05,
+      "loss": 1.1699,
+      "step": 2166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.256506447831184e-05,
+      "loss": 1.2579,
+      "step": 2167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2561547479484174e-05,
+      "loss": 1.1625,
+      "step": 2168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2558030480656506e-05,
+      "loss": 1.2263,
+      "step": 2169
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.255451348182884e-05,
+      "loss": 1.2462,
+      "step": 2170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2550996483001173e-05,
+      "loss": 1.2065,
+      "step": 2171
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2547479484173505e-05,
+      "loss": 1.2393,
+      "step": 2172
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2543962485345838e-05,
+      "loss": 1.2457,
+      "step": 2173
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2540445486518172e-05,
+      "loss": 1.2074,
+      "step": 2174
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2536928487690504e-05,
+      "loss": 1.1832,
+      "step": 2175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2533411488862837e-05,
+      "loss": 1.219,
+      "step": 2176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.252989449003517e-05,
+      "loss": 1.2635,
+      "step": 2177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2526377491207504e-05,
+      "loss": 1.2363,
+      "step": 2178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2522860492379834e-05,
+      "loss": 1.2441,
+      "step": 2179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.251934349355217e-05,
+      "loss": 1.1859,
+      "step": 2180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2515826494724503e-05,
+      "loss": 1.1949,
+      "step": 2181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2512309495896833e-05,
+      "loss": 1.2125,
+      "step": 2182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2508792497069167e-05,
+      "loss": 1.1783,
+      "step": 2183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.25052754982415e-05,
+      "loss": 1.2147,
+      "step": 2184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2501758499413832e-05,
+      "loss": 1.2228,
+      "step": 2185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2498241500586166e-05,
+      "loss": 1.2071,
+      "step": 2186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.24947245017585e-05,
+      "loss": 1.2096,
+      "step": 2187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2491207502930831e-05,
+      "loss": 1.1859,
+      "step": 2188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2487690504103165e-05,
+      "loss": 1.2074,
+      "step": 2189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2484173505275498e-05,
+      "loss": 1.1805,
+      "step": 2190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.248065650644783e-05,
+      "loss": 1.2424,
+      "step": 2191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2477139507620163e-05,
+      "loss": 1.2268,
+      "step": 2192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2473622508792497e-05,
+      "loss": 1.2341,
+      "step": 2193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2470105509964829e-05,
+      "loss": 1.2137,
+      "step": 2194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2466588511137162e-05,
+      "loss": 1.1779,
+      "step": 2195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2463071512309496e-05,
+      "loss": 1.1971,
+      "step": 2196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.245955451348183e-05,
+      "loss": 1.1975,
+      "step": 2197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2456037514654161e-05,
+      "loss": 1.244,
+      "step": 2198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2452520515826495e-05,
+      "loss": 1.2217,
+      "step": 2199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2449003516998828e-05,
+      "loss": 1.1651,
+      "step": 2200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.244548651817116e-05,
+      "loss": 1.2197,
+      "step": 2201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2441969519343494e-05,
+      "loss": 1.2671,
+      "step": 2202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2438452520515827e-05,
+      "loss": 1.2739,
+      "step": 2203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2434935521688159e-05,
+      "loss": 1.2585,
+      "step": 2204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2431418522860492e-05,
+      "loss": 1.2657,
+      "step": 2205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2427901524032826e-05,
+      "loss": 1.2268,
+      "step": 2206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2424384525205158e-05,
+      "loss": 1.2002,
+      "step": 2207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2420867526377491e-05,
+      "loss": 1.2038,
+      "step": 2208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2417350527549825e-05,
+      "loss": 1.2291,
+      "step": 2209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2413833528722157e-05,
+      "loss": 1.2397,
+      "step": 2210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.241031652989449e-05,
+      "loss": 1.2357,
+      "step": 2211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2406799531066824e-05,
+      "loss": 1.215,
+      "step": 2212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2403282532239157e-05,
+      "loss": 1.2083,
+      "step": 2213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2399765533411487e-05,
+      "loss": 1.2343,
+      "step": 2214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2396248534583823e-05,
+      "loss": 1.2235,
+      "step": 2215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2392731535756156e-05,
+      "loss": 1.215,
+      "step": 2216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2389214536928486e-05,
+      "loss": 1.1713,
+      "step": 2217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.238569753810082e-05,
+      "loss": 1.21,
+      "step": 2218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2382180539273153e-05,
+      "loss": 1.2131,
+      "step": 2219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2378663540445485e-05,
+      "loss": 1.2508,
+      "step": 2220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2375146541617819e-05,
+      "loss": 1.2008,
+      "step": 2221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2371629542790152e-05,
+      "loss": 1.2274,
+      "step": 2222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2368112543962484e-05,
+      "loss": 1.2536,
+      "step": 2223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2364595545134818e-05,
+      "loss": 1.2324,
+      "step": 2224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2361078546307151e-05,
+      "loss": 1.2509,
+      "step": 2225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2357561547479483e-05,
+      "loss": 1.2459,
+      "step": 2226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2354044548651816e-05,
+      "loss": 1.179,
+      "step": 2227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.235052754982415e-05,
+      "loss": 1.2232,
+      "step": 2228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2347010550996482e-05,
+      "loss": 1.1759,
+      "step": 2229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2343493552168815e-05,
+      "loss": 1.1988,
+      "step": 2230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2339976553341149e-05,
+      "loss": 1.2548,
+      "step": 2231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2336459554513482e-05,
+      "loss": 1.2339,
+      "step": 2232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2332942555685814e-05,
+      "loss": 1.2594,
+      "step": 2233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2329425556858148e-05,
+      "loss": 1.2062,
+      "step": 2234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2325908558030481e-05,
+      "loss": 1.2161,
+      "step": 2235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2322391559202813e-05,
+      "loss": 1.1799,
+      "step": 2236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2318874560375147e-05,
+      "loss": 1.2545,
+      "step": 2237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.231535756154748e-05,
+      "loss": 1.1718,
+      "step": 2238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2311840562719812e-05,
+      "loss": 1.2226,
+      "step": 2239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2308323563892145e-05,
+      "loss": 1.1798,
+      "step": 2240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2304806565064479e-05,
+      "loss": 1.2325,
+      "step": 2241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.230128956623681e-05,
+      "loss": 1.2065,
+      "step": 2242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2297772567409144e-05,
+      "loss": 1.2253,
+      "step": 2243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2294255568581478e-05,
+      "loss": 1.2016,
+      "step": 2244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.229073856975381e-05,
+      "loss": 1.1764,
+      "step": 2245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2287221570926143e-05,
+      "loss": 1.1903,
+      "step": 2246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2283704572098477e-05,
+      "loss": 1.2284,
+      "step": 2247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.228018757327081e-05,
+      "loss": 1.2186,
+      "step": 2248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.227667057444314e-05,
+      "loss": 1.2147,
+      "step": 2249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2273153575615474e-05,
+      "loss": 1.2278,
+      "step": 2250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.226963657678781e-05,
+      "loss": 1.2218,
+      "step": 2251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.226611957796014e-05,
+      "loss": 1.2258,
+      "step": 2252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2262602579132473e-05,
+      "loss": 1.2139,
+      "step": 2253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2259085580304806e-05,
+      "loss": 1.2149,
+      "step": 2254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2255568581477138e-05,
+      "loss": 1.2178,
+      "step": 2255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2252051582649472e-05,
+      "loss": 1.1479,
+      "step": 2256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2248534583821805e-05,
+      "loss": 1.2278,
+      "step": 2257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2245017584994137e-05,
+      "loss": 1.2184,
+      "step": 2258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.224150058616647e-05,
+      "loss": 1.1996,
+      "step": 2259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2237983587338804e-05,
+      "loss": 1.1327,
+      "step": 2260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2234466588511136e-05,
+      "loss": 1.1867,
+      "step": 2261
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.223094958968347e-05,
+      "loss": 1.2265,
+      "step": 2262
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2227432590855803e-05,
+      "loss": 1.229,
+      "step": 2263
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2223915592028135e-05,
+      "loss": 1.188,
+      "step": 2264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2220398593200468e-05,
+      "loss": 1.1784,
+      "step": 2265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2216881594372802e-05,
+      "loss": 1.2367,
+      "step": 2266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2213364595545135e-05,
+      "loss": 1.2636,
+      "step": 2267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2209847596717467e-05,
+      "loss": 1.1718,
+      "step": 2268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.22063305978898e-05,
+      "loss": 1.253,
+      "step": 2269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2202813599062134e-05,
+      "loss": 1.229,
+      "step": 2270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2199296600234466e-05,
+      "loss": 1.2032,
+      "step": 2271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.21957796014068e-05,
+      "loss": 1.225,
+      "step": 2272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2192262602579133e-05,
+      "loss": 1.2215,
+      "step": 2273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2188745603751465e-05,
+      "loss": 1.2465,
+      "step": 2274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2185228604923799e-05,
+      "loss": 1.2451,
+      "step": 2275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2181711606096132e-05,
+      "loss": 1.251,
+      "step": 2276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2178194607268464e-05,
+      "loss": 1.165,
+      "step": 2277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2174677608440797e-05,
+      "loss": 1.1654,
+      "step": 2278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2171160609613131e-05,
+      "loss": 1.1686,
+      "step": 2279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2167643610785463e-05,
+      "loss": 1.1878,
+      "step": 2280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2164126611957796e-05,
+      "loss": 1.2237,
+      "step": 2281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.216060961313013e-05,
+      "loss": 1.2133,
+      "step": 2282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.215709261430246e-05,
+      "loss": 1.2236,
+      "step": 2283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2153575615474793e-05,
+      "loss": 1.2471,
+      "step": 2284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2150058616647127e-05,
+      "loss": 1.2281,
+      "step": 2285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2146541617819462e-05,
+      "loss": 1.1952,
+      "step": 2286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2143024618991792e-05,
+      "loss": 1.2214,
+      "step": 2287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2139507620164126e-05,
+      "loss": 1.209,
+      "step": 2288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.213599062133646e-05,
+      "loss": 1.2385,
+      "step": 2289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2132473622508791e-05,
+      "loss": 1.2154,
+      "step": 2290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2128956623681125e-05,
+      "loss": 1.2335,
+      "step": 2291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2125439624853458e-05,
+      "loss": 1.2318,
+      "step": 2292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.212192262602579e-05,
+      "loss": 1.2062,
+      "step": 2293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2118405627198124e-05,
+      "loss": 1.2292,
+      "step": 2294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2114888628370457e-05,
+      "loss": 1.2348,
+      "step": 2295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2111371629542789e-05,
+      "loss": 1.1781,
+      "step": 2296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2107854630715122e-05,
+      "loss": 1.2116,
+      "step": 2297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2104337631887456e-05,
+      "loss": 1.2787,
+      "step": 2298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2100820633059788e-05,
+      "loss": 1.201,
+      "step": 2299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2097303634232121e-05,
+      "loss": 1.2171,
+      "step": 2300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2093786635404455e-05,
+      "loss": 1.2303,
+      "step": 2301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2090269636576788e-05,
+      "loss": 1.2155,
+      "step": 2302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.208675263774912e-05,
+      "loss": 1.2049,
+      "step": 2303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2083235638921454e-05,
+      "loss": 1.206,
+      "step": 2304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2079718640093787e-05,
+      "loss": 1.2143,
+      "step": 2305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2076201641266119e-05,
+      "loss": 1.2249,
+      "step": 2306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2072684642438453e-05,
+      "loss": 1.2115,
+      "step": 2307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2069167643610786e-05,
+      "loss": 1.1846,
+      "step": 2308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2065650644783118e-05,
+      "loss": 1.1764,
+      "step": 2309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2062133645955452e-05,
+      "loss": 1.2162,
+      "step": 2310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2058616647127785e-05,
+      "loss": 1.226,
+      "step": 2311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2055099648300117e-05,
+      "loss": 1.2046,
+      "step": 2312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.205158264947245e-05,
+      "loss": 1.1668,
+      "step": 2313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2048065650644784e-05,
+      "loss": 1.2119,
+      "step": 2314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2044548651817116e-05,
+      "loss": 1.2261,
+      "step": 2315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.204103165298945e-05,
+      "loss": 1.1855,
+      "step": 2316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2037514654161783e-05,
+      "loss": 1.2112,
+      "step": 2317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2033997655334113e-05,
+      "loss": 1.1966,
+      "step": 2318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2030480656506446e-05,
+      "loss": 1.2119,
+      "step": 2319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.202696365767878e-05,
+      "loss": 1.2038,
+      "step": 2320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2023446658851114e-05,
+      "loss": 1.2889,
+      "step": 2321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2019929660023445e-05,
+      "loss": 1.2334,
+      "step": 2322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2016412661195779e-05,
+      "loss": 1.2213,
+      "step": 2323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2012895662368112e-05,
+      "loss": 1.222,
+      "step": 2324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2009378663540444e-05,
+      "loss": 1.2185,
+      "step": 2325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2005861664712778e-05,
+      "loss": 1.2319,
+      "step": 2326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2002344665885111e-05,
+      "loss": 1.2376,
+      "step": 2327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1998827667057443e-05,
+      "loss": 1.2381,
+      "step": 2328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1995310668229777e-05,
+      "loss": 1.2177,
+      "step": 2329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.199179366940211e-05,
+      "loss": 1.2132,
+      "step": 2330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1988276670574442e-05,
+      "loss": 1.1789,
+      "step": 2331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1984759671746776e-05,
+      "loss": 1.1751,
+      "step": 2332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1981242672919109e-05,
+      "loss": 1.2336,
+      "step": 2333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1977725674091441e-05,
+      "loss": 1.2323,
+      "step": 2334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1974208675263774e-05,
+      "loss": 1.2538,
+      "step": 2335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1970691676436108e-05,
+      "loss": 1.2132,
+      "step": 2336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1967174677608441e-05,
+      "loss": 1.1817,
+      "step": 2337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1963657678780773e-05,
+      "loss": 1.183,
+      "step": 2338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1960140679953107e-05,
+      "loss": 1.2202,
+      "step": 2339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.195662368112544e-05,
+      "loss": 1.1611,
+      "step": 2340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1953106682297772e-05,
+      "loss": 1.2355,
+      "step": 2341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1949589683470106e-05,
+      "loss": 1.2083,
+      "step": 2342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.194607268464244e-05,
+      "loss": 1.245,
+      "step": 2343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1942555685814771e-05,
+      "loss": 1.2361,
+      "step": 2344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1939038686987105e-05,
+      "loss": 1.253,
+      "step": 2345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1935521688159438e-05,
+      "loss": 1.2206,
+      "step": 2346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.193200468933177e-05,
+      "loss": 1.1951,
+      "step": 2347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1928487690504103e-05,
+      "loss": 1.2637,
+      "step": 2348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1924970691676437e-05,
+      "loss": 1.1656,
+      "step": 2349
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1921453692848769e-05,
+      "loss": 1.2135,
+      "step": 2350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1917936694021102e-05,
+      "loss": 1.2743,
+      "step": 2351
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1914419695193436e-05,
+      "loss": 1.2146,
+      "step": 2352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1910902696365766e-05,
+      "loss": 1.2232,
+      "step": 2353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.19073856975381e-05,
+      "loss": 1.1663,
+      "step": 2354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1903868698710433e-05,
+      "loss": 1.1835,
+      "step": 2355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1900351699882767e-05,
+      "loss": 1.2389,
+      "step": 2356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1896834701055098e-05,
+      "loss": 1.2216,
+      "step": 2357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1893317702227432e-05,
+      "loss": 1.2396,
+      "step": 2358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1889800703399765e-05,
+      "loss": 1.2112,
+      "step": 2359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1886283704572097e-05,
+      "loss": 1.1971,
+      "step": 2360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.188276670574443e-05,
+      "loss": 1.2206,
+      "step": 2361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1879249706916764e-05,
+      "loss": 1.2287,
+      "step": 2362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1875732708089096e-05,
+      "loss": 1.1893,
+      "step": 2363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.187221570926143e-05,
+      "loss": 1.2103,
+      "step": 2364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1868698710433763e-05,
+      "loss": 1.18,
+      "step": 2365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1865181711606095e-05,
+      "loss": 1.2202,
+      "step": 2366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1861664712778429e-05,
+      "loss": 1.1973,
+      "step": 2367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1858147713950762e-05,
+      "loss": 1.1813,
+      "step": 2368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1854630715123094e-05,
+      "loss": 1.1732,
+      "step": 2369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1851113716295427e-05,
+      "loss": 1.237,
+      "step": 2370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1847596717467761e-05,
+      "loss": 1.1929,
+      "step": 2371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1844079718640093e-05,
+      "loss": 1.1697,
+      "step": 2372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1840562719812426e-05,
+      "loss": 1.1895,
+      "step": 2373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.183704572098476e-05,
+      "loss": 1.2302,
+      "step": 2374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1833528722157093e-05,
+      "loss": 1.2571,
+      "step": 2375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1830011723329425e-05,
+      "loss": 1.2288,
+      "step": 2376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1826494724501759e-05,
+      "loss": 1.2421,
+      "step": 2377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1822977725674092e-05,
+      "loss": 1.2098,
+      "step": 2378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1819460726846424e-05,
+      "loss": 1.2195,
+      "step": 2379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1815943728018758e-05,
+      "loss": 1.2213,
+      "step": 2380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1812426729191091e-05,
+      "loss": 1.2416,
+      "step": 2381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1808909730363423e-05,
+      "loss": 1.2447,
+      "step": 2382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1805392731535756e-05,
+      "loss": 1.2336,
+      "step": 2383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.180187573270809e-05,
+      "loss": 1.2705,
+      "step": 2384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.179835873388042e-05,
+      "loss": 1.1578,
+      "step": 2385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1794841735052755e-05,
+      "loss": 1.1486,
+      "step": 2386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1791324736225089e-05,
+      "loss": 1.2257,
+      "step": 2387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1787807737397419e-05,
+      "loss": 1.2334,
+      "step": 2388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1784290738569753e-05,
+      "loss": 1.1879,
+      "step": 2389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1780773739742086e-05,
+      "loss": 1.1954,
+      "step": 2390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.177725674091442e-05,
+      "loss": 1.2174,
+      "step": 2391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1773739742086751e-05,
+      "loss": 1.2,
+      "step": 2392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1770222743259085e-05,
+      "loss": 1.2226,
+      "step": 2393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1766705744431418e-05,
+      "loss": 1.2159,
+      "step": 2394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.176318874560375e-05,
+      "loss": 1.2164,
+      "step": 2395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1759671746776084e-05,
+      "loss": 1.2146,
+      "step": 2396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1756154747948417e-05,
+      "loss": 1.1985,
+      "step": 2397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.175263774912075e-05,
+      "loss": 1.2689,
+      "step": 2398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1749120750293083e-05,
+      "loss": 1.204,
+      "step": 2399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1745603751465416e-05,
+      "loss": 1.2374,
+      "step": 2400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1742086752637748e-05,
+      "loss": 1.2317,
+      "step": 2401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1738569753810082e-05,
+      "loss": 1.1554,
+      "step": 2402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1735052754982415e-05,
+      "loss": 1.1644,
+      "step": 2403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1731535756154747e-05,
+      "loss": 1.2056,
+      "step": 2404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.172801875732708e-05,
+      "loss": 1.2016,
+      "step": 2405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1724501758499414e-05,
+      "loss": 1.1875,
+      "step": 2406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1720984759671746e-05,
+      "loss": 1.1956,
+      "step": 2407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.171746776084408e-05,
+      "loss": 1.1988,
+      "step": 2408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1713950762016413e-05,
+      "loss": 1.2198,
+      "step": 2409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1710433763188746e-05,
+      "loss": 1.2398,
+      "step": 2410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1706916764361078e-05,
+      "loss": 1.2321,
+      "step": 2411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1703399765533412e-05,
+      "loss": 1.1807,
+      "step": 2412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1699882766705745e-05,
+      "loss": 1.2446,
+      "step": 2413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1696365767878077e-05,
+      "loss": 1.2367,
+      "step": 2414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.169284876905041e-05,
+      "loss": 1.2371,
+      "step": 2415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1689331770222744e-05,
+      "loss": 1.2177,
+      "step": 2416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1685814771395076e-05,
+      "loss": 1.1892,
+      "step": 2417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.168229777256741e-05,
+      "loss": 1.218,
+      "step": 2418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1678780773739743e-05,
+      "loss": 1.2213,
+      "step": 2419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1675263774912073e-05,
+      "loss": 1.2009,
+      "step": 2420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1671746776084408e-05,
+      "loss": 1.1939,
+      "step": 2421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1668229777256742e-05,
+      "loss": 1.1366,
+      "step": 2422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1664712778429072e-05,
+      "loss": 1.2337,
+      "step": 2423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1661195779601406e-05,
+      "loss": 1.2291,
+      "step": 2424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1657678780773739e-05,
+      "loss": 1.2281,
+      "step": 2425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1654161781946073e-05,
+      "loss": 1.175,
+      "step": 2426
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1650644783118404e-05,
+      "loss": 1.231,
+      "step": 2427
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1647127784290738e-05,
+      "loss": 1.2057,
+      "step": 2428
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1643610785463071e-05,
+      "loss": 1.1817,
+      "step": 2429
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1640093786635403e-05,
+      "loss": 1.2649,
+      "step": 2430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1636576787807737e-05,
+      "loss": 1.1873,
+      "step": 2431
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.163305978898007e-05,
+      "loss": 1.2201,
+      "step": 2432
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1629542790152402e-05,
+      "loss": 1.2656,
+      "step": 2433
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1626025791324736e-05,
+      "loss": 1.2158,
+      "step": 2434
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.162250879249707e-05,
+      "loss": 1.2224,
+      "step": 2435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1618991793669401e-05,
+      "loss": 1.1933,
+      "step": 2436
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1615474794841735e-05,
+      "loss": 1.2202,
+      "step": 2437
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1611957796014068e-05,
+      "loss": 1.2007,
+      "step": 2438
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.16084407971864e-05,
+      "loss": 1.234,
+      "step": 2439
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1604923798358733e-05,
+      "loss": 1.1707,
+      "step": 2440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1601406799531067e-05,
+      "loss": 1.2242,
+      "step": 2441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1597889800703399e-05,
+      "loss": 1.2063,
+      "step": 2442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1594372801875732e-05,
+      "loss": 1.2377,
+      "step": 2443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1590855803048066e-05,
+      "loss": 1.2301,
+      "step": 2444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.15873388042204e-05,
+      "loss": 1.2121,
+      "step": 2445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1583821805392731e-05,
+      "loss": 1.1945,
+      "step": 2446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1580304806565065e-05,
+      "loss": 1.2263,
+      "step": 2447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1576787807737398e-05,
+      "loss": 1.1555,
+      "step": 2448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.157327080890973e-05,
+      "loss": 1.1979,
+      "step": 2449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1569753810082064e-05,
+      "loss": 1.1763,
+      "step": 2450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1566236811254397e-05,
+      "loss": 1.2306,
+      "step": 2451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1562719812426729e-05,
+      "loss": 1.2388,
+      "step": 2452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1559202813599063e-05,
+      "loss": 1.2459,
+      "step": 2453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1555685814771396e-05,
+      "loss": 1.2262,
+      "step": 2454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1552168815943726e-05,
+      "loss": 1.1875,
+      "step": 2455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.154865181711606e-05,
+      "loss": 1.1834,
+      "step": 2456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1545134818288395e-05,
+      "loss": 1.2284,
+      "step": 2457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1541617819460725e-05,
+      "loss": 1.2447,
+      "step": 2458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1538100820633059e-05,
+      "loss": 1.1622,
+      "step": 2459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1534583821805392e-05,
+      "loss": 1.2004,
+      "step": 2460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1531066822977726e-05,
+      "loss": 1.1897,
+      "step": 2461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1527549824150057e-05,
+      "loss": 1.2733,
+      "step": 2462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1524032825322391e-05,
+      "loss": 1.244,
+      "step": 2463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1520515826494725e-05,
+      "loss": 1.23,
+      "step": 2464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1516998827667056e-05,
+      "loss": 1.1839,
+      "step": 2465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.151348182883939e-05,
+      "loss": 1.2113,
+      "step": 2466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1509964830011723e-05,
+      "loss": 1.1638,
+      "step": 2467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1506447831184055e-05,
+      "loss": 1.2389,
+      "step": 2468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1502930832356389e-05,
+      "loss": 1.2189,
+      "step": 2469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1499413833528722e-05,
+      "loss": 1.1632,
+      "step": 2470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1495896834701054e-05,
+      "loss": 1.1867,
+      "step": 2471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1492379835873388e-05,
+      "loss": 1.25,
+      "step": 2472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1488862837045721e-05,
+      "loss": 1.2261,
+      "step": 2473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1485345838218053e-05,
+      "loss": 1.2352,
+      "step": 2474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1481828839390386e-05,
+      "loss": 1.1823,
+      "step": 2475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.147831184056272e-05,
+      "loss": 1.177,
+      "step": 2476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1474794841735052e-05,
+      "loss": 1.1609,
+      "step": 2477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1471277842907385e-05,
+      "loss": 1.2315,
+      "step": 2478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1467760844079719e-05,
+      "loss": 1.2338,
+      "step": 2479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1464243845252052e-05,
+      "loss": 1.2288,
+      "step": 2480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1460726846424384e-05,
+      "loss": 1.22,
+      "step": 2481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1457209847596718e-05,
+      "loss": 1.1932,
+      "step": 2482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1453692848769051e-05,
+      "loss": 1.2269,
+      "step": 2483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1450175849941383e-05,
+      "loss": 1.1928,
+      "step": 2484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1446658851113717e-05,
+      "loss": 1.2156,
+      "step": 2485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.144314185228605e-05,
+      "loss": 1.1691,
+      "step": 2486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1439624853458382e-05,
+      "loss": 1.2152,
+      "step": 2487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1436107854630716e-05,
+      "loss": 1.196,
+      "step": 2488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1432590855803049e-05,
+      "loss": 1.2412,
+      "step": 2489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.142907385697538e-05,
+      "loss": 1.1891,
+      "step": 2490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1425556858147713e-05,
+      "loss": 1.2634,
+      "step": 2491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1422039859320048e-05,
+      "loss": 1.2263,
+      "step": 2492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1418522860492378e-05,
+      "loss": 1.2178,
+      "step": 2493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1415005861664712e-05,
+      "loss": 1.1609,
+      "step": 2494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1411488862837045e-05,
+      "loss": 1.1864,
+      "step": 2495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1407971864009377e-05,
+      "loss": 1.1816,
+      "step": 2496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.140445486518171e-05,
+      "loss": 1.2254,
+      "step": 2497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1400937866354044e-05,
+      "loss": 1.1802,
+      "step": 2498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1397420867526378e-05,
+      "loss": 1.2292,
+      "step": 2499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.139390386869871e-05,
+      "loss": 1.1785,
+      "step": 2500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1390386869871043e-05,
+      "loss": 1.2133,
+      "step": 2501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1386869871043376e-05,
+      "loss": 1.2143,
+      "step": 2502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1383352872215708e-05,
+      "loss": 1.241,
+      "step": 2503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1379835873388042e-05,
+      "loss": 1.2147,
+      "step": 2504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1376318874560375e-05,
+      "loss": 1.2117,
+      "step": 2505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1372801875732707e-05,
+      "loss": 1.2866,
+      "step": 2506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.136928487690504e-05,
+      "loss": 1.2379,
+      "step": 2507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1365767878077374e-05,
+      "loss": 1.2435,
+      "step": 2508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1362250879249706e-05,
+      "loss": 1.2326,
+      "step": 2509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.135873388042204e-05,
+      "loss": 1.2019,
+      "step": 2510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1355216881594373e-05,
+      "loss": 1.2559,
+      "step": 2511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1351699882766705e-05,
+      "loss": 1.1764,
+      "step": 2512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1348182883939038e-05,
+      "loss": 1.204,
+      "step": 2513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1344665885111372e-05,
+      "loss": 1.2181,
+      "step": 2514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1341148886283705e-05,
+      "loss": 1.2826,
+      "step": 2515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1337631887456037e-05,
+      "loss": 1.18,
+      "step": 2516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.133411488862837e-05,
+      "loss": 1.2416,
+      "step": 2517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1330597889800704e-05,
+      "loss": 1.2203,
+      "step": 2518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1327080890973036e-05,
+      "loss": 1.2528,
+      "step": 2519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.132356389214537e-05,
+      "loss": 1.2147,
+      "step": 2520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1320046893317703e-05,
+      "loss": 1.2163,
+      "step": 2521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1316529894490035e-05,
+      "loss": 1.1619,
+      "step": 2522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1313012895662369e-05,
+      "loss": 1.1982,
+      "step": 2523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1309495896834702e-05,
+      "loss": 1.2751,
+      "step": 2524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1305978898007032e-05,
+      "loss": 1.2597,
+      "step": 2525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1302461899179366e-05,
+      "loss": 1.1602,
+      "step": 2526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1298944900351701e-05,
+      "loss": 1.2073,
+      "step": 2527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1295427901524031e-05,
+      "loss": 1.2016,
+      "step": 2528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1291910902696365e-05,
+      "loss": 1.2001,
+      "step": 2529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1288393903868698e-05,
+      "loss": 1.1938,
+      "step": 2530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.128487690504103e-05,
+      "loss": 1.2346,
+      "step": 2531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1281359906213364e-05,
+      "loss": 1.233,
+      "step": 2532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1277842907385697e-05,
+      "loss": 1.2495,
+      "step": 2533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.127432590855803e-05,
+      "loss": 1.2089,
+      "step": 2534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1270808909730362e-05,
+      "loss": 1.2458,
+      "step": 2535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1267291910902696e-05,
+      "loss": 1.2252,
+      "step": 2536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.126377491207503e-05,
+      "loss": 1.2176,
+      "step": 2537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1260257913247361e-05,
+      "loss": 1.2082,
+      "step": 2538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1256740914419695e-05,
+      "loss": 1.2149,
+      "step": 2539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1253223915592028e-05,
+      "loss": 1.1978,
+      "step": 2540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.124970691676436e-05,
+      "loss": 1.2004,
+      "step": 2541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1246189917936694e-05,
+      "loss": 1.216,
+      "step": 2542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1242672919109027e-05,
+      "loss": 1.193,
+      "step": 2543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1239155920281359e-05,
+      "loss": 1.2279,
+      "step": 2544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1235638921453693e-05,
+      "loss": 1.1971,
+      "step": 2545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1232121922626026e-05,
+      "loss": 1.2801,
+      "step": 2546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1228604923798358e-05,
+      "loss": 1.1912,
+      "step": 2547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1225087924970691e-05,
+      "loss": 1.1834,
+      "step": 2548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1221570926143025e-05,
+      "loss": 1.2314,
+      "step": 2549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1218053927315358e-05,
+      "loss": 1.2622,
+      "step": 2550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.121453692848769e-05,
+      "loss": 1.2143,
+      "step": 2551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1211019929660024e-05,
+      "loss": 1.2633,
+      "step": 2552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1207502930832357e-05,
+      "loss": 1.2206,
+      "step": 2553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.120398593200469e-05,
+      "loss": 1.1982,
+      "step": 2554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1200468933177023e-05,
+      "loss": 1.2187,
+      "step": 2555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1196951934349356e-05,
+      "loss": 1.2168,
+      "step": 2556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1193434935521688e-05,
+      "loss": 1.1873,
+      "step": 2557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1189917936694022e-05,
+      "loss": 1.2355,
+      "step": 2558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1186400937866355e-05,
+      "loss": 1.1707,
+      "step": 2559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1182883939038685e-05,
+      "loss": 1.2183,
+      "step": 2560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1179366940211019e-05,
+      "loss": 1.2107,
+      "step": 2561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1175849941383352e-05,
+      "loss": 1.1446,
+      "step": 2562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1172332942555684e-05,
+      "loss": 1.2136,
+      "step": 2563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1168815943728018e-05,
+      "loss": 1.1893,
+      "step": 2564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1165298944900351e-05,
+      "loss": 1.2049,
+      "step": 2565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1161781946072683e-05,
+      "loss": 1.2056,
+      "step": 2566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1158264947245017e-05,
+      "loss": 1.2315,
+      "step": 2567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.115474794841735e-05,
+      "loss": 1.221,
+      "step": 2568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1151230949589684e-05,
+      "loss": 1.2204,
+      "step": 2569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1147713950762015e-05,
+      "loss": 1.2185,
+      "step": 2570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1144196951934349e-05,
+      "loss": 1.2024,
+      "step": 2571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1140679953106682e-05,
+      "loss": 1.177,
+      "step": 2572
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1137162954279014e-05,
+      "loss": 1.2126,
+      "step": 2573
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1133645955451348e-05,
+      "loss": 1.1607,
+      "step": 2574
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1130128956623681e-05,
+      "loss": 1.2469,
+      "step": 2575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1126611957796013e-05,
+      "loss": 1.2385,
+      "step": 2576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1123094958968347e-05,
+      "loss": 1.2067,
+      "step": 2577
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.111957796014068e-05,
+      "loss": 1.1888,
+      "step": 2578
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1116060961313012e-05,
+      "loss": 1.2049,
+      "step": 2579
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1112543962485346e-05,
+      "loss": 1.2158,
+      "step": 2580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1109026963657679e-05,
+      "loss": 1.2389,
+      "step": 2581
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1105509964830011e-05,
+      "loss": 1.217,
+      "step": 2582
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1101992966002344e-05,
+      "loss": 1.1763,
+      "step": 2583
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1098475967174678e-05,
+      "loss": 1.2375,
+      "step": 2584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.109495896834701e-05,
+      "loss": 1.1008,
+      "step": 2585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1091441969519343e-05,
+      "loss": 1.0658,
+      "step": 2586
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1087924970691677e-05,
+      "loss": 1.1057,
+      "step": 2587
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.108440797186401e-05,
+      "loss": 1.0239,
+      "step": 2588
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1080890973036342e-05,
+      "loss": 1.043,
+      "step": 2589
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1077373974208676e-05,
+      "loss": 1.01,
+      "step": 2590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.107385697538101e-05,
+      "loss": 1.0039,
+      "step": 2591
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1070339976553341e-05,
+      "loss": 1.0205,
+      "step": 2592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1066822977725675e-05,
+      "loss": 1.0073,
+      "step": 2593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1063305978898008e-05,
+      "loss": 1.0189,
+      "step": 2594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1059788980070338e-05,
+      "loss": 1.0417,
+      "step": 2595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1056271981242672e-05,
+      "loss": 1.0064,
+      "step": 2596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1052754982415005e-05,
+      "loss": 0.9953,
+      "step": 2597
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1049237983587337e-05,
+      "loss": 0.9374,
+      "step": 2598
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.104572098475967e-05,
+      "loss": 0.9611,
+      "step": 2599
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1042203985932004e-05,
+      "loss": 0.9612,
+      "step": 2600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1038686987104336e-05,
+      "loss": 0.9956,
+      "step": 2601
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.103516998827667e-05,
+      "loss": 1.0075,
+      "step": 2602
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1031652989449003e-05,
+      "loss": 1.0056,
+      "step": 2603
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1028135990621337e-05,
+      "loss": 0.9592,
+      "step": 2604
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1024618991793668e-05,
+      "loss": 0.969,
+      "step": 2605
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1021101992966002e-05,
+      "loss": 1.0084,
+      "step": 2606
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1017584994138335e-05,
+      "loss": 0.991,
+      "step": 2607
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1014067995310667e-05,
+      "loss": 0.9796,
+      "step": 2608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1010550996483e-05,
+      "loss": 0.9483,
+      "step": 2609
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1007033997655334e-05,
+      "loss": 0.9856,
+      "step": 2610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1003516998827666e-05,
+      "loss": 0.9763,
+      "step": 2611
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1e-05,
+      "loss": 0.9934,
+      "step": 2612
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0996483001172332e-05,
+      "loss": 1.0343,
+      "step": 2613
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0992966002344665e-05,
+      "loss": 0.9622,
+      "step": 2614
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0989449003516999e-05,
+      "loss": 0.9642,
+      "step": 2615
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.098593200468933e-05,
+      "loss": 0.9729,
+      "step": 2616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0982415005861664e-05,
+      "loss": 1.0196,
+      "step": 2617
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0978898007033997e-05,
+      "loss": 0.9748,
+      "step": 2618
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0975381008206331e-05,
+      "loss": 1.0003,
+      "step": 2619
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0971864009378663e-05,
+      "loss": 1.0,
+      "step": 2620
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0968347010550996e-05,
+      "loss": 0.9305,
+      "step": 2621
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.096483001172333e-05,
+      "loss": 0.9695,
+      "step": 2622
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0961313012895662e-05,
+      "loss": 0.9446,
+      "step": 2623
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0957796014067995e-05,
+      "loss": 1.0304,
+      "step": 2624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0954279015240329e-05,
+      "loss": 0.9348,
+      "step": 2625
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.095076201641266e-05,
+      "loss": 0.9917,
+      "step": 2626
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0947245017584994e-05,
+      "loss": 1.0248,
+      "step": 2627
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0943728018757328e-05,
+      "loss": 0.9462,
+      "step": 2628
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.094021101992966e-05,
+      "loss": 0.9502,
+      "step": 2629
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0936694021101993e-05,
+      "loss": 1.014,
+      "step": 2630
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0933177022274325e-05,
+      "loss": 1.0043,
+      "step": 2631
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0929660023446658e-05,
+      "loss": 0.9949,
+      "step": 2632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0926143024618992e-05,
+      "loss": 1.016,
+      "step": 2633
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0922626025791324e-05,
+      "loss": 0.9835,
+      "step": 2634
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0919109026963657e-05,
+      "loss": 0.9612,
+      "step": 2635
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.091559202813599e-05,
+      "loss": 0.9849,
+      "step": 2636
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0912075029308323e-05,
+      "loss": 0.9915,
+      "step": 2637
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0908558030480656e-05,
+      "loss": 0.9885,
+      "step": 2638
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.090504103165299e-05,
+      "loss": 0.9732,
+      "step": 2639
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0901524032825321e-05,
+      "loss": 0.9504,
+      "step": 2640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0898007033997655e-05,
+      "loss": 0.9725,
+      "step": 2641
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0894490035169989e-05,
+      "loss": 0.9509,
+      "step": 2642
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.089097303634232e-05,
+      "loss": 1.0249,
+      "step": 2643
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0887456037514654e-05,
+      "loss": 0.9932,
+      "step": 2644
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0883939038686987e-05,
+      "loss": 0.9735,
+      "step": 2645
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0880422039859321e-05,
+      "loss": 0.9752,
+      "step": 2646
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0876905041031653e-05,
+      "loss": 0.9611,
+      "step": 2647
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0873388042203985e-05,
+      "loss": 1.009,
+      "step": 2648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0869871043376318e-05,
+      "loss": 0.9673,
+      "step": 2649
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0866354044548652e-05,
+      "loss": 0.9432,
+      "step": 2650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0862837045720983e-05,
+      "loss": 1.0099,
+      "step": 2651
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0859320046893317e-05,
+      "loss": 0.9531,
+      "step": 2652
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.085580304806565e-05,
+      "loss": 0.9761,
+      "step": 2653
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0852286049237984e-05,
+      "loss": 1.0012,
+      "step": 2654
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0848769050410316e-05,
+      "loss": 0.9744,
+      "step": 2655
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.084525205158265e-05,
+      "loss": 0.9927,
+      "step": 2656
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0841735052754983e-05,
+      "loss": 0.9832,
+      "step": 2657
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0838218053927315e-05,
+      "loss": 0.9647,
+      "step": 2658
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0834701055099648e-05,
+      "loss": 0.9228,
+      "step": 2659
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0831184056271982e-05,
+      "loss": 0.9567,
+      "step": 2660
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0827667057444314e-05,
+      "loss": 1.026,
+      "step": 2661
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0824150058616647e-05,
+      "loss": 1.0056,
+      "step": 2662
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.082063305978898e-05,
+      "loss": 0.97,
+      "step": 2663
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0817116060961312e-05,
+      "loss": 1.0034,
+      "step": 2664
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0813599062133646e-05,
+      "loss": 0.9379,
+      "step": 2665
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0810082063305978e-05,
+      "loss": 0.9642,
+      "step": 2666
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0806565064478311e-05,
+      "loss": 0.9655,
+      "step": 2667
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0803048065650645e-05,
+      "loss": 0.9795,
+      "step": 2668
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0799531066822977e-05,
+      "loss": 0.9775,
+      "step": 2669
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.079601406799531e-05,
+      "loss": 0.9443,
+      "step": 2670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0792497069167644e-05,
+      "loss": 0.999,
+      "step": 2671
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0788980070339976e-05,
+      "loss": 0.9725,
+      "step": 2672
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0785463071512309e-05,
+      "loss": 0.9328,
+      "step": 2673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0781946072684643e-05,
+      "loss": 0.9996,
+      "step": 2674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0778429073856974e-05,
+      "loss": 1.0129,
+      "step": 2675
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0774912075029308e-05,
+      "loss": 0.9464,
+      "step": 2676
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0771395076201642e-05,
+      "loss": 0.9785,
+      "step": 2677
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0767878077373973e-05,
+      "loss": 0.9884,
+      "step": 2678
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0764361078546307e-05,
+      "loss": 0.9824,
+      "step": 2679
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.076084407971864e-05,
+      "loss": 0.9471,
+      "step": 2680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0757327080890974e-05,
+      "loss": 0.9216,
+      "step": 2681
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0753810082063306e-05,
+      "loss": 0.9545,
+      "step": 2682
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0750293083235638e-05,
+      "loss": 0.9863,
+      "step": 2683
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0746776084407971e-05,
+      "loss": 1.0105,
+      "step": 2684
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0743259085580305e-05,
+      "loss": 1.0172,
+      "step": 2685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0739742086752636e-05,
+      "loss": 1.0148,
+      "step": 2686
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.073622508792497e-05,
+      "loss": 0.9947,
+      "step": 2687
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0732708089097304e-05,
+      "loss": 0.9046,
+      "step": 2688
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0729191090269637e-05,
+      "loss": 0.9671,
+      "step": 2689
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0725674091441969e-05,
+      "loss": 1.0172,
+      "step": 2690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0722157092614302e-05,
+      "loss": 0.9888,
+      "step": 2691
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0718640093786636e-05,
+      "loss": 1.0366,
+      "step": 2692
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0715123094958968e-05,
+      "loss": 0.9786,
+      "step": 2693
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0711606096131301e-05,
+      "loss": 0.9773,
+      "step": 2694
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0708089097303635e-05,
+      "loss": 0.9543,
+      "step": 2695
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0704572098475967e-05,
+      "loss": 1.0028,
+      "step": 2696
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.07010550996483e-05,
+      "loss": 0.9788,
+      "step": 2697
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0697538100820634e-05,
+      "loss": 0.9626,
+      "step": 2698
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0694021101992966e-05,
+      "loss": 0.9816,
+      "step": 2699
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0690504103165299e-05,
+      "loss": 0.9684,
+      "step": 2700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.0686987104337631e-05,
+      "loss": 0.9914,
+      "step": 2701
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0683470105509964e-05,
+      "loss": 0.9644,
+      "step": 2702
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0679953106682298e-05,
+      "loss": 0.9728,
+      "step": 2703
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.067643610785463e-05,
+      "loss": 1.0286,
+      "step": 2704
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0672919109026963e-05,
+      "loss": 0.9925,
+      "step": 2705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0669402110199297e-05,
+      "loss": 0.9782,
+      "step": 2706
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0665885111371629e-05,
+      "loss": 0.9667,
+      "step": 2707
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0662368112543962e-05,
+      "loss": 1.0066,
+      "step": 2708
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0658851113716296e-05,
+      "loss": 0.956,
+      "step": 2709
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0655334114888628e-05,
+      "loss": 1.0144,
+      "step": 2710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0651817116060961e-05,
+      "loss": 1.009,
+      "step": 2711
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0648300117233295e-05,
+      "loss": 1.0414,
+      "step": 2712
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0644783118405626e-05,
+      "loss": 0.9384,
+      "step": 2713
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.064126611957796e-05,
+      "loss": 0.9946,
+      "step": 2714
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0637749120750293e-05,
+      "loss": 0.9501,
+      "step": 2715
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0634232121922627e-05,
+      "loss": 1.0172,
+      "step": 2716
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0630715123094959e-05,
+      "loss": 0.9728,
+      "step": 2717
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.062719812426729e-05,
+      "loss": 0.9528,
+      "step": 2718
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0623681125439624e-05,
+      "loss": 0.9536,
+      "step": 2719
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0620164126611958e-05,
+      "loss": 0.9823,
+      "step": 2720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.061664712778429e-05,
+      "loss": 0.9729,
+      "step": 2721
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0613130128956623e-05,
+      "loss": 0.9927,
+      "step": 2722
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0609613130128957e-05,
+      "loss": 1.0211,
+      "step": 2723
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.060609613130129e-05,
+      "loss": 0.9031,
+      "step": 2724
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0602579132473622e-05,
+      "loss": 0.9468,
+      "step": 2725
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0599062133645955e-05,
+      "loss": 0.9775,
+      "step": 2726
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.0595545134818289e-05,
+      "loss": 1.0116,
+      "step": 2727
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.059202813599062e-05,
+      "loss": 0.9645,
+      "step": 2728
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0588511137162954e-05,
+      "loss": 0.9229,
+      "step": 2729
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0584994138335288e-05,
+      "loss": 0.9581,
+      "step": 2730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.058147713950762e-05,
+      "loss": 1.0119,
+      "step": 2731
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0577960140679953e-05,
+      "loss": 0.9934,
+      "step": 2732
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0574443141852287e-05,
+      "loss": 1.0044,
+      "step": 2733
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0570926143024619e-05,
+      "loss": 0.9599,
+      "step": 2734
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0567409144196952e-05,
+      "loss": 1.0056,
+      "step": 2735
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0563892145369284e-05,
+      "loss": 0.9503,
+      "step": 2736
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0560375146541617e-05,
+      "loss": 0.9811,
+      "step": 2737
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0556858147713951e-05,
+      "loss": 0.9738,
+      "step": 2738
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0553341148886283e-05,
+      "loss": 0.9597,
+      "step": 2739
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0549824150058616e-05,
+      "loss": 1.0102,
+      "step": 2740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.054630715123095e-05,
+      "loss": 1.0382,
+      "step": 2741
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0542790152403282e-05,
+      "loss": 0.9523,
+      "step": 2742
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0539273153575615e-05,
+      "loss": 1.0016,
+      "step": 2743
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0535756154747949e-05,
+      "loss": 0.9702,
+      "step": 2744
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.053223915592028e-05,
+      "loss": 1.0049,
+      "step": 2745
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0528722157092614e-05,
+      "loss": 0.9896,
+      "step": 2746
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0525205158264948e-05,
+      "loss": 0.9897,
+      "step": 2747
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.052168815943728e-05,
+      "loss": 0.9405,
+      "step": 2748
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0518171160609613e-05,
+      "loss": 0.9994,
+      "step": 2749
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0514654161781945e-05,
+      "loss": 1.0152,
+      "step": 2750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.051113716295428e-05,
+      "loss": 0.9657,
+      "step": 2751
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0507620164126612e-05,
+      "loss": 0.9441,
+      "step": 2752
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.0504103165298944e-05,
+      "loss": 0.9795,
+      "step": 2753
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0500586166471277e-05,
+      "loss": 1.0491,
+      "step": 2754
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.049706916764361e-05,
+      "loss": 0.9885,
+      "step": 2755
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0493552168815943e-05,
+      "loss": 0.9424,
+      "step": 2756
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0490035169988276e-05,
+      "loss": 0.9653,
+      "step": 2757
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.048651817116061e-05,
+      "loss": 0.978,
+      "step": 2758
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0483001172332943e-05,
+      "loss": 0.9566,
+      "step": 2759
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0479484173505275e-05,
+      "loss": 1.0095,
+      "step": 2760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0475967174677608e-05,
+      "loss": 0.9895,
+      "step": 2761
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0472450175849942e-05,
+      "loss": 0.9884,
+      "step": 2762
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0468933177022274e-05,
+      "loss": 0.988,
+      "step": 2763
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0465416178194607e-05,
+      "loss": 1.0161,
+      "step": 2764
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0461899179366941e-05,
+      "loss": 0.9795,
+      "step": 2765
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0458382180539273e-05,
+      "loss": 0.9762,
+      "step": 2766
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0454865181711605e-05,
+      "loss": 0.9975,
+      "step": 2767
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.045134818288394e-05,
+      "loss": 0.973,
+      "step": 2768
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0447831184056272e-05,
+      "loss": 0.9753,
+      "step": 2769
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0444314185228605e-05,
+      "loss": 0.987,
+      "step": 2770
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0440797186400937e-05,
+      "loss": 0.9424,
+      "step": 2771
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.043728018757327e-05,
+      "loss": 0.927,
+      "step": 2772
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0433763188745604e-05,
+      "loss": 0.9778,
+      "step": 2773
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0430246189917936e-05,
+      "loss": 1.0043,
+      "step": 2774
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.042672919109027e-05,
+      "loss": 0.9694,
+      "step": 2775
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0423212192262603e-05,
+      "loss": 0.9623,
+      "step": 2776
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0419695193434935e-05,
+      "loss": 0.9754,
+      "step": 2777
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.0416178194607268e-05,
+      "loss": 0.9546,
+      "step": 2778
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0412661195779602e-05,
+      "loss": 0.9874,
+      "step": 2779
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0409144196951934e-05,
+      "loss": 0.995,
+      "step": 2780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0405627198124267e-05,
+      "loss": 0.9741,
+      "step": 2781
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.04021101992966e-05,
+      "loss": 1.0033,
+      "step": 2782
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0398593200468932e-05,
+      "loss": 1.0075,
+      "step": 2783
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0395076201641266e-05,
+      "loss": 0.9436,
+      "step": 2784
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0391559202813598e-05,
+      "loss": 0.9847,
+      "step": 2785
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0388042203985933e-05,
+      "loss": 0.9496,
+      "step": 2786
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0384525205158265e-05,
+      "loss": 0.971,
+      "step": 2787
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0381008206330597e-05,
+      "loss": 0.9847,
+      "step": 2788
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.037749120750293e-05,
+      "loss": 0.9863,
+      "step": 2789
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0373974208675264e-05,
+      "loss": 0.9546,
+      "step": 2790
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0370457209847596e-05,
+      "loss": 0.9916,
+      "step": 2791
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0366940211019929e-05,
+      "loss": 1.005,
+      "step": 2792
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0363423212192263e-05,
+      "loss": 0.9676,
+      "step": 2793
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0359906213364594e-05,
+      "loss": 0.972,
+      "step": 2794
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0356389214536928e-05,
+      "loss": 1.0111,
+      "step": 2795
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0352872215709261e-05,
+      "loss": 0.9737,
+      "step": 2796
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0349355216881595e-05,
+      "loss": 0.9545,
+      "step": 2797
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0345838218053927e-05,
+      "loss": 1.001,
+      "step": 2798
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.034232121922626e-05,
+      "loss": 0.9498,
+      "step": 2799
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0338804220398594e-05,
+      "loss": 0.9773,
+      "step": 2800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0335287221570926e-05,
+      "loss": 0.9709,
+      "step": 2801
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0331770222743258e-05,
+      "loss": 0.9645,
+      "step": 2802
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0328253223915591e-05,
+      "loss": 0.9543,
+      "step": 2803
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0324736225087925e-05,
+      "loss": 0.9462,
+      "step": 2804
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0321219226260258e-05,
+      "loss": 0.9703,
+      "step": 2805
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.031770222743259e-05,
+      "loss": 0.9769,
+      "step": 2806
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0314185228604923e-05,
+      "loss": 0.9712,
+      "step": 2807
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0310668229777257e-05,
+      "loss": 0.9648,
+      "step": 2808
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0307151230949589e-05,
+      "loss": 0.956,
+      "step": 2809
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0303634232121922e-05,
+      "loss": 1.0066,
+      "step": 2810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0300117233294256e-05,
+      "loss": 0.9917,
+      "step": 2811
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0296600234466588e-05,
+      "loss": 1.0082,
+      "step": 2812
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0293083235638921e-05,
+      "loss": 0.9828,
+      "step": 2813
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0289566236811255e-05,
+      "loss": 0.96,
+      "step": 2814
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0286049237983587e-05,
+      "loss": 0.9628,
+      "step": 2815
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.028253223915592e-05,
+      "loss": 0.9996,
+      "step": 2816
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0279015240328254e-05,
+      "loss": 1.0007,
+      "step": 2817
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0275498241500585e-05,
+      "loss": 1.026,
+      "step": 2818
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0271981242672919e-05,
+      "loss": 0.9552,
+      "step": 2819
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.026846424384525e-05,
+      "loss": 0.9995,
+      "step": 2820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0264947245017586e-05,
+      "loss": 0.9967,
+      "step": 2821
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0261430246189918e-05,
+      "loss": 0.9906,
+      "step": 2822
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.025791324736225e-05,
+      "loss": 0.9769,
+      "step": 2823
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0254396248534583e-05,
+      "loss": 0.9599,
+      "step": 2824
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0250879249706917e-05,
+      "loss": 1.0029,
+      "step": 2825
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0247362250879249e-05,
+      "loss": 0.9749,
+      "step": 2826
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0243845252051582e-05,
+      "loss": 0.9636,
+      "step": 2827
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0240328253223916e-05,
+      "loss": 1.0026,
+      "step": 2828
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0236811254396247e-05,
+      "loss": 0.9848,
+      "step": 2829
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.0233294255568581e-05,
+      "loss": 0.9231,
+      "step": 2830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0229777256740914e-05,
+      "loss": 1.0032,
+      "step": 2831
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0226260257913248e-05,
+      "loss": 0.9866,
+      "step": 2832
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.022274325908558e-05,
+      "loss": 1.0122,
+      "step": 2833
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0219226260257913e-05,
+      "loss": 0.985,
+      "step": 2834
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0215709261430247e-05,
+      "loss": 0.9974,
+      "step": 2835
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0212192262602579e-05,
+      "loss": 0.9903,
+      "step": 2836
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.020867526377491e-05,
+      "loss": 0.9842,
+      "step": 2837
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0205158264947244e-05,
+      "loss": 1.0201,
+      "step": 2838
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0201641266119578e-05,
+      "loss": 0.9908,
+      "step": 2839
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0198124267291911e-05,
+      "loss": 0.9658,
+      "step": 2840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0194607268464243e-05,
+      "loss": 1.0511,
+      "step": 2841
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0191090269636576e-05,
+      "loss": 1.0527,
+      "step": 2842
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.018757327080891e-05,
+      "loss": 0.9814,
+      "step": 2843
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0184056271981242e-05,
+      "loss": 1.0355,
+      "step": 2844
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0180539273153575e-05,
+      "loss": 0.9584,
+      "step": 2845
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0177022274325909e-05,
+      "loss": 0.9572,
+      "step": 2846
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.017350527549824e-05,
+      "loss": 0.9822,
+      "step": 2847
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0169988276670574e-05,
+      "loss": 0.9916,
+      "step": 2848
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0166471277842908e-05,
+      "loss": 1.0131,
+      "step": 2849
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.016295427901524e-05,
+      "loss": 1.0267,
+      "step": 2850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0159437280187573e-05,
+      "loss": 0.9272,
+      "step": 2851
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0155920281359907e-05,
+      "loss": 0.936,
+      "step": 2852
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0152403282532238e-05,
+      "loss": 0.9492,
+      "step": 2853
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0148886283704572e-05,
+      "loss": 0.9661,
+      "step": 2854
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0145369284876904e-05,
+      "loss": 0.95,
+      "step": 2855
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.0141852286049237e-05,
+      "loss": 0.9672,
+      "step": 2856
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0138335287221571e-05,
+      "loss": 0.9731,
+      "step": 2857
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0134818288393903e-05,
+      "loss": 0.9883,
+      "step": 2858
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0131301289566236e-05,
+      "loss": 0.9774,
+      "step": 2859
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.012778429073857e-05,
+      "loss": 1.0117,
+      "step": 2860
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0124267291910902e-05,
+      "loss": 0.9932,
+      "step": 2861
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0120750293083235e-05,
+      "loss": 0.9617,
+      "step": 2862
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0117233294255569e-05,
+      "loss": 0.9554,
+      "step": 2863
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.01137162954279e-05,
+      "loss": 1.0022,
+      "step": 2864
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0110199296600234e-05,
+      "loss": 1.0005,
+      "step": 2865
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0106682297772568e-05,
+      "loss": 0.9803,
+      "step": 2866
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0103165298944901e-05,
+      "loss": 1.0063,
+      "step": 2867
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0099648300117233e-05,
+      "loss": 0.9437,
+      "step": 2868
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0096131301289566e-05,
+      "loss": 0.9871,
+      "step": 2869
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.00926143024619e-05,
+      "loss": 0.9332,
+      "step": 2870
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0089097303634232e-05,
+      "loss": 0.9324,
+      "step": 2871
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0085580304806564e-05,
+      "loss": 1.014,
+      "step": 2872
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0082063305978897e-05,
+      "loss": 0.9849,
+      "step": 2873
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.007854630715123e-05,
+      "loss": 1.0176,
+      "step": 2874
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0075029308323564e-05,
+      "loss": 0.9788,
+      "step": 2875
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0071512309495896e-05,
+      "loss": 0.9922,
+      "step": 2876
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.006799531066823e-05,
+      "loss": 0.9989,
+      "step": 2877
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0064478311840563e-05,
+      "loss": 1.026,
+      "step": 2878
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0060961313012895e-05,
+      "loss": 0.9989,
+      "step": 2879
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0057444314185228e-05,
+      "loss": 0.9582,
+      "step": 2880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0053927315357562e-05,
+      "loss": 0.9978,
+      "step": 2881
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.0050410316529894e-05,
+      "loss": 0.9406,
+      "step": 2882
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0046893317702227e-05,
+      "loss": 0.9332,
+      "step": 2883
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.004337631887456e-05,
+      "loss": 0.9964,
+      "step": 2884
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0039859320046893e-05,
+      "loss": 1.0012,
+      "step": 2885
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0036342321219226e-05,
+      "loss": 0.9856,
+      "step": 2886
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.003282532239156e-05,
+      "loss": 1.0232,
+      "step": 2887
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0029308323563892e-05,
+      "loss": 0.9856,
+      "step": 2888
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0025791324736225e-05,
+      "loss": 1.0042,
+      "step": 2889
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0022274325908557e-05,
+      "loss": 1.0126,
+      "step": 2890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.001875732708089e-05,
+      "loss": 0.9478,
+      "step": 2891
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0015240328253224e-05,
+      "loss": 1.0136,
+      "step": 2892
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0011723329425556e-05,
+      "loss": 0.9785,
+      "step": 2893
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.000820633059789e-05,
+      "loss": 1.0008,
+      "step": 2894
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0004689331770223e-05,
+      "loss": 0.9961,
+      "step": 2895
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.0001172332942555e-05,
+      "loss": 0.9402,
+      "step": 2896
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.997655334114888e-06,
+      "loss": 1.0223,
+      "step": 2897
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.994138335287222e-06,
+      "loss": 0.972,
+      "step": 2898
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.990621336459553e-06,
+      "loss": 0.989,
+      "step": 2899
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.987104337631887e-06,
+      "loss": 0.9938,
+      "step": 2900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.98358733880422e-06,
+      "loss": 0.951,
+      "step": 2901
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.980070339976554e-06,
+      "loss": 0.988,
+      "step": 2902
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.976553341148886e-06,
+      "loss": 0.9966,
+      "step": 2903
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.97303634232122e-06,
+      "loss": 0.9405,
+      "step": 2904
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.969519343493553e-06,
+      "loss": 0.9997,
+      "step": 2905
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.966002344665885e-06,
+      "loss": 0.9722,
+      "step": 2906
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.962485345838217e-06,
+      "loss": 0.9571,
+      "step": 2907
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.95896834701055e-06,
+      "loss": 0.9386,
+      "step": 2908
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.955451348182884e-06,
+      "loss": 0.9988,
+      "step": 2909
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.951934349355217e-06,
+      "loss": 1.0306,
+      "step": 2910
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.948417350527549e-06,
+      "loss": 0.95,
+      "step": 2911
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.944900351699883e-06,
+      "loss": 0.9855,
+      "step": 2912
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.941383352872216e-06,
+      "loss": 1.0035,
+      "step": 2913
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.937866354044548e-06,
+      "loss": 0.9658,
+      "step": 2914
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.934349355216881e-06,
+      "loss": 0.9435,
+      "step": 2915
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.930832356389215e-06,
+      "loss": 0.9852,
+      "step": 2916
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.927315357561547e-06,
+      "loss": 0.9812,
+      "step": 2917
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.92379835873388e-06,
+      "loss": 0.9874,
+      "step": 2918
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.920281359906214e-06,
+      "loss": 0.9551,
+      "step": 2919
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.916764361078546e-06,
+      "loss": 0.9819,
+      "step": 2920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.91324736225088e-06,
+      "loss": 0.9934,
+      "step": 2921
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.909730363423213e-06,
+      "loss": 0.964,
+      "step": 2922
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.906213364595545e-06,
+      "loss": 0.994,
+      "step": 2923
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.902696365767878e-06,
+      "loss": 0.9892,
+      "step": 2924
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.89917936694021e-06,
+      "loss": 1.0321,
+      "step": 2925
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.895662368112543e-06,
+      "loss": 1.0245,
+      "step": 2926
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.892145369284877e-06,
+      "loss": 0.9815,
+      "step": 2927
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.888628370457209e-06,
+      "loss": 0.979,
+      "step": 2928
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.885111371629542e-06,
+      "loss": 0.9818,
+      "step": 2929
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.881594372801876e-06,
+      "loss": 0.9965,
+      "step": 2930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.878077373974208e-06,
+      "loss": 0.9946,
+      "step": 2931
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.874560375146541e-06,
+      "loss": 0.9727,
+      "step": 2932
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.871043376318875e-06,
+      "loss": 0.9614,
+      "step": 2933
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.867526377491207e-06,
+      "loss": 0.932,
+      "step": 2934
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.86400937866354e-06,
+      "loss": 1.0313,
+      "step": 2935
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.860492379835874e-06,
+      "loss": 0.9756,
+      "step": 2936
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.856975381008207e-06,
+      "loss": 0.9894,
+      "step": 2937
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.853458382180539e-06,
+      "loss": 1.0288,
+      "step": 2938
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.849941383352872e-06,
+      "loss": 0.9686,
+      "step": 2939
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.846424384525206e-06,
+      "loss": 0.976,
+      "step": 2940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.842907385697538e-06,
+      "loss": 1.0273,
+      "step": 2941
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.83939038686987e-06,
+      "loss": 0.9945,
+      "step": 2942
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.835873388042203e-06,
+      "loss": 0.9847,
+      "step": 2943
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.832356389214537e-06,
+      "loss": 1.0101,
+      "step": 2944
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.828839390386869e-06,
+      "loss": 1.0149,
+      "step": 2945
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.825322391559202e-06,
+      "loss": 1.0154,
+      "step": 2946
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.821805392731536e-06,
+      "loss": 1.0133,
+      "step": 2947
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.818288393903869e-06,
+      "loss": 0.9949,
+      "step": 2948
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.814771395076201e-06,
+      "loss": 0.9947,
+      "step": 2949
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.811254396248534e-06,
+      "loss": 0.9886,
+      "step": 2950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.807737397420868e-06,
+      "loss": 0.959,
+      "step": 2951
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.8042203985932e-06,
+      "loss": 0.954,
+      "step": 2952
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.800703399765533e-06,
+      "loss": 1.0077,
+      "step": 2953
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.797186400937867e-06,
+      "loss": 0.9666,
+      "step": 2954
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.793669402110199e-06,
+      "loss": 0.9581,
+      "step": 2955
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.790152403282532e-06,
+      "loss": 0.9468,
+      "step": 2956
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.786635404454866e-06,
+      "loss": 0.9715,
+      "step": 2957
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.783118405627198e-06,
+      "loss": 0.989,
+      "step": 2958
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.779601406799531e-06,
+      "loss": 0.9698,
+      "step": 2959
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.776084407971863e-06,
+      "loss": 0.9815,
+      "step": 2960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.772567409144196e-06,
+      "loss": 0.9725,
+      "step": 2961
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.76905041031653e-06,
+      "loss": 0.9882,
+      "step": 2962
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.765533411488862e-06,
+      "loss": 0.9638,
+      "step": 2963
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.762016412661195e-06,
+      "loss": 0.9888,
+      "step": 2964
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.758499413833529e-06,
+      "loss": 0.9755,
+      "step": 2965
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.75498241500586e-06,
+      "loss": 0.9584,
+      "step": 2966
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.751465416178194e-06,
+      "loss": 0.9713,
+      "step": 2967
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.747948417350528e-06,
+      "loss": 0.99,
+      "step": 2968
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.74443141852286e-06,
+      "loss": 1.0085,
+      "step": 2969
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.740914419695193e-06,
+      "loss": 0.9348,
+      "step": 2970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.737397420867527e-06,
+      "loss": 0.9851,
+      "step": 2971
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.73388042203986e-06,
+      "loss": 0.9645,
+      "step": 2972
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.730363423212192e-06,
+      "loss": 0.9369,
+      "step": 2973
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.726846424384525e-06,
+      "loss": 0.9598,
+      "step": 2974
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.723329425556859e-06,
+      "loss": 1.018,
+      "step": 2975
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.71981242672919e-06,
+      "loss": 0.9854,
+      "step": 2976
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.716295427901523e-06,
+      "loss": 0.9792,
+      "step": 2977
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.712778429073856e-06,
+      "loss": 0.9843,
+      "step": 2978
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.70926143024619e-06,
+      "loss": 1.0126,
+      "step": 2979
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.705744431418522e-06,
+      "loss": 0.9504,
+      "step": 2980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.702227432590855e-06,
+      "loss": 1.008,
+      "step": 2981
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.698710433763189e-06,
+      "loss": 0.9246,
+      "step": 2982
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.695193434935522e-06,
+      "loss": 0.9977,
+      "step": 2983
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.691676436107854e-06,
+      "loss": 1.0106,
+      "step": 2984
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.688159437280187e-06,
+      "loss": 0.9824,
+      "step": 2985
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.684642438452521e-06,
+      "loss": 0.9731,
+      "step": 2986
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.681125439624853e-06,
+      "loss": 1.014,
+      "step": 2987
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.677608440797186e-06,
+      "loss": 1.0155,
+      "step": 2988
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.67409144196952e-06,
+      "loss": 0.9769,
+      "step": 2989
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.670574443141852e-06,
+      "loss": 0.9636,
+      "step": 2990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.667057444314185e-06,
+      "loss": 0.978,
+      "step": 2991
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.663540445486519e-06,
+      "loss": 0.9454,
+      "step": 2992
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.66002344665885e-06,
+      "loss": 0.9529,
+      "step": 2993
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.656506447831184e-06,
+      "loss": 0.9633,
+      "step": 2994
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.652989449003516e-06,
+      "loss": 0.962,
+      "step": 2995
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.64947245017585e-06,
+      "loss": 0.9148,
+      "step": 2996
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.645955451348183e-06,
+      "loss": 1.0287,
+      "step": 2997
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.642438452520515e-06,
+      "loss": 0.9577,
+      "step": 2998
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.638921453692848e-06,
+      "loss": 0.9472,
+      "step": 2999
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.635404454865182e-06,
+      "loss": 0.9962,
+      "step": 3000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5170,
+  "num_train_epochs": 2,
+  "save_steps": 100,
+  "total_flos": 9335111417856000.0,
+  "trial_name": null,
+  "trial_params": null
+}