diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12364 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 2054,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.697825381369685e-06,
+      "loss": 0.6817,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.695650762739371e-06,
+      "loss": 0.7537,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.693476144109056e-06,
+      "loss": 0.7196,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.691301525478741e-06,
+      "loss": 0.6994,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.689126906848426e-06,
+      "loss": 0.6939,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.686952288218111e-06,
+      "loss": 0.7524,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.684777669587797e-06,
+      "loss": 0.6865,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.682603050957482e-06,
+      "loss": 0.6987,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.680428432327167e-06,
+      "loss": 0.702,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6782538136968525e-06,
+      "loss": 0.6809,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.676079195066537e-06,
+      "loss": 0.7565,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.673904576436222e-06,
+      "loss": 0.6589,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.671729957805908e-06,
+      "loss": 0.6834,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.669555339175593e-06,
+      "loss": 0.7139,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.667380720545278e-06,
+      "loss": 0.7025,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.665206101914963e-06,
+      "loss": 0.7349,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6630314832846475e-06,
+      "loss": 0.6992,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.660856864654333e-06,
+      "loss": 0.688,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.658682246024018e-06,
+      "loss": 0.668,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.656507627393703e-06,
+      "loss": 0.7267,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.654333008763389e-06,
+      "loss": 0.6675,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.652158390133074e-06,
+      "loss": 0.6939,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.649983771502759e-06,
+      "loss": 0.6894,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.647809152872444e-06,
+      "loss": 0.6965,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.645634534242129e-06,
+      "loss": 0.6531,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.643459915611815e-06,
+      "loss": 0.6634,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.6412852969815e-06,
+      "loss": 0.676,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.639110678351185e-06,
+      "loss": 0.688,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.63693605972087e-06,
+      "loss": 0.7112,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.634761441090555e-06,
+      "loss": 0.6822,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.63258682246024e-06,
+      "loss": 0.6909,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.630412203829926e-06,
+      "loss": 0.7432,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.628237585199611e-06,
+      "loss": 0.6704,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.6260629665692964e-06,
+      "loss": 0.6834,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.623888347938981e-06,
+      "loss": 0.704,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.621713729308666e-06,
+      "loss": 0.6974,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.619539110678352e-06,
+      "loss": 0.709,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.617364492048037e-06,
+      "loss": 0.6872,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.615189873417722e-06,
+      "loss": 0.7139,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.6130152547874074e-06,
+      "loss": 0.6982,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.610840636157092e-06,
+      "loss": 0.7359,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.608666017526777e-06,
+      "loss": 0.689,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.606491398896462e-06,
+      "loss": 0.6657,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.604316780266147e-06,
+      "loss": 0.6759,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.602142161635833e-06,
+      "loss": 0.6848,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.599967543005518e-06,
+      "loss": 0.6946,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5977929243752025e-06,
+      "loss": 0.6891,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.595618305744888e-06,
+      "loss": 0.7169,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.593443687114573e-06,
+      "loss": 0.717,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.591269068484259e-06,
+      "loss": 0.7185,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.589094449853944e-06,
+      "loss": 0.6761,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.586919831223629e-06,
+      "loss": 0.6744,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.584745212593314e-06,
+      "loss": 0.6424,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.582570593962999e-06,
+      "loss": 0.7126,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.580395975332684e-06,
+      "loss": 0.684,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.57822135670237e-06,
+      "loss": 0.6807,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.576046738072055e-06,
+      "loss": 0.6616,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.57387211944174e-06,
+      "loss": 0.7378,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.571697500811425e-06,
+      "loss": 0.706,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.56952288218111e-06,
+      "loss": 0.7133,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.567348263550796e-06,
+      "loss": 0.6801,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.565173644920481e-06,
+      "loss": 0.6712,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.562999026290166e-06,
+      "loss": 0.7194,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.5608244076598514e-06,
+      "loss": 0.669,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.558649789029536e-06,
+      "loss": 0.6852,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.556475170399221e-06,
+      "loss": 0.6844,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.554300551768907e-06,
+      "loss": 0.7062,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.552125933138592e-06,
+      "loss": 0.6768,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.549951314508277e-06,
+      "loss": 0.6509,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5477766958779624e-06,
+      "loss": 0.6986,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5456020772476465e-06,
+      "loss": 0.6719,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.543427458617332e-06,
+      "loss": 0.6789,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.541252839987017e-06,
+      "loss": 0.7002,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.539078221356702e-06,
+      "loss": 0.6713,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.536903602726388e-06,
+      "loss": 0.7214,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.534728984096073e-06,
+      "loss": 0.6683,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.532554365465758e-06,
+      "loss": 0.7005,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.530379746835443e-06,
+      "loss": 0.7465,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.528205128205128e-06,
+      "loss": 0.688,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.526030509574814e-06,
+      "loss": 0.6977,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.523855890944499e-06,
+      "loss": 0.7163,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.521681272314184e-06,
+      "loss": 0.6752,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.519506653683869e-06,
+      "loss": 0.7083,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.517332035053554e-06,
+      "loss": 0.6764,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.515157416423239e-06,
+      "loss": 0.6634,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.512982797792925e-06,
+      "loss": 0.6775,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.51080817916261e-06,
+      "loss": 0.6948,
+      "step": 87
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.508633560532295e-06,
+      "loss": 0.6572,
+      "step": 88
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.50645894190198e-06,
+      "loss": 0.683,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.504284323271665e-06,
+      "loss": 0.683,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.502109704641351e-06,
+      "loss": 0.6664,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.499935086011036e-06,
+      "loss": 0.672,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.497760467380721e-06,
+      "loss": 0.6948,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.495585848750406e-06,
+      "loss": 0.6448,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.493411230120091e-06,
+      "loss": 0.6893,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.491236611489776e-06,
+      "loss": 0.695,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.489061992859462e-06,
+      "loss": 0.6905,
+      "step": 97
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.486887374229147e-06,
+      "loss": 0.648,
+      "step": 98
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.484712755598832e-06,
+      "loss": 0.5802,
+      "step": 99
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.4825381369685166e-06,
+      "loss": 0.7537,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.4803635183382015e-06,
+      "loss": 0.6394,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.478188899707887e-06,
+      "loss": 0.597,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.476014281077572e-06,
+      "loss": 0.7145,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.473839662447258e-06,
+      "loss": 0.8375,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.471665043816943e-06,
+      "loss": 0.54,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.4694904251866276e-06,
+      "loss": 0.942,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.467315806556313e-06,
+      "loss": 0.535,
+      "step": 107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.465141187925998e-06,
+      "loss": 0.6681,
+      "step": 108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.462966569295683e-06,
+      "loss": 0.7981,
+      "step": 109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.460791950665369e-06,
+      "loss": 0.596,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.458617332035054e-06,
+      "loss": 0.8073,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.4564427134047386e-06,
+      "loss": 0.8823,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.454268094774424e-06,
+      "loss": 0.8097,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.452093476144109e-06,
+      "loss": 0.6947,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.449918857513795e-06,
+      "loss": 0.707,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.44774423888348e-06,
+      "loss": 0.7067,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.445569620253165e-06,
+      "loss": 0.7263,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.44339500162285e-06,
+      "loss": 0.677,
+      "step": 118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.441220382992535e-06,
+      "loss": 0.6485,
+      "step": 119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.43904576436222e-06,
+      "loss": 0.677,
+      "step": 120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.436871145731906e-06,
+      "loss": 0.6978,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.434696527101591e-06,
+      "loss": 0.6728,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.432521908471276e-06,
+      "loss": 0.704,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.430347289840961e-06,
+      "loss": 0.6869,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.428172671210646e-06,
+      "loss": 0.6894,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.425998052580332e-06,
+      "loss": 0.6754,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.423823433950016e-06,
+      "loss": 0.7029,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 6.421648815319701e-06,
+      "loss": 0.6961,
+      "step": 128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.419474196689387e-06,
+      "loss": 0.6436,
+      "step": 129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.4172995780590716e-06,
+      "loss": 0.6904,
+      "step": 130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.415124959428757e-06,
+      "loss": 0.7,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.412950340798442e-06,
+      "loss": 0.6927,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.410775722168127e-06,
+      "loss": 0.7345,
+      "step": 133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.408601103537813e-06,
+      "loss": 0.6933,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.406426484907498e-06,
+      "loss": 0.6817,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.4042518662771826e-06,
+      "loss": 0.6793,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.402077247646868e-06,
+      "loss": 0.6896,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.399902629016553e-06,
+      "loss": 0.6773,
+      "step": 138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.397728010386238e-06,
+      "loss": 0.67,
+      "step": 139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.395553391755924e-06,
+      "loss": 0.6419,
+      "step": 140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.393378773125609e-06,
+      "loss": 0.7069,
+      "step": 141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.391204154495294e-06,
+      "loss": 0.7281,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.389029535864979e-06,
+      "loss": 0.7483,
+      "step": 143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.386854917234664e-06,
+      "loss": 0.6819,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.38468029860435e-06,
+      "loss": 0.7842,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.382505679974035e-06,
+      "loss": 0.6638,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.38033106134372e-06,
+      "loss": 0.7358,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.378156442713405e-06,
+      "loss": 0.633,
+      "step": 148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.37598182408309e-06,
+      "loss": 0.6798,
+      "step": 149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.373807205452775e-06,
+      "loss": 0.7087,
+      "step": 150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.371632586822461e-06,
+      "loss": 0.7364,
+      "step": 151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.369457968192146e-06,
+      "loss": 0.734,
+      "step": 152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.3672833495618315e-06,
+      "loss": 0.7072,
+      "step": 153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.365108730931516e-06,
+      "loss": 0.6859,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.3629341123012004e-06,
+      "loss": 0.7614,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.360759493670886e-06,
+      "loss": 0.6699,
+      "step": 156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.358584875040571e-06,
+      "loss": 0.7218,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.356410256410257e-06,
+      "loss": 0.6992,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.354235637779942e-06,
+      "loss": 0.7125,
+      "step": 159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.3520610191496265e-06,
+      "loss": 0.7053,
+      "step": 160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.349886400519312e-06,
+      "loss": 0.6759,
+      "step": 161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.347711781888997e-06,
+      "loss": 0.7245,
+      "step": 162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.345537163258682e-06,
+      "loss": 0.6999,
+      "step": 163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.343362544628368e-06,
+      "loss": 0.6828,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.341187925998053e-06,
+      "loss": 0.6982,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.3390133073677375e-06,
+      "loss": 0.6917,
+      "step": 166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.336838688737423e-06,
+      "loss": 0.6995,
+      "step": 167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.334664070107108e-06,
+      "loss": 0.6786,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.332489451476794e-06,
+      "loss": 0.7042,
+      "step": 169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.330314832846479e-06,
+      "loss": 0.6695,
+      "step": 170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.328140214216164e-06,
+      "loss": 0.6734,
+      "step": 171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.325965595585849e-06,
+      "loss": 0.6693,
+      "step": 172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.323790976955534e-06,
+      "loss": 0.702,
+      "step": 173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.321616358325219e-06,
+      "loss": 0.7353,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.319441739694905e-06,
+      "loss": 0.6684,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.31726712106459e-06,
+      "loss": 0.6705,
+      "step": 176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.315092502434275e-06,
+      "loss": 0.7501,
+      "step": 177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.31291788380396e-06,
+      "loss": 0.6852,
+      "step": 178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.310743265173645e-06,
+      "loss": 0.6826,
+      "step": 179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.308568646543331e-06,
+      "loss": 0.7265,
+      "step": 180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.306394027913016e-06,
+      "loss": 0.6919,
+      "step": 181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.304219409282701e-06,
+      "loss": 0.669,
+      "step": 182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.302044790652386e-06,
+      "loss": 0.7,
+      "step": 183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.2998701720220705e-06,
+      "loss": 0.6707,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.297695553391755e-06,
+      "loss": 0.6898,
+      "step": 185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.295520934761441e-06,
+      "loss": 0.6524,
+      "step": 186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.293346316131126e-06,
+      "loss": 0.6854,
+      "step": 187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.291171697500812e-06,
+      "loss": 0.6808,
+      "step": 188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.288997078870497e-06,
+      "loss": 0.6746,
+      "step": 189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.2868224602401815e-06,
+      "loss": 0.6695,
+      "step": 190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.284647841609867e-06,
+      "loss": 0.7198,
+      "step": 191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.282473222979552e-06,
+      "loss": 0.7092,
+      "step": 192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.280298604349237e-06,
+      "loss": 0.6842,
+      "step": 193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.278123985718923e-06,
+      "loss": 0.6539,
+      "step": 194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.275949367088608e-06,
+      "loss": 0.7132,
+      "step": 195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.273774748458293e-06,
+      "loss": 0.6843,
+      "step": 196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.271600129827978e-06,
+      "loss": 0.6462,
+      "step": 197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.269425511197663e-06,
+      "loss": 0.6793,
+      "step": 198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.267250892567349e-06,
+      "loss": 0.6961,
+      "step": 199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.265076273937034e-06,
+      "loss": 0.6428,
+      "step": 200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.262901655306719e-06,
+      "loss": 0.7302,
+      "step": 201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.260727036676404e-06,
+      "loss": 0.7474,
+      "step": 202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.258552418046089e-06,
+      "loss": 0.6575,
+      "step": 203
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.256377799415774e-06,
+      "loss": 0.7031,
+      "step": 204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.25420318078546e-06,
+      "loss": 0.6312,
+      "step": 205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.252028562155145e-06,
+      "loss": 0.6271,
+      "step": 206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.2498539435248305e-06,
+      "loss": 0.6852,
+      "step": 207
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.247679324894515e-06,
+      "loss": 0.7118,
+      "step": 208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.2455047062642e-06,
+      "loss": 0.6791,
+      "step": 209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.243330087633886e-06,
+      "loss": 0.7101,
+      "step": 210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.24115546900357e-06,
+      "loss": 0.6421,
+      "step": 211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.238980850373255e-06,
+      "loss": 0.6635,
+      "step": 212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.236806231742941e-06,
+      "loss": 0.6681,
+      "step": 213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.2346316131126255e-06,
+      "loss": 0.7591,
+      "step": 214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.232456994482311e-06,
+      "loss": 0.7669,
+      "step": 215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.230282375851996e-06,
+      "loss": 0.6582,
+      "step": 216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.228107757221681e-06,
+      "loss": 0.7408,
+      "step": 217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.225933138591367e-06,
+      "loss": 0.691,
+      "step": 218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.223758519961052e-06,
+      "loss": 0.7693,
+      "step": 219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.2215839013307365e-06,
+      "loss": 0.7287,
+      "step": 220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.219409282700422e-06,
+      "loss": 0.7002,
+      "step": 221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.217234664070107e-06,
+      "loss": 0.6897,
+      "step": 222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.215060045439793e-06,
+      "loss": 0.7378,
+      "step": 223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.212885426809478e-06,
+      "loss": 0.7274,
+      "step": 224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.210710808179163e-06,
+      "loss": 0.7402,
+      "step": 225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.208536189548848e-06,
+      "loss": 0.6741,
+      "step": 226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.206361570918533e-06,
+      "loss": 0.6304,
+      "step": 227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.204186952288218e-06,
+      "loss": 0.6794,
+      "step": 228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.202012333657904e-06,
+      "loss": 0.7346,
+      "step": 229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.199837715027589e-06,
+      "loss": 0.7532,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.197663096397274e-06,
+      "loss": 0.6656,
+      "step": 231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.195488477766959e-06,
+      "loss": 0.7755,
+      "step": 232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.193313859136644e-06,
+      "loss": 0.6467,
+      "step": 233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.19113924050633e-06,
+      "loss": 0.6489,
+      "step": 234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.188964621876015e-06,
+      "loss": 0.7459,
+      "step": 235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1867900032457e-06,
+      "loss": 0.7985,
+      "step": 236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1846153846153855e-06,
+      "loss": 0.7159,
+      "step": 237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.18244076598507e-06,
+      "loss": 0.6911,
+      "step": 238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.180266147354754e-06,
+      "loss": 0.6933,
+      "step": 239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.17809152872444e-06,
+      "loss": 0.7089,
+      "step": 240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.175916910094125e-06,
+      "loss": 0.6579,
+      "step": 241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.173742291463811e-06,
+      "loss": 0.6235,
+      "step": 242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.171567672833496e-06,
+      "loss": 0.7281,
+      "step": 243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.1693930542031805e-06,
+      "loss": 0.6966,
+      "step": 244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.167218435572866e-06,
+      "loss": 0.7223,
+      "step": 245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.165043816942551e-06,
+      "loss": 0.6443,
+      "step": 246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.162869198312236e-06,
+      "loss": 0.7366,
+      "step": 247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.160694579681922e-06,
+      "loss": 0.6639,
+      "step": 248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.158519961051607e-06,
+      "loss": 0.7145,
+      "step": 249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.156345342421292e-06,
+      "loss": 0.6591,
+      "step": 250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.154170723790977e-06,
+      "loss": 0.7186,
+      "step": 251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.151996105160662e-06,
+      "loss": 0.6564,
+      "step": 252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.149821486530348e-06,
+      "loss": 0.6564,
+      "step": 253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.147646867900033e-06,
+      "loss": 0.6509,
+      "step": 254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.145472249269718e-06,
+      "loss": 0.6573,
+      "step": 255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.143297630639403e-06,
+      "loss": 0.6902,
+      "step": 256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.141123012009088e-06,
+      "loss": 0.6568,
+      "step": 257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.138948393378773e-06,
+      "loss": 0.6305,
+      "step": 258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.136773774748459e-06,
+      "loss": 0.6847,
+      "step": 259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.134599156118144e-06,
+      "loss": 0.6922,
+      "step": 260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.1324245374878295e-06,
+      "loss": 0.7145,
+      "step": 261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.130249918857514e-06,
+      "loss": 0.6176,
+      "step": 262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.128075300227199e-06,
+      "loss": 0.6821,
+      "step": 263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.125900681596885e-06,
+      "loss": 0.6038,
+      "step": 264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.12372606296657e-06,
+      "loss": 0.6889,
+      "step": 265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.121551444336255e-06,
+      "loss": 0.6869,
+      "step": 266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.11937682570594e-06,
+      "loss": 0.7549,
+      "step": 267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.1172022070756245e-06,
+      "loss": 0.6722,
+      "step": 268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.11502758844531e-06,
+      "loss": 0.7115,
+      "step": 269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.112852969814995e-06,
+      "loss": 0.6763,
+      "step": 270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.11067835118468e-06,
+      "loss": 0.6467,
+      "step": 271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.108503732554366e-06,
+      "loss": 0.7488,
+      "step": 272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.106329113924051e-06,
+      "loss": 0.7111,
+      "step": 273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.1041544952937355e-06,
+      "loss": 0.6402,
+      "step": 274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.101979876663421e-06,
+      "loss": 0.7278,
+      "step": 275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.099805258033106e-06,
+      "loss": 0.7112,
+      "step": 276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.097630639402792e-06,
+      "loss": 0.6325,
+      "step": 277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.095456020772477e-06,
+      "loss": 0.6684,
+      "step": 278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.093281402142162e-06,
+      "loss": 0.7079,
+      "step": 279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.091106783511847e-06,
+      "loss": 0.6579,
+      "step": 280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.088932164881532e-06,
+      "loss": 0.7102,
+      "step": 281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.086757546251217e-06,
+      "loss": 0.7542,
+      "step": 282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.084582927620903e-06,
+      "loss": 0.6386,
+      "step": 283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.082408308990588e-06,
+      "loss": 0.6847,
+      "step": 284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.080233690360273e-06,
+      "loss": 0.6777,
+      "step": 285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.078059071729958e-06,
+      "loss": 0.6413,
+      "step": 286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.075884453099643e-06,
+      "loss": 0.6617,
+      "step": 287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.073709834469329e-06,
+      "loss": 0.67,
+      "step": 288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.071535215839014e-06,
+      "loss": 0.7105,
+      "step": 289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.069360597208699e-06,
+      "loss": 0.7512,
+      "step": 290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.0671859785783845e-06,
+      "loss": 0.5662,
+      "step": 291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.065011359948069e-06,
+      "loss": 0.6144,
+      "step": 292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.062836741317754e-06,
+      "loss": 0.7352,
+      "step": 293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.06066212268744e-06,
+      "loss": 0.6725,
+      "step": 294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.058487504057124e-06,
+      "loss": 0.7458,
+      "step": 295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.05631288542681e-06,
+      "loss": 0.7219,
+      "step": 296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.054138266796495e-06,
+      "loss": 0.7504,
+      "step": 297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.0519636481661795e-06,
+      "loss": 0.6026,
+      "step": 298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.049789029535865e-06,
+      "loss": 0.7382,
+      "step": 299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.04761441090555e-06,
+      "loss": 0.6983,
+      "step": 300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.045439792275235e-06,
+      "loss": 0.6626,
+      "step": 301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.043265173644921e-06,
+      "loss": 0.6555,
+      "step": 302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.041090555014606e-06,
+      "loss": 0.5479,
+      "step": 303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.038915936384291e-06,
+      "loss": 0.6566,
+      "step": 304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.036741317753976e-06,
+      "loss": 0.6355,
+      "step": 305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.034566699123661e-06,
+      "loss": 0.6805,
+      "step": 306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.032392080493347e-06,
+      "loss": 0.685,
+      "step": 307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.030217461863032e-06,
+      "loss": 0.6586,
+      "step": 308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.028042843232717e-06,
+      "loss": 0.5246,
+      "step": 309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.025868224602402e-06,
+      "loss": 0.598,
+      "step": 310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.023693605972087e-06,
+      "loss": 0.7123,
+      "step": 311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.021518987341772e-06,
+      "loss": 0.625,
+      "step": 312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.019344368711458e-06,
+      "loss": 0.8931,
+      "step": 313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.017169750081143e-06,
+      "loss": 0.7768,
+      "step": 314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.0149951314508284e-06,
+      "loss": 0.678,
+      "step": 315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.012820512820513e-06,
+      "loss": 0.6016,
+      "step": 316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.010645894190198e-06,
+      "loss": 0.6505,
+      "step": 317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.008471275559884e-06,
+      "loss": 0.6694,
+      "step": 318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.006296656929569e-06,
+      "loss": 0.5313,
+      "step": 319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.004122038299254e-06,
+      "loss": 0.8063,
+      "step": 320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.0019474196689394e-06,
+      "loss": 0.5285,
+      "step": 321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.999772801038624e-06,
+      "loss": 0.6384,
+      "step": 322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.997598182408309e-06,
+      "loss": 0.7882,
+      "step": 323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.995423563777994e-06,
+      "loss": 0.8053,
+      "step": 324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.993248945147679e-06,
+      "loss": 0.6211,
+      "step": 325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.991074326517365e-06,
+      "loss": 0.5979,
+      "step": 326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.98889970788705e-06,
+      "loss": 0.7548,
+      "step": 327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9867250892567345e-06,
+      "loss": 0.5512,
+      "step": 328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.98455047062642e-06,
+      "loss": 0.8488,
+      "step": 329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.982375851996105e-06,
+      "loss": 0.6185,
+      "step": 330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.98020123336579e-06,
+      "loss": 0.7427,
+      "step": 331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.978026614735476e-06,
+      "loss": 0.6198,
+      "step": 332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.975851996105161e-06,
+      "loss": 0.5759,
+      "step": 333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.973677377474846e-06,
+      "loss": 0.6949,
+      "step": 334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.971502758844531e-06,
+      "loss": 0.7228,
+      "step": 335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.969328140214216e-06,
+      "loss": 0.4717,
+      "step": 336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.967153521583902e-06,
+      "loss": 0.5662,
+      "step": 337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.964978902953587e-06,
+      "loss": 0.7433,
+      "step": 338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.962804284323272e-06,
+      "loss": 0.751,
+      "step": 339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.960629665692957e-06,
+      "loss": 0.5067,
+      "step": 340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.958455047062642e-06,
+      "loss": 0.6156,
+      "step": 341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.956280428432328e-06,
+      "loss": 0.8232,
+      "step": 342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.954105809802013e-06,
+      "loss": 0.6026,
+      "step": 343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.951931191171698e-06,
+      "loss": 0.5063,
+      "step": 344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9497565725413834e-06,
+      "loss": 0.4486,
+      "step": 345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.947581953911068e-06,
+      "loss": 0.7322,
+      "step": 346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.945407335280753e-06,
+      "loss": 0.4134,
+      "step": 347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.943232716650439e-06,
+      "loss": 0.6513,
+      "step": 348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.941058098020124e-06,
+      "loss": 0.4255,
+      "step": 349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.938883479389809e-06,
+      "loss": 0.7527,
+      "step": 350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.936708860759494e-06,
+      "loss": 0.8193,
+      "step": 351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9345342421291785e-06,
+      "loss": 0.7079,
+      "step": 352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.932359623498864e-06,
+      "loss": 0.4958,
+      "step": 353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.930185004868549e-06,
+      "loss": 0.6688,
+      "step": 354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.928010386238234e-06,
+      "loss": 0.8112,
+      "step": 355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.92583576760792e-06,
+      "loss": 0.4942,
+      "step": 356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.923661148977605e-06,
+      "loss": 0.6791,
+      "step": 357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9214865303472895e-06,
+      "loss": 0.4866,
+      "step": 358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.919311911716975e-06,
+      "loss": 0.8851,
+      "step": 359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.91713729308666e-06,
+      "loss": 0.5416,
+      "step": 360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.914962674456346e-06,
+      "loss": 0.5943,
+      "step": 361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.912788055826031e-06,
+      "loss": 0.6421,
+      "step": 362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.910613437195716e-06,
+      "loss": 0.7519,
+      "step": 363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.908438818565401e-06,
+      "loss": 0.3705,
+      "step": 364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.906264199935086e-06,
+      "loss": 0.5362,
+      "step": 365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.904089581304771e-06,
+      "loss": 0.6865,
+      "step": 366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.901914962674457e-06,
+      "loss": 1.1517,
+      "step": 367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.899740344044142e-06,
+      "loss": 0.6418,
+      "step": 368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.897565725413827e-06,
+      "loss": 0.6568,
+      "step": 369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.895391106783512e-06,
+      "loss": 0.5802,
+      "step": 370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.893216488153197e-06,
+      "loss": 0.5214,
+      "step": 371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.891041869522883e-06,
+      "loss": 0.7986,
+      "step": 372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.888867250892568e-06,
+      "loss": 0.6174,
+      "step": 373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.886692632262253e-06,
+      "loss": 0.7808,
+      "step": 374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.884518013631938e-06,
+      "loss": 0.5835,
+      "step": 375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.882343395001623e-06,
+      "loss": 0.8945,
+      "step": 376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.880168776371308e-06,
+      "loss": 1.0499,
+      "step": 377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877994157740994e-06,
+      "loss": 0.614,
+      "step": 378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875819539110678e-06,
+      "loss": 0.5511,
+      "step": 379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873644920480364e-06,
+      "loss": 0.4187,
+      "step": 380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8714703018500486e-06,
+      "loss": 0.6396,
+      "step": 381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8692956832197335e-06,
+      "loss": 0.5064,
+      "step": 382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.867121064589419e-06,
+      "loss": 0.5233,
+      "step": 383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.864946445959104e-06,
+      "loss": 0.6117,
+      "step": 384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.862771827328789e-06,
+      "loss": 0.5639,
+      "step": 385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.860597208698475e-06,
+      "loss": 0.7857,
+      "step": 386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8584225900681596e-06,
+      "loss": 0.4921,
+      "step": 387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.856247971437845e-06,
+      "loss": 1.6215,
+      "step": 388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.85407335280753e-06,
+      "loss": 1.2481,
+      "step": 389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.851898734177215e-06,
+      "loss": 0.4154,
+      "step": 390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.849724115546901e-06,
+      "loss": 0.6367,
+      "step": 391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.847549496916586e-06,
+      "loss": 0.8198,
+      "step": 392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8453748782862706e-06,
+      "loss": 0.4773,
+      "step": 393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.843200259655956e-06,
+      "loss": 0.5976,
+      "step": 394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.841025641025641e-06,
+      "loss": 0.9637,
+      "step": 395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.838851022395327e-06,
+      "loss": 0.8518,
+      "step": 396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.836676403765012e-06,
+      "loss": 0.8171,
+      "step": 397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.834501785134697e-06,
+      "loss": 0.6219,
+      "step": 398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.832327166504382e-06,
+      "loss": 1.0139,
+      "step": 399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.830152547874067e-06,
+      "loss": 0.6445,
+      "step": 400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.827977929243752e-06,
+      "loss": 0.7279,
+      "step": 401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.825803310613438e-06,
+      "loss": 0.7077,
+      "step": 402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.823628691983123e-06,
+      "loss": 0.6896,
+      "step": 403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.821454073352808e-06,
+      "loss": 0.5071,
+      "step": 404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.819279454722493e-06,
+      "loss": 0.6078,
+      "step": 405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.817104836092178e-06,
+      "loss": 0.7835,
+      "step": 406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.814930217461863e-06,
+      "loss": 0.6846,
+      "step": 407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.812755598831548e-06,
+      "loss": 0.6087,
+      "step": 408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.810580980201233e-06,
+      "loss": 0.5838,
+      "step": 409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.808406361570919e-06,
+      "loss": 0.6586,
+      "step": 410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8062317429406036e-06,
+      "loss": 0.689,
+      "step": 411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8040571243102884e-06,
+      "loss": 0.6887,
+      "step": 412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.801882505679974e-06,
+      "loss": 0.6177,
+      "step": 413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.799707887049659e-06,
+      "loss": 0.8463,
+      "step": 414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.797533268419345e-06,
+      "loss": 0.6534,
+      "step": 415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.79535864978903e-06,
+      "loss": 0.5039,
+      "step": 416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.7931840311587146e-06,
+      "loss": 0.668,
+      "step": 417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.7910094125284e-06,
+      "loss": 0.7015,
+      "step": 418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.788834793898085e-06,
+      "loss": 0.6359,
+      "step": 419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.78666017526777e-06,
+      "loss": 0.7213,
+      "step": 420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.784485556637456e-06,
+      "loss": 0.5299,
+      "step": 421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.782310938007141e-06,
+      "loss": 0.8645,
+      "step": 422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.780136319376826e-06,
+      "loss": 0.79,
+      "step": 423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.777961700746511e-06,
+      "loss": 0.7072,
+      "step": 424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.775787082116196e-06,
+      "loss": 0.6093,
+      "step": 425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.773612463485882e-06,
+      "loss": 0.6383,
+      "step": 426
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.771437844855567e-06,
+      "loss": 0.6471,
+      "step": 427
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.769263226225252e-06,
+      "loss": 0.7369,
+      "step": 428
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.767088607594937e-06,
+      "loss": 0.8972,
+      "step": 429
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.764913988964622e-06,
+      "loss": 0.6141,
+      "step": 430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.762739370334307e-06,
+      "loss": 0.4332,
+      "step": 431
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.760564751703993e-06,
+      "loss": 0.7163,
+      "step": 432
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.758390133073678e-06,
+      "loss": 0.731,
+      "step": 433
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7562155144433635e-06,
+      "loss": 0.343,
+      "step": 434
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7540408958130475e-06,
+      "loss": 0.5051,
+      "step": 435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.7518662771827324e-06,
+      "loss": 0.7766,
+      "step": 436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.749691658552418e-06,
+      "loss": 0.8407,
+      "step": 437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.747517039922103e-06,
+      "loss": 0.4702,
+      "step": 438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.745342421291788e-06,
+      "loss": 0.5035,
+      "step": 439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.743167802661474e-06,
+      "loss": 1.2953,
+      "step": 440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7409931840311585e-06,
+      "loss": 0.7479,
+      "step": 441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.738818565400844e-06,
+      "loss": 0.5726,
+      "step": 442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.736643946770529e-06,
+      "loss": 1.1675,
+      "step": 443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.734469328140214e-06,
+      "loss": 0.3684,
+      "step": 444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7322947095099e-06,
+      "loss": 0.3732,
+      "step": 445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.730120090879585e-06,
+      "loss": 1.0015,
+      "step": 446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.7279454722492695e-06,
+      "loss": 0.7234,
+      "step": 447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.725770853618955e-06,
+      "loss": 0.4563,
+      "step": 448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.72359623498864e-06,
+      "loss": 0.4289,
+      "step": 449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.721421616358325e-06,
+      "loss": 0.6039,
+      "step": 450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.719246997728011e-06,
+      "loss": 0.7158,
+      "step": 451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.717072379097696e-06,
+      "loss": 0.5208,
+      "step": 452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.714897760467381e-06,
+      "loss": 0.8491,
+      "step": 453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.712723141837066e-06,
+      "loss": 0.9613,
+      "step": 454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.710548523206751e-06,
+      "loss": 0.3421,
+      "step": 455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.708373904576437e-06,
+      "loss": 1.005,
+      "step": 456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.706199285946122e-06,
+      "loss": 0.8077,
+      "step": 457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.704024667315807e-06,
+      "loss": 0.5251,
+      "step": 458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.701850048685492e-06,
+      "loss": 0.8211,
+      "step": 459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.699675430055177e-06,
+      "loss": 0.7557,
+      "step": 460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.697500811424863e-06,
+      "loss": 0.3736,
+      "step": 461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.695326192794548e-06,
+      "loss": 0.8946,
+      "step": 462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.693151574164232e-06,
+      "loss": 0.4698,
+      "step": 463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.690976955533918e-06,
+      "loss": 0.8967,
+      "step": 464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.6888023369036025e-06,
+      "loss": 0.8604,
+      "step": 465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.686627718273287e-06,
+      "loss": 0.327,
+      "step": 466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.684453099642973e-06,
+      "loss": 0.3945,
+      "step": 467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.682278481012658e-06,
+      "loss": 0.6342,
+      "step": 468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.680103862382344e-06,
+      "loss": 0.4046,
+      "step": 469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.677929243752029e-06,
+      "loss": 0.6428,
+      "step": 470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6757546251217135e-06,
+      "loss": 0.3221,
+      "step": 471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.673580006491399e-06,
+      "loss": 0.5099,
+      "step": 472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.671405387861084e-06,
+      "loss": 0.4256,
+      "step": 473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.669230769230769e-06,
+      "loss": 0.513,
+      "step": 474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.667056150600455e-06,
+      "loss": 0.4951,
+      "step": 475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.66488153197014e-06,
+      "loss": 0.582,
+      "step": 476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6627069133398245e-06,
+      "loss": 0.8297,
+      "step": 477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.66053229470951e-06,
+      "loss": 0.6451,
+      "step": 478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.658357676079195e-06,
+      "loss": 0.9137,
+      "step": 479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.656183057448881e-06,
+      "loss": 0.4366,
+      "step": 480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.654008438818566e-06,
+      "loss": 0.451,
+      "step": 481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.651833820188251e-06,
+      "loss": 0.3201,
+      "step": 482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.649659201557936e-06,
+      "loss": 0.4316,
+      "step": 483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.647484582927621e-06,
+      "loss": 0.6117,
+      "step": 484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.645309964297306e-06,
+      "loss": 0.8498,
+      "step": 485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.643135345666992e-06,
+      "loss": 0.9178,
+      "step": 486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.640960727036677e-06,
+      "loss": 0.8558,
+      "step": 487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6387861084063625e-06,
+      "loss": 0.5783,
+      "step": 488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.636611489776047e-06,
+      "loss": 0.6868,
+      "step": 489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.634436871145732e-06,
+      "loss": 0.4511,
+      "step": 490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.632262252515417e-06,
+      "loss": 0.483,
+      "step": 491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.630087633885102e-06,
+      "loss": 0.2477,
+      "step": 492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.627913015254787e-06,
+      "loss": 0.5201,
+      "step": 493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625738396624473e-06,
+      "loss": 0.146,
+      "step": 494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6235637779941575e-06,
+      "loss": 0.6224,
+      "step": 495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.621389159363843e-06,
+      "loss": 1.2075,
+      "step": 496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.619214540733528e-06,
+      "loss": 0.2707,
+      "step": 497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.617039922103213e-06,
+      "loss": 1.3528,
+      "step": 498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.614865303472899e-06,
+      "loss": 0.5508,
+      "step": 499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.612690684842584e-06,
+      "loss": 0.6851,
+      "step": 500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6105160662122685e-06,
+      "loss": 1.1705,
+      "step": 501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.608341447581954e-06,
+      "loss": 0.7836,
+      "step": 502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.606166828951639e-06,
+      "loss": 0.7645,
+      "step": 503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.603992210321324e-06,
+      "loss": 1.328,
+      "step": 504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.60181759169101e-06,
+      "loss": 0.8667,
+      "step": 505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.599642973060695e-06,
+      "loss": 0.6212,
+      "step": 506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.59746835443038e-06,
+      "loss": 0.6058,
+      "step": 507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.595293735800065e-06,
+      "loss": 0.4459,
+      "step": 508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.59311911716975e-06,
+      "loss": 0.6998,
+      "step": 509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.590944498539436e-06,
+      "loss": 0.9197,
+      "step": 510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.588769879909121e-06,
+      "loss": 0.7997,
+      "step": 511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.586595261278806e-06,
+      "loss": 0.4432,
+      "step": 512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.584420642648491e-06,
+      "loss": 0.3966,
+      "step": 513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.582246024018176e-06,
+      "loss": 0.6756,
+      "step": 514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.580071405387862e-06,
+      "loss": 0.6868,
+      "step": 515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.577896786757547e-06,
+      "loss": 0.3916,
+      "step": 516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.575722168127232e-06,
+      "loss": 0.8049,
+      "step": 517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5735475494969175e-06,
+      "loss": 0.6368,
+      "step": 518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.5713729308666015e-06,
+      "loss": 0.5471,
+      "step": 519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.569198312236286e-06,
+      "loss": 0.6219,
+      "step": 520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.567023693605972e-06,
+      "loss": 0.5048,
+      "step": 521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.564849074975657e-06,
+      "loss": 0.8082,
+      "step": 522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.562674456345343e-06,
+      "loss": 0.6017,
+      "step": 523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.560499837715028e-06,
+      "loss": 0.9122,
+      "step": 524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.5583252190847125e-06,
+      "loss": 0.6628,
+      "step": 525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.556150600454398e-06,
+      "loss": 0.8783,
+      "step": 526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.553975981824083e-06,
+      "loss": 0.3672,
+      "step": 527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.551801363193768e-06,
+      "loss": 0.8132,
+      "step": 528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.549626744563454e-06,
+      "loss": 0.4025,
+      "step": 529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.547452125933139e-06,
+      "loss": 0.9535,
+      "step": 530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5452775073028235e-06,
+      "loss": 0.7606,
+      "step": 531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.543102888672509e-06,
+      "loss": 0.9575,
+      "step": 532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.540928270042194e-06,
+      "loss": 0.3016,
+      "step": 533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.53875365141188e-06,
+      "loss": 0.3813,
+      "step": 534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.536579032781565e-06,
+      "loss": 0.9094,
+      "step": 535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.53440441415125e-06,
+      "loss": 0.8219,
+      "step": 536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.532229795520935e-06,
+      "loss": 0.8679,
+      "step": 537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.53005517689062e-06,
+      "loss": 0.8519,
+      "step": 538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.527880558260305e-06,
+      "loss": 0.4173,
+      "step": 539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.525705939629991e-06,
+      "loss": 0.5046,
+      "step": 540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.523531320999676e-06,
+      "loss": 0.4344,
+      "step": 541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.5213567023693615e-06,
+      "loss": 1.0172,
+      "step": 542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.519182083739046e-06,
+      "loss": 0.8564,
+      "step": 543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.517007465108731e-06,
+      "loss": 0.7331,
+      "step": 544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.514832846478417e-06,
+      "loss": 0.54,
+      "step": 545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.512658227848102e-06,
+      "loss": 0.3085,
+      "step": 546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.510483609217786e-06,
+      "loss": 0.8383,
+      "step": 547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.508308990587472e-06,
+      "loss": 0.2892,
+      "step": 548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.5061343719571565e-06,
+      "loss": 0.7057,
+      "step": 549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.503959753326842e-06,
+      "loss": 0.6728,
+      "step": 550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.501785134696527e-06,
+      "loss": 0.7673,
+      "step": 551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.499610516066212e-06,
+      "loss": 0.3565,
+      "step": 552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.497435897435898e-06,
+      "loss": 0.3954,
+      "step": 553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.495261278805583e-06,
+      "loss": 0.408,
+      "step": 554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.4930866601752675e-06,
+      "loss": 0.6456,
+      "step": 555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.490912041544953e-06,
+      "loss": 0.8927,
+      "step": 556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.488737422914638e-06,
+      "loss": 0.5696,
+      "step": 557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.486562804284323e-06,
+      "loss": 0.8333,
+      "step": 558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.484388185654009e-06,
+      "loss": 0.7372,
+      "step": 559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.482213567023694e-06,
+      "loss": 0.9361,
+      "step": 560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.480038948393379e-06,
+      "loss": 0.2863,
+      "step": 561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.477864329763064e-06,
+      "loss": 0.5857,
+      "step": 562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.475689711132749e-06,
+      "loss": 0.5074,
+      "step": 563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.473515092502435e-06,
+      "loss": 0.9961,
+      "step": 564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.47134047387212e-06,
+      "loss": 0.4076,
+      "step": 565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.469165855241805e-06,
+      "loss": 0.6892,
+      "step": 566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.46699123661149e-06,
+      "loss": 0.7044,
+      "step": 567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.464816617981175e-06,
+      "loss": 0.7182,
+      "step": 568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.46264199935086e-06,
+      "loss": 0.6114,
+      "step": 569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.460467380720546e-06,
+      "loss": 0.7207,
+      "step": 570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.458292762090231e-06,
+      "loss": 0.4064,
+      "step": 571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.4561181434599165e-06,
+      "loss": 0.4072,
+      "step": 572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.453943524829601e-06,
+      "loss": 0.9663,
+      "step": 573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.451768906199286e-06,
+      "loss": 0.7824,
+      "step": 574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.449594287568971e-06,
+      "loss": 0.656,
+      "step": 575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.447419668938656e-06,
+      "loss": 0.5473,
+      "step": 576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.445245050308342e-06,
+      "loss": 0.5885,
+      "step": 577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.443070431678027e-06,
+      "loss": 0.5812,
+      "step": 578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.4408958130477115e-06,
+      "loss": 0.4859,
+      "step": 579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.438721194417397e-06,
+      "loss": 0.503,
+      "step": 580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.436546575787082e-06,
+      "loss": 0.4859,
+      "step": 581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.434371957156767e-06,
+      "loss": 0.64,
+      "step": 582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.432197338526453e-06,
+      "loss": 0.8453,
+      "step": 583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.430022719896138e-06,
+      "loss": 0.5392,
+      "step": 584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.4278481012658225e-06,
+      "loss": 0.5151,
+      "step": 585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.425673482635508e-06,
+      "loss": 0.429,
+      "step": 586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.423498864005193e-06,
+      "loss": 0.845,
+      "step": 587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.421324245374879e-06,
+      "loss": 0.4062,
+      "step": 588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.419149626744564e-06,
+      "loss": 1.0288,
+      "step": 589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.416975008114249e-06,
+      "loss": 0.4775,
+      "step": 590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.414800389483934e-06,
+      "loss": 0.4885,
+      "step": 591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.412625770853619e-06,
+      "loss": 0.6327,
+      "step": 592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.410451152223304e-06,
+      "loss": 0.8387,
+      "step": 593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.40827653359299e-06,
+      "loss": 0.378,
+      "step": 594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.406101914962675e-06,
+      "loss": 0.5718,
+      "step": 595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.40392729633236e-06,
+      "loss": 0.984,
+      "step": 596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.401752677702045e-06,
+      "loss": 0.6255,
+      "step": 597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.39957805907173e-06,
+      "loss": 0.6387,
+      "step": 598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.397403440441416e-06,
+      "loss": 0.8767,
+      "step": 599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.395228821811101e-06,
+      "loss": 0.5569,
+      "step": 600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.393054203180786e-06,
+      "loss": 0.9314,
+      "step": 601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.3908795845504714e-06,
+      "loss": 0.7887,
+      "step": 602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.3887049659201555e-06,
+      "loss": 0.7154,
+      "step": 603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.386530347289841e-06,
+      "loss": 0.9171,
+      "step": 604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.384355728659526e-06,
+      "loss": 0.717,
+      "step": 605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.382181110029211e-06,
+      "loss": 0.7696,
+      "step": 606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.380006491398897e-06,
+      "loss": 0.933,
+      "step": 607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.377831872768582e-06,
+      "loss": 0.9263,
+      "step": 608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.3756572541382665e-06,
+      "loss": 0.7274,
+      "step": 609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.373482635507952e-06,
+      "loss": 0.5304,
+      "step": 610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.371308016877637e-06,
+      "loss": 0.7307,
+      "step": 611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.369133398247322e-06,
+      "loss": 0.6505,
+      "step": 612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.366958779617008e-06,
+      "loss": 0.4953,
+      "step": 613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.364784160986693e-06,
+      "loss": 0.476,
+      "step": 614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.362609542356378e-06,
+      "loss": 0.8029,
+      "step": 615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.360434923726063e-06,
+      "loss": 0.4807,
+      "step": 616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.358260305095748e-06,
+      "loss": 0.8052,
+      "step": 617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.356085686465434e-06,
+      "loss": 0.7679,
+      "step": 618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.353911067835119e-06,
+      "loss": 0.6029,
+      "step": 619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.351736449204804e-06,
+      "loss": 0.4029,
+      "step": 620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.349561830574489e-06,
+      "loss": 0.4933,
+      "step": 621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.347387211944174e-06,
+      "loss": 0.9018,
+      "step": 622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.345212593313859e-06,
+      "loss": 0.9361,
+      "step": 623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.343037974683545e-06,
+      "loss": 0.5228,
+      "step": 624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.34086335605323e-06,
+      "loss": 0.6455,
+      "step": 625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.3386887374229154e-06,
+      "loss": 0.7976,
+      "step": 626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.3365141187926e-06,
+      "loss": 0.9137,
+      "step": 627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.334339500162285e-06,
+      "loss": 0.5295,
+      "step": 628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.332164881531971e-06,
+      "loss": 0.3583,
+      "step": 629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.329990262901656e-06,
+      "loss": 0.4378,
+      "step": 630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.327815644271341e-06,
+      "loss": 0.8146,
+      "step": 631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.325641025641026e-06,
+      "loss": 1.1217,
+      "step": 632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.3234664070107105e-06,
+      "loss": 0.5095,
+      "step": 633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.321291788380396e-06,
+      "loss": 0.5723,
+      "step": 634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.319117169750081e-06,
+      "loss": 0.6934,
+      "step": 635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.316942551119766e-06,
+      "loss": 0.6608,
+      "step": 636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.314767932489452e-06,
+      "loss": 0.742,
+      "step": 637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.312593313859137e-06,
+      "loss": 0.5727,
+      "step": 638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.3104186952288215e-06,
+      "loss": 0.6059,
+      "step": 639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.308244076598507e-06,
+      "loss": 0.7034,
+      "step": 640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.306069457968192e-06,
+      "loss": 0.5547,
+      "step": 641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.303894839337878e-06,
+      "loss": 0.4323,
+      "step": 642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.301720220707563e-06,
+      "loss": 0.5574,
+      "step": 643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.299545602077248e-06,
+      "loss": 0.618,
+      "step": 644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.297370983446933e-06,
+      "loss": 0.7421,
+      "step": 645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.295196364816618e-06,
+      "loss": 0.5734,
+      "step": 646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.293021746186303e-06,
+      "loss": 0.961,
+      "step": 647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.290847127555989e-06,
+      "loss": 0.5159,
+      "step": 648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.288672508925674e-06,
+      "loss": 0.3965,
+      "step": 649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.2864978902953586e-06,
+      "loss": 0.5856,
+      "step": 650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.284323271665044e-06,
+      "loss": 0.6071,
+      "step": 651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.282148653034729e-06,
+      "loss": 0.7695,
+      "step": 652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.279974034404415e-06,
+      "loss": 0.478,
+      "step": 653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.2777994157741e-06,
+      "loss": 0.401,
+      "step": 654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.275624797143785e-06,
+      "loss": 0.7177,
+      "step": 655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.27345017851347e-06,
+      "loss": 0.4992,
+      "step": 656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.271275559883155e-06,
+      "loss": 0.8146,
+      "step": 657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.26910094125284e-06,
+      "loss": 0.7804,
+      "step": 658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.266926322622525e-06,
+      "loss": 0.6714,
+      "step": 659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.26475170399221e-06,
+      "loss": 0.4029,
+      "step": 660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.262577085361896e-06,
+      "loss": 0.7184,
+      "step": 661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.2604024667315806e-06,
+      "loss": 0.9155,
+      "step": 662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.2582278481012655e-06,
+      "loss": 0.7927,
+      "step": 663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.256053229470951e-06,
+      "loss": 0.4092,
+      "step": 664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.253878610840636e-06,
+      "loss": 0.5085,
+      "step": 665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.251703992210321e-06,
+      "loss": 0.5118,
+      "step": 666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.249529373580007e-06,
+      "loss": 0.7449,
+      "step": 667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.2473547549496916e-06,
+      "loss": 0.6152,
+      "step": 668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.245180136319377e-06,
+      "loss": 0.6996,
+      "step": 669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.243005517689062e-06,
+      "loss": 0.4755,
+      "step": 670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.240830899058747e-06,
+      "loss": 0.956,
+      "step": 671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.238656280428433e-06,
+      "loss": 1.311,
+      "step": 672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.236481661798118e-06,
+      "loss": 0.573,
+      "step": 673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2343070431678026e-06,
+      "loss": 0.4689,
+      "step": 674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.232132424537488e-06,
+      "loss": 0.429,
+      "step": 675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.229957805907173e-06,
+      "loss": 0.6171,
+      "step": 676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.227783187276858e-06,
+      "loss": 0.4672,
+      "step": 677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.225608568646544e-06,
+      "loss": 0.8311,
+      "step": 678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.223433950016229e-06,
+      "loss": 0.4033,
+      "step": 679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.221259331385914e-06,
+      "loss": 0.484,
+      "step": 680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.219084712755599e-06,
+      "loss": 0.8499,
+      "step": 681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.216910094125284e-06,
+      "loss": 0.4774,
+      "step": 682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21473547549497e-06,
+      "loss": 0.4476,
+      "step": 683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212560856864655e-06,
+      "loss": 0.8264,
+      "step": 684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21038623823434e-06,
+      "loss": 0.3093,
+      "step": 685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208211619604025e-06,
+      "loss": 0.3038,
+      "step": 686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2060370009737094e-06,
+      "loss": 0.8135,
+      "step": 687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203862382343395e-06,
+      "loss": 0.5101,
+      "step": 688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20168776371308e-06,
+      "loss": 0.6784,
+      "step": 689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199513145082765e-06,
+      "loss": 0.6572,
+      "step": 690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.197338526452451e-06,
+      "loss": 0.4295,
+      "step": 691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1951639078221356e-06,
+      "loss": 0.4787,
+      "step": 692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1929892891918204e-06,
+      "loss": 0.6033,
+      "step": 693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.190814670561506e-06,
+      "loss": 0.8097,
+      "step": 694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.188640051931191e-06,
+      "loss": 0.6858,
+      "step": 695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.186465433300877e-06,
+      "loss": 0.3206,
+      "step": 696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.184290814670562e-06,
+      "loss": 0.4983,
+      "step": 697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.1821161960402466e-06,
+      "loss": 0.5178,
+      "step": 698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.179941577409932e-06,
+      "loss": 1.3049,
+      "step": 699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.177766958779617e-06,
+      "loss": 0.8485,
+      "step": 700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.175592340149302e-06,
+      "loss": 0.2688,
+      "step": 701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.173417721518988e-06,
+      "loss": 0.6689,
+      "step": 702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.171243102888673e-06,
+      "loss": 0.4606,
+      "step": 703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1690684842583576e-06,
+      "loss": 0.8123,
+      "step": 704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.166893865628043e-06,
+      "loss": 0.593,
+      "step": 705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.164719246997728e-06,
+      "loss": 0.2778,
+      "step": 706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.162544628367414e-06,
+      "loss": 0.2661,
+      "step": 707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.160370009737099e-06,
+      "loss": 0.5596,
+      "step": 708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.158195391106784e-06,
+      "loss": 0.3728,
+      "step": 709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.156020772476469e-06,
+      "loss": 0.404,
+      "step": 710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.153846153846154e-06,
+      "loss": 0.7592,
+      "step": 711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.151671535215839e-06,
+      "loss": 0.3681,
+      "step": 712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.149496916585525e-06,
+      "loss": 0.8277,
+      "step": 713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.14732229795521e-06,
+      "loss": 0.8271,
+      "step": 714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.145147679324895e-06,
+      "loss": 0.7555,
+      "step": 715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1429730606945795e-06,
+      "loss": 0.1591,
+      "step": 716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1407984420642644e-06,
+      "loss": 0.3388,
+      "step": 717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.13862382343395e-06,
+      "loss": 0.6007,
+      "step": 718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.136449204803635e-06,
+      "loss": 1.2185,
+      "step": 719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.13427458617332e-06,
+      "loss": 0.7508,
+      "step": 720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.132099967543006e-06,
+      "loss": 0.2876,
+      "step": 721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1299253489126905e-06,
+      "loss": 0.3259,
+      "step": 722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.127750730282376e-06,
+      "loss": 0.2907,
+      "step": 723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.125576111652061e-06,
+      "loss": 0.5569,
+      "step": 724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.123401493021746e-06,
+      "loss": 0.3668,
+      "step": 725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.121226874391432e-06,
+      "loss": 0.6158,
+      "step": 726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.119052255761117e-06,
+      "loss": 0.6663,
+      "step": 727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.1168776371308015e-06,
+      "loss": 0.1553,
+      "step": 728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.114703018500487e-06,
+      "loss": 0.9632,
+      "step": 729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.112528399870172e-06,
+      "loss": 0.9459,
+      "step": 730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.110353781239857e-06,
+      "loss": 0.626,
+      "step": 731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.108179162609543e-06,
+      "loss": 0.3753,
+      "step": 732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.106004543979228e-06,
+      "loss": 0.5703,
+      "step": 733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.103829925348913e-06,
+      "loss": 1.2143,
+      "step": 734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.101655306718598e-06,
+      "loss": 0.6008,
+      "step": 735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.099480688088283e-06,
+      "loss": 0.299,
+      "step": 736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.097306069457969e-06,
+      "loss": 0.8267,
+      "step": 737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.095131450827654e-06,
+      "loss": 0.7137,
+      "step": 738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.092956832197339e-06,
+      "loss": 0.5183,
+      "step": 739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.090782213567024e-06,
+      "loss": 0.6909,
+      "step": 740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.088607594936709e-06,
+      "loss": 1.1085,
+      "step": 741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.086432976306394e-06,
+      "loss": 0.4188,
+      "step": 742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.084258357676079e-06,
+      "loss": 0.2469,
+      "step": 743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.082083739045764e-06,
+      "loss": 1.0088,
+      "step": 744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.07990912041545e-06,
+      "loss": 0.5174,
+      "step": 745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0777345017851345e-06,
+      "loss": 1.3383,
+      "step": 746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.075559883154819e-06,
+      "loss": 0.1993,
+      "step": 747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.073385264524505e-06,
+      "loss": 0.2427,
+      "step": 748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.07121064589419e-06,
+      "loss": 0.5734,
+      "step": 749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.069036027263876e-06,
+      "loss": 0.7477,
+      "step": 750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.066861408633561e-06,
+      "loss": 1.3359,
+      "step": 751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.0646867900032455e-06,
+      "loss": 0.9589,
+      "step": 752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.062512171372931e-06,
+      "loss": 0.5497,
+      "step": 753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.060337552742616e-06,
+      "loss": 1.4287,
+      "step": 754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.058162934112301e-06,
+      "loss": 0.2537,
+      "step": 755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.055988315481987e-06,
+      "loss": 0.6623,
+      "step": 756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.053813696851672e-06,
+      "loss": 0.661,
+      "step": 757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0516390782213565e-06,
+      "loss": 0.3741,
+      "step": 758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.049464459591042e-06,
+      "loss": 0.3395,
+      "step": 759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.047289840960727e-06,
+      "loss": 0.1317,
+      "step": 760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.045115222330413e-06,
+      "loss": 1.5234,
+      "step": 761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.042940603700098e-06,
+      "loss": 0.4715,
+      "step": 762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.040765985069783e-06,
+      "loss": 0.9497,
+      "step": 763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.038591366439468e-06,
+      "loss": 1.174,
+      "step": 764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.036416747809153e-06,
+      "loss": 0.4522,
+      "step": 765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.034242129178838e-06,
+      "loss": 0.2711,
+      "step": 766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.032067510548524e-06,
+      "loss": 0.3585,
+      "step": 767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.029892891918209e-06,
+      "loss": 0.4807,
+      "step": 768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.027718273287894e-06,
+      "loss": 0.8312,
+      "step": 769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.025543654657579e-06,
+      "loss": 0.2248,
+      "step": 770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.023369036027263e-06,
+      "loss": 0.4017,
+      "step": 771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.021194417396949e-06,
+      "loss": 1.077,
+      "step": 772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.019019798766634e-06,
+      "loss": 0.1824,
+      "step": 773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.016845180136319e-06,
+      "loss": 0.7331,
+      "step": 774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.014670561506005e-06,
+      "loss": 0.5354,
+      "step": 775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0124959428756895e-06,
+      "loss": 0.6025,
+      "step": 776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.010321324245375e-06,
+      "loss": 0.452,
+      "step": 777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.00814670561506e-06,
+      "loss": 0.3012,
+      "step": 778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.005972086984745e-06,
+      "loss": 0.8775,
+      "step": 779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.003797468354431e-06,
+      "loss": 1.0354,
+      "step": 780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.001622849724116e-06,
+      "loss": 0.9584,
+      "step": 781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9994482310938005e-06,
+      "loss": 0.5379,
+      "step": 782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.997273612463486e-06,
+      "loss": 1.0512,
+      "step": 783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995098993833171e-06,
+      "loss": 0.4904,
+      "step": 784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.992924375202856e-06,
+      "loss": 0.2676,
+      "step": 785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.990749756572542e-06,
+      "loss": 0.5796,
+      "step": 786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.988575137942227e-06,
+      "loss": 0.2645,
+      "step": 787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.986400519311912e-06,
+      "loss": 0.2532,
+      "step": 788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.984225900681597e-06,
+      "loss": 0.2672,
+      "step": 789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.982051282051282e-06,
+      "loss": 0.6482,
+      "step": 790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.979876663420968e-06,
+      "loss": 0.837,
+      "step": 791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.977702044790653e-06,
+      "loss": 0.4346,
+      "step": 792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.975527426160338e-06,
+      "loss": 0.1616,
+      "step": 793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.973352807530023e-06,
+      "loss": 0.4202,
+      "step": 794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.971178188899708e-06,
+      "loss": 0.8416,
+      "step": 795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.969003570269393e-06,
+      "loss": 0.1818,
+      "step": 796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.966828951639079e-06,
+      "loss": 0.5735,
+      "step": 797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.964654333008764e-06,
+      "loss": 1.0394,
+      "step": 798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.962479714378449e-06,
+      "loss": 0.2301,
+      "step": 799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9603050957481335e-06,
+      "loss": 0.3942,
+      "step": 800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.958130477117818e-06,
+      "loss": 0.9715,
+      "step": 801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.955955858487504e-06,
+      "loss": 0.2734,
+      "step": 802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.953781239857189e-06,
+      "loss": 0.4842,
+      "step": 803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.951606621226875e-06,
+      "loss": 0.4678,
+      "step": 804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.94943200259656e-06,
+      "loss": 0.7898,
+      "step": 805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9472573839662445e-06,
+      "loss": 0.7249,
+      "step": 806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.94508276533593e-06,
+      "loss": 0.1642,
+      "step": 807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.942908146705615e-06,
+      "loss": 0.8107,
+      "step": 808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9407335280753e-06,
+      "loss": 0.565,
+      "step": 809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.938558909444986e-06,
+      "loss": 0.5233,
+      "step": 810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.936384290814671e-06,
+      "loss": 0.6964,
+      "step": 811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9342096721843555e-06,
+      "loss": 1.2015,
+      "step": 812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.932035053554041e-06,
+      "loss": 0.8237,
+      "step": 813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.929860434923726e-06,
+      "loss": 0.6167,
+      "step": 814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.927685816293412e-06,
+      "loss": 0.8351,
+      "step": 815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.925511197663097e-06,
+      "loss": 0.6126,
+      "step": 816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.923336579032782e-06,
+      "loss": 0.1205,
+      "step": 817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.921161960402467e-06,
+      "loss": 0.3117,
+      "step": 818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.918987341772152e-06,
+      "loss": 0.7643,
+      "step": 819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.916812723141837e-06,
+      "loss": 0.7078,
+      "step": 820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.914638104511523e-06,
+      "loss": 0.7261,
+      "step": 821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.912463485881208e-06,
+      "loss": 0.3285,
+      "step": 822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.910288867250893e-06,
+      "loss": 0.6502,
+      "step": 823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.908114248620578e-06,
+      "loss": 0.4695,
+      "step": 824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.905939629990263e-06,
+      "loss": 0.1863,
+      "step": 825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.903765011359949e-06,
+      "loss": 0.3179,
+      "step": 826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.901590392729633e-06,
+      "loss": 0.4708,
+      "step": 827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.899415774099318e-06,
+      "loss": 0.387,
+      "step": 828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.897241155469004e-06,
+      "loss": 0.3588,
+      "step": 829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8950665368386885e-06,
+      "loss": 0.221,
+      "step": 830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.892891918208374e-06,
+      "loss": 0.3754,
+      "step": 831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.890717299578059e-06,
+      "loss": 1.2769,
+      "step": 832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.888542680947744e-06,
+      "loss": 0.7319,
+      "step": 833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.88636806231743e-06,
+      "loss": 0.2501,
+      "step": 834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.884193443687115e-06,
+      "loss": 0.8841,
+      "step": 835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.8820188250567995e-06,
+      "loss": 0.6608,
+      "step": 836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.879844206426485e-06,
+      "loss": 0.6122,
+      "step": 837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.87766958779617e-06,
+      "loss": 0.1864,
+      "step": 838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.875494969165855e-06,
+      "loss": 0.6402,
+      "step": 839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.873320350535541e-06,
+      "loss": 0.1981,
+      "step": 840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.871145731905226e-06,
+      "loss": 0.3104,
+      "step": 841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.868971113274911e-06,
+      "loss": 1.6726,
+      "step": 842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.866796494644596e-06,
+      "loss": 0.4316,
+      "step": 843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.864621876014281e-06,
+      "loss": 0.2845,
+      "step": 844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.862447257383967e-06,
+      "loss": 0.3563,
+      "step": 845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.860272638753652e-06,
+      "loss": 1.1778,
+      "step": 846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.858098020123337e-06,
+      "loss": 0.3258,
+      "step": 847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.855923401493022e-06,
+      "loss": 1.4021,
+      "step": 848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.853748782862707e-06,
+      "loss": 1.8677,
+      "step": 849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.851574164232392e-06,
+      "loss": 0.1565,
+      "step": 850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.849399545602078e-06,
+      "loss": 0.302,
+      "step": 851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.847224926971763e-06,
+      "loss": 1.2744,
+      "step": 852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.8450503083414485e-06,
+      "loss": 0.5084,
+      "step": 853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.842875689711133e-06,
+      "loss": 1.1163,
+      "step": 854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.840701071080817e-06,
+      "loss": 0.307,
+      "step": 855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.838526452450503e-06,
+      "loss": 0.1145,
+      "step": 856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.836351833820188e-06,
+      "loss": 0.2233,
+      "step": 857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.834177215189874e-06,
+      "loss": 0.4797,
+      "step": 858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.832002596559559e-06,
+      "loss": 1.3233,
+      "step": 859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8298279779292435e-06,
+      "loss": 0.2332,
+      "step": 860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.827653359298929e-06,
+      "loss": 1.0729,
+      "step": 861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.825478740668614e-06,
+      "loss": 0.2485,
+      "step": 862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.823304122038299e-06,
+      "loss": 0.618,
+      "step": 863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.821129503407985e-06,
+      "loss": 0.3164,
+      "step": 864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.81895488477767e-06,
+      "loss": 0.9521,
+      "step": 865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.8167802661473545e-06,
+      "loss": 0.869,
+      "step": 866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.81460564751704e-06,
+      "loss": 1.0634,
+      "step": 867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.812431028886725e-06,
+      "loss": 0.4084,
+      "step": 868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.810256410256411e-06,
+      "loss": 0.925,
+      "step": 869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.808081791626096e-06,
+      "loss": 0.1579,
+      "step": 870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.805907172995781e-06,
+      "loss": 0.5433,
+      "step": 871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.803732554365466e-06,
+      "loss": 0.7258,
+      "step": 872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.801557935735151e-06,
+      "loss": 0.2844,
+      "step": 873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.799383317104836e-06,
+      "loss": 0.174,
+      "step": 874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.797208698474522e-06,
+      "loss": 0.1943,
+      "step": 875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.795034079844207e-06,
+      "loss": 1.0066,
+      "step": 876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.792859461213892e-06,
+      "loss": 1.1471,
+      "step": 877
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.790684842583577e-06,
+      "loss": 0.4109,
+      "step": 878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.788510223953262e-06,
+      "loss": 0.3564,
+      "step": 879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.786335605322948e-06,
+      "loss": 1.2234,
+      "step": 880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.784160986692633e-06,
+      "loss": 0.1595,
+      "step": 881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.781986368062318e-06,
+      "loss": 0.195,
+      "step": 882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.779811749432003e-06,
+      "loss": 0.1881,
+      "step": 883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7776371308016875e-06,
+      "loss": 0.2561,
+      "step": 884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.775462512171373e-06,
+      "loss": 0.5835,
+      "step": 885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.773287893541058e-06,
+      "loss": 0.8,
+      "step": 886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.771113274910743e-06,
+      "loss": 0.1465,
+      "step": 887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.768938656280429e-06,
+      "loss": 0.6139,
+      "step": 888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.766764037650114e-06,
+      "loss": 0.2847,
+      "step": 889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7645894190197985e-06,
+      "loss": 0.1528,
+      "step": 890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.762414800389484e-06,
+      "loss": 0.3735,
+      "step": 891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.760240181759169e-06,
+      "loss": 0.2505,
+      "step": 892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.758065563128854e-06,
+      "loss": 1.8104,
+      "step": 893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.75589094449854e-06,
+      "loss": 0.6299,
+      "step": 894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.753716325868225e-06,
+      "loss": 0.4679,
+      "step": 895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.75154170723791e-06,
+      "loss": 0.3004,
+      "step": 896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.749367088607595e-06,
+      "loss": 0.1853,
+      "step": 897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.74719246997728e-06,
+      "loss": 1.3158,
+      "step": 898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.745017851346966e-06,
+      "loss": 0.4428,
+      "step": 899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.742843232716651e-06,
+      "loss": 0.3823,
+      "step": 900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.740668614086336e-06,
+      "loss": 0.4548,
+      "step": 901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.738493995456021e-06,
+      "loss": 1.3566,
+      "step": 902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.736319376825706e-06,
+      "loss": 0.6756,
+      "step": 903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.734144758195391e-06,
+      "loss": 0.1802,
+      "step": 904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.731970139565077e-06,
+      "loss": 1.144,
+      "step": 905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.729795520934762e-06,
+      "loss": 0.4566,
+      "step": 906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7276209023044474e-06,
+      "loss": 1.3557,
+      "step": 907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725446283674132e-06,
+      "loss": 0.6472,
+      "step": 908
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.723271665043817e-06,
+      "loss": 0.6462,
+      "step": 909
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721097046413503e-06,
+      "loss": 0.5246,
+      "step": 910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.718922427783187e-06,
+      "loss": 0.1684,
+      "step": 911
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716747809152873e-06,
+      "loss": 0.7644,
+      "step": 912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.714573190522558e-06,
+      "loss": 0.7248,
+      "step": 913
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7123985718922425e-06,
+      "loss": 0.561,
+      "step": 914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.710223953261928e-06,
+      "loss": 0.14,
+      "step": 915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.708049334631613e-06,
+      "loss": 0.1878,
+      "step": 916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.705874716001298e-06,
+      "loss": 0.5559,
+      "step": 917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.703700097370984e-06,
+      "loss": 0.6929,
+      "step": 918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.701525478740669e-06,
+      "loss": 0.5416,
+      "step": 919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6993508601103535e-06,
+      "loss": 0.3252,
+      "step": 920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.697176241480039e-06,
+      "loss": 0.5236,
+      "step": 921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.695001622849724e-06,
+      "loss": 0.918,
+      "step": 922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.69282700421941e-06,
+      "loss": 0.5746,
+      "step": 923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.690652385589095e-06,
+      "loss": 0.1627,
+      "step": 924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.68847776695878e-06,
+      "loss": 0.438,
+      "step": 925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.686303148328465e-06,
+      "loss": 0.7421,
+      "step": 926
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.68412852969815e-06,
+      "loss": 0.7599,
+      "step": 927
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.681953911067835e-06,
+      "loss": 1.2018,
+      "step": 928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.679779292437521e-06,
+      "loss": 0.9076,
+      "step": 929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.677604673807206e-06,
+      "loss": 0.5365,
+      "step": 930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6754300551768906e-06,
+      "loss": 0.6405,
+      "step": 931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.673255436546576e-06,
+      "loss": 0.624,
+      "step": 932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.671080817916261e-06,
+      "loss": 0.176,
+      "step": 933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.668906199285947e-06,
+      "loss": 0.6246,
+      "step": 934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.666731580655632e-06,
+      "loss": 0.5801,
+      "step": 935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.664556962025317e-06,
+      "loss": 0.7553,
+      "step": 936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.662382343395002e-06,
+      "loss": 0.1942,
+      "step": 937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.660207724764687e-06,
+      "loss": 1.3905,
+      "step": 938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.658033106134371e-06,
+      "loss": 0.7778,
+      "step": 939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.655858487504057e-06,
+      "loss": 0.3552,
+      "step": 940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.653683868873742e-06,
+      "loss": 1.1808,
+      "step": 941
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.651509250243428e-06,
+      "loss": 0.564,
+      "step": 942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6493346316131126e-06,
+      "loss": 0.2023,
+      "step": 943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6471600129827975e-06,
+      "loss": 1.6021,
+      "step": 944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.644985394352483e-06,
+      "loss": 0.8151,
+      "step": 945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.642810775722168e-06,
+      "loss": 1.0669,
+      "step": 946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.640636157091853e-06,
+      "loss": 0.6428,
+      "step": 947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.638461538461539e-06,
+      "loss": 1.1143,
+      "step": 948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6362869198312236e-06,
+      "loss": 0.362,
+      "step": 949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.634112301200909e-06,
+      "loss": 0.7129,
+      "step": 950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.631937682570594e-06,
+      "loss": 0.326,
+      "step": 951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.629763063940279e-06,
+      "loss": 0.4315,
+      "step": 952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.627588445309965e-06,
+      "loss": 1.2947,
+      "step": 953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.62541382667965e-06,
+      "loss": 1.0211,
+      "step": 954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6232392080493346e-06,
+      "loss": 1.0548,
+      "step": 955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.62106458941902e-06,
+      "loss": 0.863,
+      "step": 956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.618889970788705e-06,
+      "loss": 0.2999,
+      "step": 957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.61671535215839e-06,
+      "loss": 0.4516,
+      "step": 958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.614540733528076e-06,
+      "loss": 0.5711,
+      "step": 959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.612366114897761e-06,
+      "loss": 0.2973,
+      "step": 960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.610191496267446e-06,
+      "loss": 0.245,
+      "step": 961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.608016877637131e-06,
+      "loss": 0.6518,
+      "step": 962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.605842259006816e-06,
+      "loss": 0.2693,
+      "step": 963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.603667640376502e-06,
+      "loss": 0.5679,
+      "step": 964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.601493021746187e-06,
+      "loss": 0.7892,
+      "step": 965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.599318403115872e-06,
+      "loss": 0.4897,
+      "step": 966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5971437844855566e-06,
+      "loss": 0.3758,
+      "step": 967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5949691658552414e-06,
+      "loss": 1.057,
+      "step": 968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.592794547224927e-06,
+      "loss": 0.5932,
+      "step": 969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.590619928594612e-06,
+      "loss": 0.7129,
+      "step": 970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.588445309964297e-06,
+      "loss": 1.2611,
+      "step": 971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.586270691333983e-06,
+      "loss": 0.7463,
+      "step": 972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5840960727036676e-06,
+      "loss": 0.2757,
+      "step": 973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5819214540733524e-06,
+      "loss": 0.241,
+      "step": 974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.579746835443038e-06,
+      "loss": 0.8623,
+      "step": 975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.577572216812723e-06,
+      "loss": 0.9086,
+      "step": 976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.575397598182409e-06,
+      "loss": 0.3858,
+      "step": 977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.573222979552094e-06,
+      "loss": 0.5411,
+      "step": 978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5710483609217786e-06,
+      "loss": 1.3029,
+      "step": 979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.568873742291464e-06,
+      "loss": 0.7585,
+      "step": 980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.566699123661149e-06,
+      "loss": 0.5282,
+      "step": 981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.564524505030834e-06,
+      "loss": 0.8928,
+      "step": 982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.56234988640052e-06,
+      "loss": 0.6644,
+      "step": 983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.560175267770205e-06,
+      "loss": 0.7858,
+      "step": 984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5580006491398896e-06,
+      "loss": 0.4752,
+      "step": 985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.555826030509575e-06,
+      "loss": 0.5732,
+      "step": 986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.55365141187926e-06,
+      "loss": 0.1952,
+      "step": 987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.551476793248946e-06,
+      "loss": 0.7934,
+      "step": 988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.549302174618631e-06,
+      "loss": 0.8435,
+      "step": 989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.547127555988316e-06,
+      "loss": 0.8742,
+      "step": 990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.544952937358001e-06,
+      "loss": 0.539,
+      "step": 991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.542778318727686e-06,
+      "loss": 0.7286,
+      "step": 992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.540603700097371e-06,
+      "loss": 0.473,
+      "step": 993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.538429081467057e-06,
+      "loss": 0.7313,
+      "step": 994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.536254462836741e-06,
+      "loss": 0.5189,
+      "step": 995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.534079844206427e-06,
+      "loss": 0.5658,
+      "step": 996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5319052255761115e-06,
+      "loss": 0.46,
+      "step": 997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5297306069457964e-06,
+      "loss": 0.5066,
+      "step": 998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.527555988315482e-06,
+      "loss": 0.3376,
+      "step": 999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.525381369685167e-06,
+      "loss": 0.7161,
+      "step": 1000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.523206751054852e-06,
+      "loss": 0.3866,
+      "step": 1001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.521032132424538e-06,
+      "loss": 0.6057,
+      "step": 1002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5188575137942225e-06,
+      "loss": 0.2491,
+      "step": 1003
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.516682895163908e-06,
+      "loss": 0.727,
+      "step": 1004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.514508276533593e-06,
+      "loss": 1.0228,
+      "step": 1005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512333657903278e-06,
+      "loss": 0.8728,
+      "step": 1006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.510159039272964e-06,
+      "loss": 1.4585,
+      "step": 1007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.507984420642649e-06,
+      "loss": 0.6929,
+      "step": 1008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5058098020123335e-06,
+      "loss": 0.6256,
+      "step": 1009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.503635183382019e-06,
+      "loss": 0.7436,
+      "step": 1010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.501460564751704e-06,
+      "loss": 1.1413,
+      "step": 1011
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.499285946121389e-06,
+      "loss": 0.7275,
+      "step": 1012
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.497111327491075e-06,
+      "loss": 0.3424,
+      "step": 1013
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.49493670886076e-06,
+      "loss": 0.4032,
+      "step": 1014
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.492762090230445e-06,
+      "loss": 0.7766,
+      "step": 1015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.49058747160013e-06,
+      "loss": 0.3074,
+      "step": 1016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.488412852969815e-06,
+      "loss": 0.9087,
+      "step": 1017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.486238234339501e-06,
+      "loss": 0.3214,
+      "step": 1018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.484063615709186e-06,
+      "loss": 0.4113,
+      "step": 1019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.481888997078871e-06,
+      "loss": 0.3448,
+      "step": 1020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.479714378448556e-06,
+      "loss": 0.5113,
+      "step": 1021
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.477539759818241e-06,
+      "loss": 0.4696,
+      "step": 1022
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.475365141187926e-06,
+      "loss": 0.5849,
+      "step": 1023
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.473190522557611e-06,
+      "loss": 0.5704,
+      "step": 1024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.471015903927296e-06,
+      "loss": 0.6497,
+      "step": 1025
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.468841285296982e-06,
+      "loss": 0.6146,
+      "step": 1026
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4666666666666665e-06,
+      "loss": 0.7887,
+      "step": 1027
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.6864035087719298,
+      "eval_f1": 0.7183191070256075,
+      "eval_loss": 0.6103039383888245,
+      "eval_precision": 0.6519666269368296,
+      "eval_recall": 0.7997076023391813,
+      "eval_runtime": 230.8484,
+      "eval_samples_per_second": 5.926,
+      "eval_steps_per_second": 1.481,
+      "step": 1027
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.464492048036351e-06,
+      "loss": 0.3109,
+      "step": 1028
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.462317429406037e-06,
+      "loss": 0.9816,
+      "step": 1029
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.460142810775722e-06,
+      "loss": 0.7894,
+      "step": 1030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.457968192145408e-06,
+      "loss": 0.3063,
+      "step": 1031
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.455793573515093e-06,
+      "loss": 0.5769,
+      "step": 1032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4536189548847775e-06,
+      "loss": 0.3828,
+      "step": 1033
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.451444336254463e-06,
+      "loss": 0.4582,
+      "step": 1034
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.449269717624148e-06,
+      "loss": 0.4331,
+      "step": 1035
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.447095098993833e-06,
+      "loss": 0.5129,
+      "step": 1036
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.444920480363519e-06,
+      "loss": 0.3616,
+      "step": 1037
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.442745861733204e-06,
+      "loss": 0.3555,
+      "step": 1038
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4405712431028885e-06,
+      "loss": 0.3799,
+      "step": 1039
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.438396624472574e-06,
+      "loss": 0.8152,
+      "step": 1040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.436222005842259e-06,
+      "loss": 0.2554,
+      "step": 1041
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.434047387211945e-06,
+      "loss": 0.52,
+      "step": 1042
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.43187276858163e-06,
+      "loss": 0.2993,
+      "step": 1043
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.429698149951315e-06,
+      "loss": 0.3516,
+      "step": 1044
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.427523531321e-06,
+      "loss": 1.3647,
+      "step": 1045
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.425348912690685e-06,
+      "loss": 0.2755,
+      "step": 1046
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.42317429406037e-06,
+      "loss": 0.5672,
+      "step": 1047
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.420999675430056e-06,
+      "loss": 0.6835,
+      "step": 1048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.418825056799741e-06,
+      "loss": 0.3674,
+      "step": 1049
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.416650438169426e-06,
+      "loss": 0.1645,
+      "step": 1050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4144758195391105e-06,
+      "loss": 0.382,
+      "step": 1051
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.412301200908795e-06,
+      "loss": 0.1675,
+      "step": 1052
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.410126582278481e-06,
+      "loss": 0.696,
+      "step": 1053
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.407951963648166e-06,
+      "loss": 1.1537,
+      "step": 1054
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.405777345017851e-06,
+      "loss": 0.6824,
+      "step": 1055
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.403602726387537e-06,
+      "loss": 0.2215,
+      "step": 1056
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4014281077572215e-06,
+      "loss": 0.2134,
+      "step": 1057
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.399253489126906e-06,
+      "loss": 0.2081,
+      "step": 1058
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.397078870496592e-06,
+      "loss": 0.6845,
+      "step": 1059
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.394904251866277e-06,
+      "loss": 0.5452,
+      "step": 1060
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.392729633235963e-06,
+      "loss": 0.8074,
+      "step": 1061
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.390555014605648e-06,
+      "loss": 0.1778,
+      "step": 1062
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3883803959753325e-06,
+      "loss": 0.1517,
+      "step": 1063
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.386205777345018e-06,
+      "loss": 0.346,
+      "step": 1064
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.384031158714703e-06,
+      "loss": 0.6213,
+      "step": 1065
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.381856540084388e-06,
+      "loss": 0.7372,
+      "step": 1066
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.379681921454074e-06,
+      "loss": 0.7468,
+      "step": 1067
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.377507302823759e-06,
+      "loss": 0.5814,
+      "step": 1068
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.375332684193444e-06,
+      "loss": 0.7892,
+      "step": 1069
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.373158065563129e-06,
+      "loss": 0.2081,
+      "step": 1070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.370983446932814e-06,
+      "loss": 0.9555,
+      "step": 1071
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.3688088283025e-06,
+      "loss": 0.6907,
+      "step": 1072
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.366634209672185e-06,
+      "loss": 0.8103,
+      "step": 1073
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.36445959104187e-06,
+      "loss": 0.1545,
+      "step": 1074
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.362284972411555e-06,
+      "loss": 1.3006,
+      "step": 1075
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.36011035378124e-06,
+      "loss": 0.7086,
+      "step": 1076
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.357935735150925e-06,
+      "loss": 0.0958,
+      "step": 1077
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.355761116520611e-06,
+      "loss": 0.669,
+      "step": 1078
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.353586497890295e-06,
+      "loss": 1.5062,
+      "step": 1079
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.351411879259981e-06,
+      "loss": 1.2717,
+      "step": 1080
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.3492372606296655e-06,
+      "loss": 0.736,
+      "step": 1081
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.34706264199935e-06,
+      "loss": 0.7876,
+      "step": 1082
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.344888023369036e-06,
+      "loss": 0.1175,
+      "step": 1083
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.342713404738721e-06,
+      "loss": 0.5374,
+      "step": 1084
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.340538786108406e-06,
+      "loss": 0.1301,
+      "step": 1085
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.338364167478092e-06,
+      "loss": 0.1423,
+      "step": 1086
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3361895488477765e-06,
+      "loss": 0.7224,
+      "step": 1087
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.334014930217462e-06,
+      "loss": 0.5453,
+      "step": 1088
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.331840311587147e-06,
+      "loss": 0.5348,
+      "step": 1089
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.329665692956832e-06,
+      "loss": 1.7808,
+      "step": 1090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.327491074326518e-06,
+      "loss": 0.1598,
+      "step": 1091
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.325316455696203e-06,
+      "loss": 0.6849,
+      "step": 1092
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.3231418370658875e-06,
+      "loss": 0.6141,
+      "step": 1093
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.320967218435573e-06,
+      "loss": 0.1993,
+      "step": 1094
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.318792599805258e-06,
+      "loss": 0.6584,
+      "step": 1095
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.316617981174944e-06,
+      "loss": 0.1826,
+      "step": 1096
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.314443362544629e-06,
+      "loss": 0.8769,
+      "step": 1097
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.312268743914314e-06,
+      "loss": 1.0919,
+      "step": 1098
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.310094125283999e-06,
+      "loss": 0.5959,
+      "step": 1099
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.307919506653684e-06,
+      "loss": 0.8602,
+      "step": 1100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.305744888023369e-06,
+      "loss": 1.6468,
+      "step": 1101
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.303570269393055e-06,
+      "loss": 0.6669,
+      "step": 1102
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.30139565076274e-06,
+      "loss": 1.6547,
+      "step": 1103
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.299221032132425e-06,
+      "loss": 0.8603,
+      "step": 1104
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.29704641350211e-06,
+      "loss": 0.2754,
+      "step": 1105
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.294871794871795e-06,
+      "loss": 0.5108,
+      "step": 1106
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.29269717624148e-06,
+      "loss": 0.7941,
+      "step": 1107
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.290522557611165e-06,
+      "loss": 1.0266,
+      "step": 1108
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.28834793898085e-06,
+      "loss": 0.3379,
+      "step": 1109
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.286173320350536e-06,
+      "loss": 0.5952,
+      "step": 1110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.2839987017202205e-06,
+      "loss": 0.6581,
+      "step": 1111
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.281824083089905e-06,
+      "loss": 0.6442,
+      "step": 1112
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.279649464459591e-06,
+      "loss": 1.376,
+      "step": 1113
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.277474845829276e-06,
+      "loss": 0.6033,
+      "step": 1114
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.275300227198962e-06,
+      "loss": 0.2756,
+      "step": 1115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.273125608568647e-06,
+      "loss": 0.1199,
+      "step": 1116
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.2709509899383315e-06,
+      "loss": 1.0733,
+      "step": 1117
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.268776371308017e-06,
+      "loss": 0.5913,
+      "step": 1118
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.266601752677702e-06,
+      "loss": 0.4488,
+      "step": 1119
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.264427134047387e-06,
+      "loss": 1.0088,
+      "step": 1120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.262252515417073e-06,
+      "loss": 0.5103,
+      "step": 1121
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.260077896786758e-06,
+      "loss": 0.1839,
+      "step": 1122
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.257903278156443e-06,
+      "loss": 0.7474,
+      "step": 1123
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.255728659526128e-06,
+      "loss": 0.7708,
+      "step": 1124
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.253554040895813e-06,
+      "loss": 0.3481,
+      "step": 1125
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.251379422265499e-06,
+      "loss": 0.1569,
+      "step": 1126
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.249204803635184e-06,
+      "loss": 0.6777,
+      "step": 1127
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.247030185004869e-06,
+      "loss": 0.503,
+      "step": 1128
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.244855566374554e-06,
+      "loss": 0.8359,
+      "step": 1129
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.242680947744239e-06,
+      "loss": 0.5689,
+      "step": 1130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.240506329113924e-06,
+      "loss": 0.9364,
+      "step": 1131
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.23833171048361e-06,
+      "loss": 0.3391,
+      "step": 1132
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.236157091853295e-06,
+      "loss": 0.8901,
+      "step": 1133
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.2339824732229805e-06,
+      "loss": 0.3277,
+      "step": 1134
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.2318078545926645e-06,
+      "loss": 0.1468,
+      "step": 1135
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.229633235962349e-06,
+      "loss": 0.6114,
+      "step": 1136
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.227458617332035e-06,
+      "loss": 0.2012,
+      "step": 1137
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.22528399870172e-06,
+      "loss": 0.9527,
+      "step": 1138
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.223109380071405e-06,
+      "loss": 0.2938,
+      "step": 1139
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.220934761441091e-06,
+      "loss": 0.6574,
+      "step": 1140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.2187601428107755e-06,
+      "loss": 0.9656,
+      "step": 1141
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.216585524180461e-06,
+      "loss": 0.2354,
+      "step": 1142
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.214410905550146e-06,
+      "loss": 0.6749,
+      "step": 1143
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.212236286919831e-06,
+      "loss": 0.7825,
+      "step": 1144
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.210061668289517e-06,
+      "loss": 0.8963,
+      "step": 1145
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.207887049659202e-06,
+      "loss": 0.1818,
+      "step": 1146
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.2057124310288865e-06,
+      "loss": 0.6603,
+      "step": 1147
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.203537812398572e-06,
+      "loss": 1.4068,
+      "step": 1148
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.201363193768257e-06,
+      "loss": 0.4764,
+      "step": 1149
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.199188575137943e-06,
+      "loss": 1.207,
+      "step": 1150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.197013956507628e-06,
+      "loss": 1.0422,
+      "step": 1151
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.194839337877313e-06,
+      "loss": 0.6252,
+      "step": 1152
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.192664719246998e-06,
+      "loss": 0.3368,
+      "step": 1153
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.190490100616683e-06,
+      "loss": 0.2895,
+      "step": 1154
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.188315481986368e-06,
+      "loss": 0.4954,
+      "step": 1155
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.186140863356054e-06,
+      "loss": 0.2343,
+      "step": 1156
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.183966244725739e-06,
+      "loss": 0.8847,
+      "step": 1157
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.181791626095424e-06,
+      "loss": 0.298,
+      "step": 1158
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.179617007465109e-06,
+      "loss": 0.5894,
+      "step": 1159
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.177442388834794e-06,
+      "loss": 0.7758,
+      "step": 1160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.17526777020448e-06,
+      "loss": 1.3766,
+      "step": 1161
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.173093151574165e-06,
+      "loss": 0.1976,
+      "step": 1162
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.170918532943849e-06,
+      "loss": 0.2468,
+      "step": 1163
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.168743914313535e-06,
+      "loss": 0.9009,
+      "step": 1164
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.1665692956832195e-06,
+      "loss": 0.483,
+      "step": 1165
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.164394677052904e-06,
+      "loss": 0.4599,
+      "step": 1166
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.16222005842259e-06,
+      "loss": 0.4607,
+      "step": 1167
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.160045439792275e-06,
+      "loss": 1.2936,
+      "step": 1168
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.157870821161961e-06,
+      "loss": 0.8442,
+      "step": 1169
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.155696202531646e-06,
+      "loss": 0.5728,
+      "step": 1170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.1535215839013305e-06,
+      "loss": 0.1599,
+      "step": 1171
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.151346965271016e-06,
+      "loss": 0.5717,
+      "step": 1172
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.149172346640701e-06,
+      "loss": 0.4938,
+      "step": 1173
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.146997728010386e-06,
+      "loss": 0.6295,
+      "step": 1174
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.144823109380072e-06,
+      "loss": 0.373,
+      "step": 1175
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.142648490749757e-06,
+      "loss": 0.8001,
+      "step": 1176
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.1404738721194415e-06,
+      "loss": 0.2317,
+      "step": 1177
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.138299253489127e-06,
+      "loss": 0.5385,
+      "step": 1178
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.136124634858812e-06,
+      "loss": 0.1533,
+      "step": 1179
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.133950016228498e-06,
+      "loss": 0.4944,
+      "step": 1180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.131775397598183e-06,
+      "loss": 0.3098,
+      "step": 1181
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.129600778967868e-06,
+      "loss": 1.2013,
+      "step": 1182
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.127426160337553e-06,
+      "loss": 0.5515,
+      "step": 1183
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.125251541707238e-06,
+      "loss": 0.3302,
+      "step": 1184
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.123076923076923e-06,
+      "loss": 1.0706,
+      "step": 1185
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.120902304446609e-06,
+      "loss": 0.6136,
+      "step": 1186
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.118727685816294e-06,
+      "loss": 0.423,
+      "step": 1187
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.1165530671859794e-06,
+      "loss": 1.1232,
+      "step": 1188
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.114378448555664e-06,
+      "loss": 0.8124,
+      "step": 1189
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.112203829925349e-06,
+      "loss": 0.5038,
+      "step": 1190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.110029211295034e-06,
+      "loss": 0.661,
+      "step": 1191
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.107854592664719e-06,
+      "loss": 0.2272,
+      "step": 1192
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.105679974034404e-06,
+      "loss": 0.5822,
+      "step": 1193
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.10350535540409e-06,
+      "loss": 0.5851,
+      "step": 1194
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.1013307367737745e-06,
+      "loss": 0.4518,
+      "step": 1195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.09915611814346e-06,
+      "loss": 1.3874,
+      "step": 1196
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.096981499513145e-06,
+      "loss": 0.3137,
+      "step": 1197
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.09480688088283e-06,
+      "loss": 1.1715,
+      "step": 1198
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.092632262252516e-06,
+      "loss": 0.7938,
+      "step": 1199
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.090457643622201e-06,
+      "loss": 0.5479,
+      "step": 1200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.0882830249918855e-06,
+      "loss": 0.6272,
+      "step": 1201
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.086108406361571e-06,
+      "loss": 0.6056,
+      "step": 1202
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.083933787731256e-06,
+      "loss": 0.4632,
+      "step": 1203
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.081759169100941e-06,
+      "loss": 0.732,
+      "step": 1204
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.079584550470627e-06,
+      "loss": 0.3284,
+      "step": 1205
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.077409931840312e-06,
+      "loss": 1.0211,
+      "step": 1206
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.075235313209997e-06,
+      "loss": 0.9852,
+      "step": 1207
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.073060694579682e-06,
+      "loss": 0.4872,
+      "step": 1208
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.070886075949367e-06,
+      "loss": 0.3784,
+      "step": 1209
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.068711457319053e-06,
+      "loss": 0.6979,
+      "step": 1210
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.066536838688738e-06,
+      "loss": 0.7283,
+      "step": 1211
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.0643622200584226e-06,
+      "loss": 0.2396,
+      "step": 1212
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.062187601428108e-06,
+      "loss": 0.2421,
+      "step": 1213
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.060012982797793e-06,
+      "loss": 0.4355,
+      "step": 1214
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.057838364167479e-06,
+      "loss": 0.6699,
+      "step": 1215
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.055663745537164e-06,
+      "loss": 0.9444,
+      "step": 1216
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.053489126906849e-06,
+      "loss": 1.1541,
+      "step": 1217
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.051314508276534e-06,
+      "loss": 0.1996,
+      "step": 1218
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.0491398896462185e-06,
+      "loss": 0.4871,
+      "step": 1219
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.046965271015903e-06,
+      "loss": 0.441,
+      "step": 1220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.044790652385589e-06,
+      "loss": 0.2213,
+      "step": 1221
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.042616033755274e-06,
+      "loss": 0.6843,
+      "step": 1222
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.04044141512496e-06,
+      "loss": 0.7933,
+      "step": 1223
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.0382667964946446e-06,
+      "loss": 0.5849,
+      "step": 1224
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.0360921778643295e-06,
+      "loss": 0.2845,
+      "step": 1225
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.033917559234015e-06,
+      "loss": 0.7754,
+      "step": 1226
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.0317429406037e-06,
+      "loss": 0.9737,
+      "step": 1227
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.029568321973385e-06,
+      "loss": 0.4848,
+      "step": 1228
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.027393703343071e-06,
+      "loss": 0.4036,
+      "step": 1229
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.0252190847127556e-06,
+      "loss": 1.1795,
+      "step": 1230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.0230444660824405e-06,
+      "loss": 0.2148,
+      "step": 1231
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.020869847452126e-06,
+      "loss": 0.5282,
+      "step": 1232
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.018695228821811e-06,
+      "loss": 0.2426,
+      "step": 1233
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.016520610191497e-06,
+      "loss": 1.0134,
+      "step": 1234
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.014345991561182e-06,
+      "loss": 0.269,
+      "step": 1235
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.0121713729308666e-06,
+      "loss": 0.5035,
+      "step": 1236
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.009996754300552e-06,
+      "loss": 0.9262,
+      "step": 1237
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.007822135670237e-06,
+      "loss": 0.436,
+      "step": 1238
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.005647517039922e-06,
+      "loss": 0.7342,
+      "step": 1239
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.003472898409608e-06,
+      "loss": 0.2399,
+      "step": 1240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.001298279779293e-06,
+      "loss": 0.5475,
+      "step": 1241
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.999123661148978e-06,
+      "loss": 1.3934,
+      "step": 1242
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.996949042518663e-06,
+      "loss": 0.245,
+      "step": 1243
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.994774423888348e-06,
+      "loss": 0.9649,
+      "step": 1244
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.992599805258034e-06,
+      "loss": 0.7621,
+      "step": 1245
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.990425186627719e-06,
+      "loss": 1.3229,
+      "step": 1246
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.988250567997403e-06,
+      "loss": 0.3839,
+      "step": 1247
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9860759493670886e-06,
+      "loss": 0.6381,
+      "step": 1248
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9839013307367734e-06,
+      "loss": 0.6446,
+      "step": 1249
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.981726712106459e-06,
+      "loss": 0.1648,
+      "step": 1250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.979552093476144e-06,
+      "loss": 0.8049,
+      "step": 1251
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.977377474845829e-06,
+      "loss": 0.9902,
+      "step": 1252
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.975202856215515e-06,
+      "loss": 0.236,
+      "step": 1253
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9730282375851996e-06,
+      "loss": 0.589,
+      "step": 1254
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9708536189548844e-06,
+      "loss": 0.519,
+      "step": 1255
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.96867900032457e-06,
+      "loss": 0.2102,
+      "step": 1256
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.966504381694255e-06,
+      "loss": 0.3616,
+      "step": 1257
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.96432976306394e-06,
+      "loss": 0.5435,
+      "step": 1258
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.962155144433626e-06,
+      "loss": 0.8029,
+      "step": 1259
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9599805258033106e-06,
+      "loss": 0.5027,
+      "step": 1260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.957805907172996e-06,
+      "loss": 0.2005,
+      "step": 1261
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.955631288542681e-06,
+      "loss": 0.2904,
+      "step": 1262
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.953456669912366e-06,
+      "loss": 0.8723,
+      "step": 1263
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.951282051282052e-06,
+      "loss": 0.6405,
+      "step": 1264
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.949107432651737e-06,
+      "loss": 0.4869,
+      "step": 1265
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9469328140214216e-06,
+      "loss": 0.4558,
+      "step": 1266
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.944758195391107e-06,
+      "loss": 0.1976,
+      "step": 1267
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.942583576760792e-06,
+      "loss": 0.3155,
+      "step": 1268
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.940408958130478e-06,
+      "loss": 0.3798,
+      "step": 1269
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.938234339500163e-06,
+      "loss": 1.0229,
+      "step": 1270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.936059720869848e-06,
+      "loss": 1.0351,
+      "step": 1271
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.933885102239533e-06,
+      "loss": 1.345,
+      "step": 1272
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.931710483609218e-06,
+      "loss": 0.7735,
+      "step": 1273
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.929535864978903e-06,
+      "loss": 0.443,
+      "step": 1274
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.927361246348588e-06,
+      "loss": 0.9085,
+      "step": 1275
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.925186627718273e-06,
+      "loss": 0.2354,
+      "step": 1276
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.923012009087959e-06,
+      "loss": 0.2617,
+      "step": 1277
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9208373904576435e-06,
+      "loss": 0.5205,
+      "step": 1278
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9186627718273284e-06,
+      "loss": 0.2108,
+      "step": 1279
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.916488153197014e-06,
+      "loss": 0.1443,
+      "step": 1280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.914313534566699e-06,
+      "loss": 0.5232,
+      "step": 1281
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.912138915936384e-06,
+      "loss": 0.8921,
+      "step": 1282
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.90996429730607e-06,
+      "loss": 0.5939,
+      "step": 1283
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9077896786757545e-06,
+      "loss": 0.2656,
+      "step": 1284
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9056150600454394e-06,
+      "loss": 0.2377,
+      "step": 1285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.903440441415125e-06,
+      "loss": 1.1884,
+      "step": 1286
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.90126582278481e-06,
+      "loss": 0.1427,
+      "step": 1287
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.899091204154496e-06,
+      "loss": 0.768,
+      "step": 1288
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.896916585524181e-06,
+      "loss": 1.2024,
+      "step": 1289
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8947419668938655e-06,
+      "loss": 0.5338,
+      "step": 1290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.892567348263551e-06,
+      "loss": 1.2169,
+      "step": 1291
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.890392729633236e-06,
+      "loss": 0.1238,
+      "step": 1292
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.888218111002921e-06,
+      "loss": 0.6943,
+      "step": 1293
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.886043492372607e-06,
+      "loss": 0.4839,
+      "step": 1294
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.883868873742292e-06,
+      "loss": 0.5613,
+      "step": 1295
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8816942551119765e-06,
+      "loss": 1.2474,
+      "step": 1296
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.879519636481662e-06,
+      "loss": 0.6765,
+      "step": 1297
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.877345017851347e-06,
+      "loss": 1.2505,
+      "step": 1298
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.875170399221033e-06,
+      "loss": 0.3931,
+      "step": 1299
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.872995780590718e-06,
+      "loss": 0.4087,
+      "step": 1300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.870821161960403e-06,
+      "loss": 0.4034,
+      "step": 1301
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.868646543330088e-06,
+      "loss": 0.7378,
+      "step": 1302
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.866471924699772e-06,
+      "loss": 0.5157,
+      "step": 1303
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.864297306069458e-06,
+      "loss": 0.305,
+      "step": 1304
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.862122687439143e-06,
+      "loss": 1.2742,
+      "step": 1305
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.859948068808828e-06,
+      "loss": 0.4423,
+      "step": 1306
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.857773450178514e-06,
+      "loss": 0.2712,
+      "step": 1307
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.8555988315481985e-06,
+      "loss": 0.4714,
+      "step": 1308
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.853424212917883e-06,
+      "loss": 0.2223,
+      "step": 1309
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.851249594287569e-06,
+      "loss": 0.756,
+      "step": 1310
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.849074975657254e-06,
+      "loss": 0.5301,
+      "step": 1311
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.846900357026939e-06,
+      "loss": 0.1789,
+      "step": 1312
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.844725738396625e-06,
+      "loss": 0.8603,
+      "step": 1313
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.8425511197663095e-06,
+      "loss": 1.1457,
+      "step": 1314
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.840376501135995e-06,
+      "loss": 0.4741,
+      "step": 1315
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.83820188250568e-06,
+      "loss": 0.4679,
+      "step": 1316
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.836027263875365e-06,
+      "loss": 0.196,
+      "step": 1317
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.833852645245051e-06,
+      "loss": 1.7679,
+      "step": 1318
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.831678026614736e-06,
+      "loss": 0.3608,
+      "step": 1319
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8295034079844205e-06,
+      "loss": 0.2501,
+      "step": 1320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.827328789354106e-06,
+      "loss": 0.6238,
+      "step": 1321
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.825154170723791e-06,
+      "loss": 0.5304,
+      "step": 1322
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.822979552093476e-06,
+      "loss": 0.2932,
+      "step": 1323
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.820804933463162e-06,
+      "loss": 0.9292,
+      "step": 1324
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.818630314832847e-06,
+      "loss": 0.4682,
+      "step": 1325
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.816455696202532e-06,
+      "loss": 0.4445,
+      "step": 1326
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8142810775722173e-06,
+      "loss": 0.1804,
+      "step": 1327
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8121064589419026e-06,
+      "loss": 0.1695,
+      "step": 1328
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8099318403115875e-06,
+      "loss": 0.1736,
+      "step": 1329
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8077572216812728e-06,
+      "loss": 0.6086,
+      "step": 1330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8055826030509572e-06,
+      "loss": 0.2189,
+      "step": 1331
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8034079844206425e-06,
+      "loss": 0.6228,
+      "step": 1332
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.801233365790328e-06,
+      "loss": 0.6557,
+      "step": 1333
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7990587471600127e-06,
+      "loss": 0.1989,
+      "step": 1334
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.796884128529698e-06,
+      "loss": 0.5742,
+      "step": 1335
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7947095098993833e-06,
+      "loss": 0.5231,
+      "step": 1336
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7925348912690682e-06,
+      "loss": 0.4706,
+      "step": 1337
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7903602726387535e-06,
+      "loss": 0.2492,
+      "step": 1338
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.788185654008439e-06,
+      "loss": 0.5333,
+      "step": 1339
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.786011035378124e-06,
+      "loss": 0.3096,
+      "step": 1340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.783836416747809e-06,
+      "loss": 0.1651,
+      "step": 1341
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7816617981174943e-06,
+      "loss": 0.5509,
+      "step": 1342
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7794871794871796e-06,
+      "loss": 1.0949,
+      "step": 1343
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.777312560856865e-06,
+      "loss": 0.3063,
+      "step": 1344
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.77513794222655e-06,
+      "loss": 0.7522,
+      "step": 1345
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.772963323596235e-06,
+      "loss": 1.0505,
+      "step": 1346
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7707887049659204e-06,
+      "loss": 0.6873,
+      "step": 1347
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7686140863356058e-06,
+      "loss": 0.6062,
+      "step": 1348
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7664394677052906e-06,
+      "loss": 0.8778,
+      "step": 1349
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.764264849074976e-06,
+      "loss": 0.6642,
+      "step": 1350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7620902304446612e-06,
+      "loss": 0.4283,
+      "step": 1351
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.759915611814346e-06,
+      "loss": 0.6805,
+      "step": 1352
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7577409931840314e-06,
+      "loss": 0.2482,
+      "step": 1353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7555663745537167e-06,
+      "loss": 0.5557,
+      "step": 1354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.753391755923402e-06,
+      "loss": 1.0391,
+      "step": 1355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.751217137293087e-06,
+      "loss": 1.0231,
+      "step": 1356
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7490425186627722e-06,
+      "loss": 0.3068,
+      "step": 1357
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7468679000324576e-06,
+      "loss": 0.591,
+      "step": 1358
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.744693281402142e-06,
+      "loss": 0.5564,
+      "step": 1359
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7425186627718273e-06,
+      "loss": 0.2124,
+      "step": 1360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.740344044141512e-06,
+      "loss": 0.6233,
+      "step": 1361
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7381694255111975e-06,
+      "loss": 0.5927,
+      "step": 1362
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.735994806880883e-06,
+      "loss": 0.2302,
+      "step": 1363
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7338201882505677e-06,
+      "loss": 0.7667,
+      "step": 1364
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.731645569620253e-06,
+      "loss": 1.0514,
+      "step": 1365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7294709509899383e-06,
+      "loss": 1.2216,
+      "step": 1366
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7272963323596236e-06,
+      "loss": 0.9763,
+      "step": 1367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7251217137293085e-06,
+      "loss": 1.2813,
+      "step": 1368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.722947095098994e-06,
+      "loss": 0.4641,
+      "step": 1369
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.720772476468679e-06,
+      "loss": 0.4489,
+      "step": 1370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7185978578383644e-06,
+      "loss": 0.5388,
+      "step": 1371
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7164232392080493e-06,
+      "loss": 0.7024,
+      "step": 1372
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7142486205777346e-06,
+      "loss": 0.6539,
+      "step": 1373
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.71207400194742e-06,
+      "loss": 0.4965,
+      "step": 1374
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7098993833171052e-06,
+      "loss": 0.6814,
+      "step": 1375
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.70772476468679e-06,
+      "loss": 0.5997,
+      "step": 1376
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7055501460564754e-06,
+      "loss": 0.9071,
+      "step": 1377
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7033755274261607e-06,
+      "loss": 0.4281,
+      "step": 1378
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7012009087958456e-06,
+      "loss": 0.1655,
+      "step": 1379
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.699026290165531e-06,
+      "loss": 0.628,
+      "step": 1380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6968516715352162e-06,
+      "loss": 0.3321,
+      "step": 1381
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6946770529049015e-06,
+      "loss": 0.7109,
+      "step": 1382
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6925024342745864e-06,
+      "loss": 0.7407,
+      "step": 1383
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6903278156442717e-06,
+      "loss": 0.9274,
+      "step": 1384
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.688153197013957e-06,
+      "loss": 0.6377,
+      "step": 1385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6859785783836424e-06,
+      "loss": 0.7968,
+      "step": 1386
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.683803959753327e-06,
+      "loss": 0.3497,
+      "step": 1387
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6816293411230117e-06,
+      "loss": 0.6647,
+      "step": 1388
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.679454722492697e-06,
+      "loss": 0.3384,
+      "step": 1389
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6772801038623823e-06,
+      "loss": 0.2381,
+      "step": 1390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.675105485232067e-06,
+      "loss": 0.4743,
+      "step": 1391
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6729308666017525e-06,
+      "loss": 0.4611,
+      "step": 1392
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.670756247971438e-06,
+      "loss": 1.2117,
+      "step": 1393
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.668581629341123e-06,
+      "loss": 0.1927,
+      "step": 1394
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.666407010710808e-06,
+      "loss": 0.4114,
+      "step": 1395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6642323920804933e-06,
+      "loss": 0.1868,
+      "step": 1396
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6620577734501786e-06,
+      "loss": 0.5862,
+      "step": 1397
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.659883154819864e-06,
+      "loss": 0.3172,
+      "step": 1398
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.657708536189549e-06,
+      "loss": 1.0954,
+      "step": 1399
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.655533917559234e-06,
+      "loss": 0.5847,
+      "step": 1400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6533592989289194e-06,
+      "loss": 0.5897,
+      "step": 1401
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6511846802986043e-06,
+      "loss": 0.2345,
+      "step": 1402
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6490100616682896e-06,
+      "loss": 0.4657,
+      "step": 1403
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.646835443037975e-06,
+      "loss": 0.8087,
+      "step": 1404
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6446608244076602e-06,
+      "loss": 0.4527,
+      "step": 1405
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.642486205777345e-06,
+      "loss": 0.3459,
+      "step": 1406
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6403115871470304e-06,
+      "loss": 0.636,
+      "step": 1407
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6381369685167157e-06,
+      "loss": 0.2535,
+      "step": 1408
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.635962349886401e-06,
+      "loss": 0.7273,
+      "step": 1409
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.633787731256086e-06,
+      "loss": 0.5343,
+      "step": 1410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6316131126257712e-06,
+      "loss": 0.25,
+      "step": 1411
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6294384939954565e-06,
+      "loss": 0.5694,
+      "step": 1412
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.627263875365142e-06,
+      "loss": 0.1905,
+      "step": 1413
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6250892567348267e-06,
+      "loss": 1.6782,
+      "step": 1414
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.622914638104511e-06,
+      "loss": 0.9065,
+      "step": 1415
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6207400194741965e-06,
+      "loss": 0.5423,
+      "step": 1416
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.618565400843882e-06,
+      "loss": 0.5107,
+      "step": 1417
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6163907822135667e-06,
+      "loss": 0.5791,
+      "step": 1418
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.614216163583252e-06,
+      "loss": 1.2505,
+      "step": 1419
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6120415449529373e-06,
+      "loss": 0.6963,
+      "step": 1420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6098669263226226e-06,
+      "loss": 0.2916,
+      "step": 1421
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6076923076923075e-06,
+      "loss": 0.9528,
+      "step": 1422
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.605517689061993e-06,
+      "loss": 0.7178,
+      "step": 1423
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.603343070431678e-06,
+      "loss": 0.2396,
+      "step": 1424
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6011684518013634e-06,
+      "loss": 0.2137,
+      "step": 1425
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.5989938331710483e-06,
+      "loss": 0.704,
+      "step": 1426
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.5968192145407336e-06,
+      "loss": 1.0245,
+      "step": 1427
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.594644595910419e-06,
+      "loss": 0.4829,
+      "step": 1428
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.592469977280104e-06,
+      "loss": 0.5998,
+      "step": 1429
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.590295358649789e-06,
+      "loss": 0.5391,
+      "step": 1430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.5881207400194744e-06,
+      "loss": 0.6741,
+      "step": 1431
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.5859461213891597e-06,
+      "loss": 0.3476,
+      "step": 1432
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5837715027588446e-06,
+      "loss": 0.586,
+      "step": 1433
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.58159688412853e-06,
+      "loss": 0.5487,
+      "step": 1434
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5794222654982152e-06,
+      "loss": 0.7509,
+      "step": 1435
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5772476468679005e-06,
+      "loss": 0.7132,
+      "step": 1436
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5750730282375854e-06,
+      "loss": 0.251,
+      "step": 1437
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5728984096072707e-06,
+      "loss": 1.2891,
+      "step": 1438
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.570723790976956e-06,
+      "loss": 0.4756,
+      "step": 1439
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5685491723466413e-06,
+      "loss": 0.6359,
+      "step": 1440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5663745537163262e-06,
+      "loss": 0.2223,
+      "step": 1441
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5641999350860115e-06,
+      "loss": 0.4016,
+      "step": 1442
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.562025316455696e-06,
+      "loss": 0.5692,
+      "step": 1443
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5598506978253813e-06,
+      "loss": 0.2999,
+      "step": 1444
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.557676079195066e-06,
+      "loss": 0.6773,
+      "step": 1445
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5555014605647515e-06,
+      "loss": 0.3295,
+      "step": 1446
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5533268419344368e-06,
+      "loss": 0.6061,
+      "step": 1447
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.551152223304122e-06,
+      "loss": 0.4883,
+      "step": 1448
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.548977604673807e-06,
+      "loss": 0.7853,
+      "step": 1449
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5468029860434923e-06,
+      "loss": 0.7866,
+      "step": 1450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5446283674131776e-06,
+      "loss": 1.152,
+      "step": 1451
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.542453748782863e-06,
+      "loss": 0.7537,
+      "step": 1452
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5402791301525478e-06,
+      "loss": 0.7868,
+      "step": 1453
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.538104511522233e-06,
+      "loss": 0.2184,
+      "step": 1454
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5359298928919184e-06,
+      "loss": 0.5216,
+      "step": 1455
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5337552742616033e-06,
+      "loss": 0.7088,
+      "step": 1456
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5315806556312886e-06,
+      "loss": 0.6644,
+      "step": 1457
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.529406037000974e-06,
+      "loss": 0.9646,
+      "step": 1458
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.527231418370659e-06,
+      "loss": 0.4316,
+      "step": 1459
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.525056799740344e-06,
+      "loss": 0.6176,
+      "step": 1460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5228821811100294e-06,
+      "loss": 0.3861,
+      "step": 1461
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5207075624797147e-06,
+      "loss": 0.5541,
+      "step": 1462
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5185329438494e-06,
+      "loss": 0.2034,
+      "step": 1463
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.516358325219085e-06,
+      "loss": 0.6132,
+      "step": 1464
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.51418370658877e-06,
+      "loss": 0.529,
+      "step": 1465
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5120090879584555e-06,
+      "loss": 0.8352,
+      "step": 1466
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.509834469328141e-06,
+      "loss": 1.0191,
+      "step": 1467
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5076598506978257e-06,
+      "loss": 0.5831,
+      "step": 1468
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.505485232067511e-06,
+      "loss": 0.605,
+      "step": 1469
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5033106134371963e-06,
+      "loss": 0.3338,
+      "step": 1470
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5011359948068808e-06,
+      "loss": 0.2905,
+      "step": 1471
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4989613761765657e-06,
+      "loss": 0.8946,
+      "step": 1472
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.496786757546251e-06,
+      "loss": 0.4814,
+      "step": 1473
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4946121389159363e-06,
+      "loss": 0.2576,
+      "step": 1474
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4924375202856216e-06,
+      "loss": 0.6808,
+      "step": 1475
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4902629016553065e-06,
+      "loss": 0.3909,
+      "step": 1476
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4880882830249918e-06,
+      "loss": 0.8056,
+      "step": 1477
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.485913664394677e-06,
+      "loss": 0.6118,
+      "step": 1478
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4837390457643624e-06,
+      "loss": 1.1933,
+      "step": 1479
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4815644271340473e-06,
+      "loss": 0.5352,
+      "step": 1480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4793898085037326e-06,
+      "loss": 0.1758,
+      "step": 1481
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.477215189873418e-06,
+      "loss": 0.9582,
+      "step": 1482
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4750405712431028e-06,
+      "loss": 0.3496,
+      "step": 1483
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.472865952612788e-06,
+      "loss": 0.6701,
+      "step": 1484
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4706913339824734e-06,
+      "loss": 0.4354,
+      "step": 1485
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4685167153521587e-06,
+      "loss": 0.5563,
+      "step": 1486
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4663420967218436e-06,
+      "loss": 0.337,
+      "step": 1487
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.464167478091529e-06,
+      "loss": 0.688,
+      "step": 1488
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.461992859461214e-06,
+      "loss": 0.9607,
+      "step": 1489
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4598182408308995e-06,
+      "loss": 1.023,
+      "step": 1490
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4576436222005844e-06,
+      "loss": 0.3413,
+      "step": 1491
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4554690035702697e-06,
+      "loss": 0.6795,
+      "step": 1492
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.453294384939955e-06,
+      "loss": 0.3687,
+      "step": 1493
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4511197663096403e-06,
+      "loss": 0.1841,
+      "step": 1494
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.448945147679325e-06,
+      "loss": 1.0469,
+      "step": 1495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4467705290490105e-06,
+      "loss": 0.7369,
+      "step": 1496
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.444595910418696e-06,
+      "loss": 0.5249,
+      "step": 1497
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4424212917883807e-06,
+      "loss": 0.3473,
+      "step": 1498
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.440246673158065e-06,
+      "loss": 0.2815,
+      "step": 1499
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4380720545277505e-06,
+      "loss": 0.2844,
+      "step": 1500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4358974358974358e-06,
+      "loss": 0.3991,
+      "step": 1501
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.433722817267121e-06,
+      "loss": 1.4901,
+      "step": 1502
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.431548198636806e-06,
+      "loss": 0.9676,
+      "step": 1503
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4293735800064913e-06,
+      "loss": 0.2063,
+      "step": 1504
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4271989613761766e-06,
+      "loss": 0.2396,
+      "step": 1505
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.425024342745862e-06,
+      "loss": 0.7507,
+      "step": 1506
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4228497241155468e-06,
+      "loss": 0.5016,
+      "step": 1507
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.420675105485232e-06,
+      "loss": 0.577,
+      "step": 1508
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4185004868549174e-06,
+      "loss": 0.7748,
+      "step": 1509
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4163258682246023e-06,
+      "loss": 1.3372,
+      "step": 1510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4141512495942876e-06,
+      "loss": 0.1896,
+      "step": 1511
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.411976630963973e-06,
+      "loss": 0.1619,
+      "step": 1512
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.409802012333658e-06,
+      "loss": 0.2978,
+      "step": 1513
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.407627393703343e-06,
+      "loss": 0.8033,
+      "step": 1514
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4054527750730284e-06,
+      "loss": 0.3456,
+      "step": 1515
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4032781564427137e-06,
+      "loss": 0.6564,
+      "step": 1516
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.401103537812399e-06,
+      "loss": 1.2641,
+      "step": 1517
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.398928919182084e-06,
+      "loss": 0.3201,
+      "step": 1518
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.396754300551769e-06,
+      "loss": 0.4,
+      "step": 1519
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3945796819214545e-06,
+      "loss": 0.5989,
+      "step": 1520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3924050632911394e-06,
+      "loss": 0.6138,
+      "step": 1521
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3902304446608247e-06,
+      "loss": 0.2408,
+      "step": 1522
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.38805582603051e-06,
+      "loss": 0.2512,
+      "step": 1523
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3858812074001953e-06,
+      "loss": 0.2285,
+      "step": 1524
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.38370658876988e-06,
+      "loss": 0.1957,
+      "step": 1525
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3815319701395655e-06,
+      "loss": 1.0675,
+      "step": 1526
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.37935735150925e-06,
+      "loss": 1.0087,
+      "step": 1527
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3771827328789353e-06,
+      "loss": 0.4913,
+      "step": 1528
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3750081142486206e-06,
+      "loss": 0.2528,
+      "step": 1529
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3728334956183054e-06,
+      "loss": 1.1833,
+      "step": 1530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3706588769879908e-06,
+      "loss": 0.4546,
+      "step": 1531
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.368484258357676e-06,
+      "loss": 0.3776,
+      "step": 1532
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3663096397273614e-06,
+      "loss": 0.9789,
+      "step": 1533
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3641350210970463e-06,
+      "loss": 0.3841,
+      "step": 1534
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.3619604024667316e-06,
+      "loss": 0.6061,
+      "step": 1535
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.359785783836417e-06,
+      "loss": 0.137,
+      "step": 1536
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.3576111652061017e-06,
+      "loss": 0.7307,
+      "step": 1537
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.355436546575787e-06,
+      "loss": 0.3985,
+      "step": 1538
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.3532619279454724e-06,
+      "loss": 0.139,
+      "step": 1539
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.3510873093151577e-06,
+      "loss": 0.8653,
+      "step": 1540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.3489126906848426e-06,
+      "loss": 0.2781,
+      "step": 1541
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.346738072054528e-06,
+      "loss": 0.6494,
+      "step": 1542
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.344563453424213e-06,
+      "loss": 0.6549,
+      "step": 1543
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.3423888347938985e-06,
+      "loss": 1.0516,
+      "step": 1544
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.3402142161635834e-06,
+      "loss": 0.2252,
+      "step": 1545
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3380395975332687e-06,
+      "loss": 0.3622,
+      "step": 1546
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.335864978902954e-06,
+      "loss": 1.1243,
+      "step": 1547
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.333690360272639e-06,
+      "loss": 0.7168,
+      "step": 1548
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3315157416423237e-06,
+      "loss": 1.5494,
+      "step": 1549
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.329341123012009e-06,
+      "loss": 0.9929,
+      "step": 1550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3271665043816944e-06,
+      "loss": 0.2277,
+      "step": 1551
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3249918857513797e-06,
+      "loss": 0.4799,
+      "step": 1552
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.3228172671210646e-06,
+      "loss": 0.8569,
+      "step": 1553
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.32064264849075e-06,
+      "loss": 0.9625,
+      "step": 1554
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.318468029860435e-06,
+      "loss": 0.2683,
+      "step": 1555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.31629341123012e-06,
+      "loss": 0.4818,
+      "step": 1556
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3141187925998054e-06,
+      "loss": 0.3251,
+      "step": 1557
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3119441739694907e-06,
+      "loss": 0.1598,
+      "step": 1558
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.309769555339176e-06,
+      "loss": 0.4353,
+      "step": 1559
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.307594936708861e-06,
+      "loss": 0.7073,
+      "step": 1560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.305420318078546e-06,
+      "loss": 1.0136,
+      "step": 1561
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.303245699448231e-06,
+      "loss": 1.607,
+      "step": 1562
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.3010710808179164e-06,
+      "loss": 0.5151,
+      "step": 1563
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2988964621876012e-06,
+      "loss": 0.1699,
+      "step": 1564
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.2967218435572865e-06,
+      "loss": 0.1774,
+      "step": 1565
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.294547224926972e-06,
+      "loss": 0.3715,
+      "step": 1566
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.292372606296657e-06,
+      "loss": 0.1276,
+      "step": 1567
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.290197987666342e-06,
+      "loss": 1.4657,
+      "step": 1568
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2880233690360274e-06,
+      "loss": 0.8251,
+      "step": 1569
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2858487504057127e-06,
+      "loss": 1.4002,
+      "step": 1570
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.283674131775398e-06,
+      "loss": 0.8527,
+      "step": 1571
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.281499513145083e-06,
+      "loss": 0.645,
+      "step": 1572
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.279324894514768e-06,
+      "loss": 0.2342,
+      "step": 1573
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2771502758844535e-06,
+      "loss": 0.2607,
+      "step": 1574
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2749756572541383e-06,
+      "loss": 0.1804,
+      "step": 1575
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.2728010386238232e-06,
+      "loss": 0.1924,
+      "step": 1576
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.2706264199935085e-06,
+      "loss": 0.1989,
+      "step": 1577
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.268451801363194e-06,
+      "loss": 0.8006,
+      "step": 1578
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.266277182732879e-06,
+      "loss": 1.3682,
+      "step": 1579
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.264102564102564e-06,
+      "loss": 1.0342,
+      "step": 1580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.2619279454722493e-06,
+      "loss": 0.1939,
+      "step": 1581
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.2597533268419347e-06,
+      "loss": 0.2651,
+      "step": 1582
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.2575787082116195e-06,
+      "loss": 0.7167,
+      "step": 1583
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.255404089581305e-06,
+      "loss": 0.9601,
+      "step": 1584
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.25322947095099e-06,
+      "loss": 0.3151,
+      "step": 1585
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.2510548523206755e-06,
+      "loss": 0.3832,
+      "step": 1586
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.2488802336903603e-06,
+      "loss": 0.1447,
+      "step": 1587
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.2467056150600457e-06,
+      "loss": 0.1678,
+      "step": 1588
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.244530996429731e-06,
+      "loss": 0.125,
+      "step": 1589
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.242356377799416e-06,
+      "loss": 0.1166,
+      "step": 1590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.2401817591691007e-06,
+      "loss": 0.2747,
+      "step": 1591
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.238007140538786e-06,
+      "loss": 1.1748,
+      "step": 1592
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.2358325219084713e-06,
+      "loss": 0.8016,
+      "step": 1593
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.2336579032781566e-06,
+      "loss": 0.928,
+      "step": 1594
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.2314832846478415e-06,
+      "loss": 0.4366,
+      "step": 1595
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.229308666017527e-06,
+      "loss": 0.827,
+      "step": 1596
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.227134047387212e-06,
+      "loss": 0.1264,
+      "step": 1597
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.2249594287568975e-06,
+      "loss": 0.419,
+      "step": 1598
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.2227848101265823e-06,
+      "loss": 0.6697,
+      "step": 1599
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.2206101914962676e-06,
+      "loss": 0.5792,
+      "step": 1600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.218435572865953e-06,
+      "loss": 1.0438,
+      "step": 1601
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.216260954235638e-06,
+      "loss": 1.0783,
+      "step": 1602
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.214086335605323e-06,
+      "loss": 0.7875,
+      "step": 1603
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.211911716975008e-06,
+      "loss": 1.6148,
+      "step": 1604
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.2097370983446933e-06,
+      "loss": 0.7418,
+      "step": 1605
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.2075624797143786e-06,
+      "loss": 0.7094,
+      "step": 1606
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.2053878610840635e-06,
+      "loss": 0.1046,
+      "step": 1607
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.203213242453749e-06,
+      "loss": 0.687,
+      "step": 1608
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.201038623823434e-06,
+      "loss": 0.1054,
+      "step": 1609
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.198864005193119e-06,
+      "loss": 0.2989,
+      "step": 1610
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.1966893865628043e-06,
+      "loss": 0.1174,
+      "step": 1611
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.1945147679324896e-06,
+      "loss": 0.8367,
+      "step": 1612
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.192340149302175e-06,
+      "loss": 0.7503,
+      "step": 1613
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.19016553067186e-06,
+      "loss": 0.528,
+      "step": 1614
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.187990912041545e-06,
+      "loss": 1.0772,
+      "step": 1615
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.1858162934112304e-06,
+      "loss": 0.5713,
+      "step": 1616
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.1836416747809158e-06,
+      "loss": 0.5166,
+      "step": 1617
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.1814670561506002e-06,
+      "loss": 0.3163,
+      "step": 1618
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.1792924375202855e-06,
+      "loss": 1.1754,
+      "step": 1619
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.177117818889971e-06,
+      "loss": 0.1819,
+      "step": 1620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.174943200259656e-06,
+      "loss": 0.6786,
+      "step": 1621
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.172768581629341e-06,
+      "loss": 0.5654,
+      "step": 1622
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.1705939629990263e-06,
+      "loss": 0.1095,
+      "step": 1623
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.1684193443687116e-06,
+      "loss": 0.1173,
+      "step": 1624
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.166244725738397e-06,
+      "loss": 0.2153,
+      "step": 1625
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.164070107108082e-06,
+      "loss": 0.1321,
+      "step": 1626
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.161895488477767e-06,
+      "loss": 0.3468,
+      "step": 1627
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1597208698474524e-06,
+      "loss": 0.8013,
+      "step": 1628
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1575462512171373e-06,
+      "loss": 0.1415,
+      "step": 1629
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1553716325868226e-06,
+      "loss": 0.7184,
+      "step": 1630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.153197013956508e-06,
+      "loss": 0.8302,
+      "step": 1631
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.151022395326193e-06,
+      "loss": 0.6137,
+      "step": 1632
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1488477766958777e-06,
+      "loss": 0.5272,
+      "step": 1633
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.146673158065563e-06,
+      "loss": 0.8411,
+      "step": 1634
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1444985394352483e-06,
+      "loss": 0.8241,
+      "step": 1635
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1423239208049336e-06,
+      "loss": 1.6225,
+      "step": 1636
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1401493021746185e-06,
+      "loss": 0.2872,
+      "step": 1637
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.137974683544304e-06,
+      "loss": 0.7054,
+      "step": 1638
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.135800064913989e-06,
+      "loss": 0.8773,
+      "step": 1639
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1336254462836744e-06,
+      "loss": 1.3413,
+      "step": 1640
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1314508276533593e-06,
+      "loss": 1.3213,
+      "step": 1641
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1292762090230446e-06,
+      "loss": 0.2019,
+      "step": 1642
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.12710159039273e-06,
+      "loss": 0.223,
+      "step": 1643
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1249269717624152e-06,
+      "loss": 0.1516,
+      "step": 1644
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1227523531321e-06,
+      "loss": 1.1001,
+      "step": 1645
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.120577734501785e-06,
+      "loss": 0.2215,
+      "step": 1646
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1184031158714703e-06,
+      "loss": 0.969,
+      "step": 1647
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1162284972411556e-06,
+      "loss": 1.7772,
+      "step": 1648
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1140538786108405e-06,
+      "loss": 1.1904,
+      "step": 1649
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.111879259980526e-06,
+      "loss": 0.7227,
+      "step": 1650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.109704641350211e-06,
+      "loss": 0.3042,
+      "step": 1651
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1075300227198964e-06,
+      "loss": 0.7751,
+      "step": 1652
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1053554040895813e-06,
+      "loss": 0.9768,
+      "step": 1653
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.1031807854592666e-06,
+      "loss": 1.1337,
+      "step": 1654
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.101006166828952e-06,
+      "loss": 0.8491,
+      "step": 1655
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.098831548198637e-06,
+      "loss": 0.216,
+      "step": 1656
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.096656929568322e-06,
+      "loss": 0.5896,
+      "step": 1657
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0944823109380074e-06,
+      "loss": 1.8013,
+      "step": 1658
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0923076923076927e-06,
+      "loss": 0.3476,
+      "step": 1659
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.090133073677377e-06,
+      "loss": 0.4969,
+      "step": 1660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0879584550470625e-06,
+      "loss": 0.1101,
+      "step": 1661
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.085783836416748e-06,
+      "loss": 0.1768,
+      "step": 1662
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.083609217786433e-06,
+      "loss": 0.4508,
+      "step": 1663
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.081434599156118e-06,
+      "loss": 0.5797,
+      "step": 1664
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0792599805258033e-06,
+      "loss": 0.7574,
+      "step": 1665
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0770853618954886e-06,
+      "loss": 0.1353,
+      "step": 1666
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.074910743265174e-06,
+      "loss": 0.1331,
+      "step": 1667
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.072736124634859e-06,
+      "loss": 0.715,
+      "step": 1668
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.070561506004544e-06,
+      "loss": 0.226,
+      "step": 1669
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0683868873742294e-06,
+      "loss": 0.6703,
+      "step": 1670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0662122687439147e-06,
+      "loss": 0.9339,
+      "step": 1671
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0640376501135996e-06,
+      "loss": 0.5263,
+      "step": 1672
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.061863031483285e-06,
+      "loss": 0.6599,
+      "step": 1673
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.05968841285297e-06,
+      "loss": 0.9768,
+      "step": 1674
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.057513794222655e-06,
+      "loss": 0.4664,
+      "step": 1675
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.05533917559234e-06,
+      "loss": 0.9366,
+      "step": 1676
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0531645569620253e-06,
+      "loss": 0.4389,
+      "step": 1677
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0509899383317106e-06,
+      "loss": 0.1882,
+      "step": 1678
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.048815319701396e-06,
+      "loss": 0.5649,
+      "step": 1679
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.046640701071081e-06,
+      "loss": 0.3318,
+      "step": 1680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.044466082440766e-06,
+      "loss": 0.6915,
+      "step": 1681
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.0422914638104514e-06,
+      "loss": 0.4993,
+      "step": 1682
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.0401168451801363e-06,
+      "loss": 0.3083,
+      "step": 1683
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.0379422265498216e-06,
+      "loss": 0.8945,
+      "step": 1684
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.035767607919507e-06,
+      "loss": 0.6171,
+      "step": 1685
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.0335929892891922e-06,
+      "loss": 0.2228,
+      "step": 1686
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.031418370658877e-06,
+      "loss": 1.2121,
+      "step": 1687
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.029243752028562e-06,
+      "loss": 0.7469,
+      "step": 1688
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.0270691333982473e-06,
+      "loss": 1.1014,
+      "step": 1689
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0248945147679326e-06,
+      "loss": 0.6515,
+      "step": 1690
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0227198961376175e-06,
+      "loss": 0.1378,
+      "step": 1691
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.020545277507303e-06,
+      "loss": 0.6735,
+      "step": 1692
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.018370658876988e-06,
+      "loss": 0.2607,
+      "step": 1693
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0161960402466734e-06,
+      "loss": 0.278,
+      "step": 1694
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0140214216163583e-06,
+      "loss": 0.8057,
+      "step": 1695
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0118468029860436e-06,
+      "loss": 0.1862,
+      "step": 1696
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.009672184355729e-06,
+      "loss": 0.7058,
+      "step": 1697
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0074975657254142e-06,
+      "loss": 0.8939,
+      "step": 1698
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.005322947095099e-06,
+      "loss": 0.6033,
+      "step": 1699
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.0031483284647844e-06,
+      "loss": 0.8517,
+      "step": 1700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.0009737098344697e-06,
+      "loss": 0.3663,
+      "step": 1701
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9987990912041546e-06,
+      "loss": 1.4332,
+      "step": 1702
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9966244725738395e-06,
+      "loss": 0.7101,
+      "step": 1703
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.994449853943525e-06,
+      "loss": 0.58,
+      "step": 1704
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.99227523531321e-06,
+      "loss": 0.5542,
+      "step": 1705
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.990100616682895e-06,
+      "loss": 0.8286,
+      "step": 1706
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9879259980525803e-06,
+      "loss": 0.5152,
+      "step": 1707
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9857513794222656e-06,
+      "loss": 0.5466,
+      "step": 1708
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.983576760791951e-06,
+      "loss": 0.5067,
+      "step": 1709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.981402142161636e-06,
+      "loss": 0.4532,
+      "step": 1710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.979227523531321e-06,
+      "loss": 1.3259,
+      "step": 1711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9770529049010064e-06,
+      "loss": 0.7623,
+      "step": 1712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9748782862706917e-06,
+      "loss": 1.4749,
+      "step": 1713
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9727036676403766e-06,
+      "loss": 0.4109,
+      "step": 1714
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.970529049010062e-06,
+      "loss": 0.9895,
+      "step": 1715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.968354430379747e-06,
+      "loss": 1.2965,
+      "step": 1716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.966179811749432e-06,
+      "loss": 0.4855,
+      "step": 1717
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.964005193119117e-06,
+      "loss": 0.2794,
+      "step": 1718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9618305744888023e-06,
+      "loss": 0.5319,
+      "step": 1719
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9596559558584876e-06,
+      "loss": 0.1738,
+      "step": 1720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.957481337228173e-06,
+      "loss": 0.3323,
+      "step": 1721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.955306718597858e-06,
+      "loss": 0.8032,
+      "step": 1722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.953132099967543e-06,
+      "loss": 0.5139,
+      "step": 1723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9509574813372284e-06,
+      "loss": 0.5355,
+      "step": 1724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9487828627069137e-06,
+      "loss": 0.206,
+      "step": 1725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9466082440765986e-06,
+      "loss": 0.4158,
+      "step": 1726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.944433625446284e-06,
+      "loss": 0.8315,
+      "step": 1727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.942259006815969e-06,
+      "loss": 1.1115,
+      "step": 1728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.940084388185654e-06,
+      "loss": 0.6436,
+      "step": 1729
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.937909769555339e-06,
+      "loss": 0.662,
+      "step": 1730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9357351509250243e-06,
+      "loss": 0.9602,
+      "step": 1731
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9335605322947096e-06,
+      "loss": 0.5671,
+      "step": 1732
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9313859136643945e-06,
+      "loss": 0.4134,
+      "step": 1733
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9292112950340798e-06,
+      "loss": 0.4681,
+      "step": 1734
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.927036676403765e-06,
+      "loss": 1.1406,
+      "step": 1735
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9248620577734504e-06,
+      "loss": 0.5212,
+      "step": 1736
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9226874391431353e-06,
+      "loss": 0.3379,
+      "step": 1737
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9205128205128206e-06,
+      "loss": 0.5936,
+      "step": 1738
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.918338201882506e-06,
+      "loss": 0.7496,
+      "step": 1739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.916163583252191e-06,
+      "loss": 0.6887,
+      "step": 1740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.913988964621876e-06,
+      "loss": 0.2305,
+      "step": 1741
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9118143459915614e-06,
+      "loss": 0.497,
+      "step": 1742
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9096397273612467e-06,
+      "loss": 0.3875,
+      "step": 1743
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9074651087309316e-06,
+      "loss": 0.5513,
+      "step": 1744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9052904901006165e-06,
+      "loss": 0.5512,
+      "step": 1745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9031158714703018e-06,
+      "loss": 0.1822,
+      "step": 1746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.900941252839987e-06,
+      "loss": 1.2388,
+      "step": 1747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8987666342096724e-06,
+      "loss": 0.5859,
+      "step": 1748
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8965920155793573e-06,
+      "loss": 0.7318,
+      "step": 1749
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8944173969490426e-06,
+      "loss": 0.3785,
+      "step": 1750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.892242778318728e-06,
+      "loss": 0.2711,
+      "step": 1751
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.890068159688413e-06,
+      "loss": 0.4193,
+      "step": 1752
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.887893541058098e-06,
+      "loss": 0.7008,
+      "step": 1753
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8857189224277834e-06,
+      "loss": 0.3589,
+      "step": 1754
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8835443037974687e-06,
+      "loss": 0.9096,
+      "step": 1755
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8813696851671536e-06,
+      "loss": 0.6009,
+      "step": 1756
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.879195066536839e-06,
+      "loss": 0.8066,
+      "step": 1757
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8770204479065238e-06,
+      "loss": 0.3101,
+      "step": 1758
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.874845829276209e-06,
+      "loss": 0.2709,
+      "step": 1759
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.872671210645894e-06,
+      "loss": 0.2672,
+      "step": 1760
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8704965920155793e-06,
+      "loss": 0.2494,
+      "step": 1761
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8683219733852646e-06,
+      "loss": 0.2142,
+      "step": 1762
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.86614735475495e-06,
+      "loss": 0.9398,
+      "step": 1763
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8639727361246348e-06,
+      "loss": 0.1537,
+      "step": 1764
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.86179811749432e-06,
+      "loss": 0.7992,
+      "step": 1765
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8596234988640054e-06,
+      "loss": 0.2112,
+      "step": 1766
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8574488802336907e-06,
+      "loss": 1.3768,
+      "step": 1767
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8552742616033756e-06,
+      "loss": 0.3142,
+      "step": 1768
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.853099642973061e-06,
+      "loss": 0.1575,
+      "step": 1769
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.850925024342746e-06,
+      "loss": 0.5289,
+      "step": 1770
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8487504057124315e-06,
+      "loss": 0.5225,
+      "step": 1771
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.846575787082116e-06,
+      "loss": 0.224,
+      "step": 1772
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8444011684518013e-06,
+      "loss": 0.4493,
+      "step": 1773
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8422265498214866e-06,
+      "loss": 0.462,
+      "step": 1774
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.840051931191172e-06,
+      "loss": 0.6066,
+      "step": 1775
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8378773125608568e-06,
+      "loss": 0.3646,
+      "step": 1776
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.835702693930542e-06,
+      "loss": 1.3415,
+      "step": 1777
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8335280753002274e-06,
+      "loss": 0.1735,
+      "step": 1778
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8313534566699123e-06,
+      "loss": 0.1766,
+      "step": 1779
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8291788380395976e-06,
+      "loss": 0.4669,
+      "step": 1780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.827004219409283e-06,
+      "loss": 0.6097,
+      "step": 1781
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.824829600778968e-06,
+      "loss": 0.1938,
+      "step": 1782
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.822654982148653e-06,
+      "loss": 0.8229,
+      "step": 1783
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8204803635183384e-06,
+      "loss": 0.7941,
+      "step": 1784
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8183057448880237e-06,
+      "loss": 0.4738,
+      "step": 1785
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8161311262577086e-06,
+      "loss": 0.1759,
+      "step": 1786
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8139565076273935e-06,
+      "loss": 0.7616,
+      "step": 1787
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8117818889970788e-06,
+      "loss": 0.3603,
+      "step": 1788
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.809607270366764e-06,
+      "loss": 1.2688,
+      "step": 1789
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8074326517364494e-06,
+      "loss": 0.716,
+      "step": 1790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8052580331061343e-06,
+      "loss": 0.2103,
+      "step": 1791
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8030834144758196e-06,
+      "loss": 0.4635,
+      "step": 1792
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.800908795845505e-06,
+      "loss": 0.3885,
+      "step": 1793
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.79873417721519e-06,
+      "loss": 0.0976,
+      "step": 1794
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.796559558584875e-06,
+      "loss": 0.3105,
+      "step": 1795
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.7943849399545604e-06,
+      "loss": 0.1996,
+      "step": 1796
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.7922103213242457e-06,
+      "loss": 0.5196,
+      "step": 1797
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.790035702693931e-06,
+      "loss": 0.1245,
+      "step": 1798
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.787861084063616e-06,
+      "loss": 0.6902,
+      "step": 1799
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.7856864654333008e-06,
+      "loss": 0.5587,
+      "step": 1800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.783511846802986e-06,
+      "loss": 1.4809,
+      "step": 1801
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.7813372281726714e-06,
+      "loss": 2.141,
+      "step": 1802
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7791626095423563e-06,
+      "loss": 1.0053,
+      "step": 1803
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7769879909120416e-06,
+      "loss": 0.1946,
+      "step": 1804
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.774813372281727e-06,
+      "loss": 0.9417,
+      "step": 1805
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7726387536514118e-06,
+      "loss": 0.9407,
+      "step": 1806
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.770464135021097e-06,
+      "loss": 0.9436,
+      "step": 1807
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7682895163907824e-06,
+      "loss": 1.0208,
+      "step": 1808
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7661148977604677e-06,
+      "loss": 0.9123,
+      "step": 1809
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7639402791301526e-06,
+      "loss": 1.2752,
+      "step": 1810
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.761765660499838e-06,
+      "loss": 0.4396,
+      "step": 1811
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.759591041869523e-06,
+      "loss": 0.3271,
+      "step": 1812
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7574164232392085e-06,
+      "loss": 0.3916,
+      "step": 1813
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.755241804608893e-06,
+      "loss": 0.3302,
+      "step": 1814
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7530671859785783e-06,
+      "loss": 0.7244,
+      "step": 1815
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7508925673482636e-06,
+      "loss": 0.142,
+      "step": 1816
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.748717948717949e-06,
+      "loss": 0.2638,
+      "step": 1817
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7465433300876337e-06,
+      "loss": 0.1343,
+      "step": 1818
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.744368711457319e-06,
+      "loss": 0.1841,
+      "step": 1819
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7421940928270044e-06,
+      "loss": 0.5043,
+      "step": 1820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7400194741966897e-06,
+      "loss": 1.0787,
+      "step": 1821
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7378448555663746e-06,
+      "loss": 0.6264,
+      "step": 1822
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.73567023693606e-06,
+      "loss": 0.7277,
+      "step": 1823
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.733495618305745e-06,
+      "loss": 0.7707,
+      "step": 1824
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.73132099967543e-06,
+      "loss": 1.1305,
+      "step": 1825
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7291463810451154e-06,
+      "loss": 0.5549,
+      "step": 1826
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7269717624148007e-06,
+      "loss": 0.2501,
+      "step": 1827
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7247971437844856e-06,
+      "loss": 1.0761,
+      "step": 1828
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.722622525154171e-06,
+      "loss": 1.0408,
+      "step": 1829
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7204479065238557e-06,
+      "loss": 0.299,
+      "step": 1830
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.718273287893541e-06,
+      "loss": 0.543,
+      "step": 1831
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7160986692632264e-06,
+      "loss": 0.721,
+      "step": 1832
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7139240506329112e-06,
+      "loss": 0.3383,
+      "step": 1833
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7117494320025966e-06,
+      "loss": 0.5009,
+      "step": 1834
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.709574813372282e-06,
+      "loss": 0.4922,
+      "step": 1835
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.707400194741967e-06,
+      "loss": 1.1778,
+      "step": 1836
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.705225576111652e-06,
+      "loss": 0.9733,
+      "step": 1837
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7030509574813374e-06,
+      "loss": 1.056,
+      "step": 1838
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7008763388510227e-06,
+      "loss": 0.5827,
+      "step": 1839
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.698701720220708e-06,
+      "loss": 0.3782,
+      "step": 1840
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.696527101590393e-06,
+      "loss": 0.926,
+      "step": 1841
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.6943524829600777e-06,
+      "loss": 0.7373,
+      "step": 1842
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.692177864329763e-06,
+      "loss": 1.4195,
+      "step": 1843
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6900032456994484e-06,
+      "loss": 0.293,
+      "step": 1844
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6878286270691332e-06,
+      "loss": 0.1935,
+      "step": 1845
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6856540084388185e-06,
+      "loss": 0.5112,
+      "step": 1846
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.683479389808504e-06,
+      "loss": 0.2647,
+      "step": 1847
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.681304771178189e-06,
+      "loss": 0.1876,
+      "step": 1848
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.679130152547874e-06,
+      "loss": 0.714,
+      "step": 1849
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6769555339175594e-06,
+      "loss": 0.6403,
+      "step": 1850
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6747809152872447e-06,
+      "loss": 0.6282,
+      "step": 1851
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6726062966569295e-06,
+      "loss": 1.2782,
+      "step": 1852
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.670431678026615e-06,
+      "loss": 0.7926,
+      "step": 1853
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6682570593963e-06,
+      "loss": 0.3597,
+      "step": 1854
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6660824407659855e-06,
+      "loss": 0.4956,
+      "step": 1855
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6639078221356703e-06,
+      "loss": 0.3427,
+      "step": 1856
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6617332035053552e-06,
+      "loss": 0.5669,
+      "step": 1857
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6595585848750405e-06,
+      "loss": 0.1525,
+      "step": 1858
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.657383966244726e-06,
+      "loss": 0.2563,
+      "step": 1859
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6552093476144107e-06,
+      "loss": 0.3367,
+      "step": 1860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.653034728984096e-06,
+      "loss": 0.3383,
+      "step": 1861
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6508601103537813e-06,
+      "loss": 0.8514,
+      "step": 1862
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6486854917234667e-06,
+      "loss": 0.4259,
+      "step": 1863
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.6465108730931515e-06,
+      "loss": 0.5565,
+      "step": 1864
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.644336254462837e-06,
+      "loss": 0.2644,
+      "step": 1865
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.642161635832522e-06,
+      "loss": 0.9492,
+      "step": 1866
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6399870172022075e-06,
+      "loss": 0.5877,
+      "step": 1867
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6378123985718923e-06,
+      "loss": 0.4605,
+      "step": 1868
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6356377799415777e-06,
+      "loss": 0.5606,
+      "step": 1869
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6334631613112625e-06,
+      "loss": 0.4515,
+      "step": 1870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.631288542680948e-06,
+      "loss": 0.9,
+      "step": 1871
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6291139240506327e-06,
+      "loss": 0.3622,
+      "step": 1872
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.626939305420318e-06,
+      "loss": 0.4222,
+      "step": 1873
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6247646867900033e-06,
+      "loss": 0.3277,
+      "step": 1874
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6225900681596886e-06,
+      "loss": 0.2891,
+      "step": 1875
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6204154495293735e-06,
+      "loss": 0.4587,
+      "step": 1876
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.618240830899059e-06,
+      "loss": 0.3775,
+      "step": 1877
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.616066212268744e-06,
+      "loss": 0.6936,
+      "step": 1878
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.613891593638429e-06,
+      "loss": 0.3857,
+      "step": 1879
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6117169750081143e-06,
+      "loss": 1.4727,
+      "step": 1880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6095423563777996e-06,
+      "loss": 0.4805,
+      "step": 1881
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.607367737747485e-06,
+      "loss": 0.8118,
+      "step": 1882
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.60519311911717e-06,
+      "loss": 0.3996,
+      "step": 1883
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6030185004868547e-06,
+      "loss": 0.6356,
+      "step": 1884
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.60084388185654e-06,
+      "loss": 1.1128,
+      "step": 1885
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.5986692632262253e-06,
+      "loss": 0.5054,
+      "step": 1886
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.5964946445959102e-06,
+      "loss": 0.2494,
+      "step": 1887
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.5943200259655955e-06,
+      "loss": 0.1567,
+      "step": 1888
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.592145407335281e-06,
+      "loss": 0.4983,
+      "step": 1889
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.589970788704966e-06,
+      "loss": 1.4113,
+      "step": 1890
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.587796170074651e-06,
+      "loss": 0.668,
+      "step": 1891
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.5856215514443363e-06,
+      "loss": 0.4232,
+      "step": 1892
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.5834469328140216e-06,
+      "loss": 0.7534,
+      "step": 1893
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.581272314183707e-06,
+      "loss": 0.3851,
+      "step": 1894
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.579097695553392e-06,
+      "loss": 1.1842,
+      "step": 1895
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.576923076923077e-06,
+      "loss": 0.6825,
+      "step": 1896
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.5747484582927624e-06,
+      "loss": 1.1232,
+      "step": 1897
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.5725738396624473e-06,
+      "loss": 0.265,
+      "step": 1898
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.5703992210321322e-06,
+      "loss": 0.1693,
+      "step": 1899
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.5682246024018175e-06,
+      "loss": 0.2915,
+      "step": 1900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.566049983771503e-06,
+      "loss": 1.2666,
+      "step": 1901
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.563875365141188e-06,
+      "loss": 0.6916,
+      "step": 1902
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.561700746510873e-06,
+      "loss": 0.5307,
+      "step": 1903
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.5595261278805583e-06,
+      "loss": 0.3,
+      "step": 1904
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.5573515092502436e-06,
+      "loss": 1.1493,
+      "step": 1905
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5551768906199285e-06,
+      "loss": 0.2042,
+      "step": 1906
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.553002271989614e-06,
+      "loss": 1.0124,
+      "step": 1907
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.550827653359299e-06,
+      "loss": 0.3547,
+      "step": 1908
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5486530347289844e-06,
+      "loss": 0.9331,
+      "step": 1909
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5464784160986693e-06,
+      "loss": 0.8238,
+      "step": 1910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5443037974683546e-06,
+      "loss": 0.9588,
+      "step": 1911
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5421291788380395e-06,
+      "loss": 0.9465,
+      "step": 1912
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.539954560207725e-06,
+      "loss": 0.2731,
+      "step": 1913
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5377799415774097e-06,
+      "loss": 0.8579,
+      "step": 1914
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.535605322947095e-06,
+      "loss": 0.243,
+      "step": 1915
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5334307043167803e-06,
+      "loss": 0.7847,
+      "step": 1916
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5312560856864656e-06,
+      "loss": 0.9601,
+      "step": 1917
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5290814670561505e-06,
+      "loss": 0.3946,
+      "step": 1918
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.526906848425836e-06,
+      "loss": 0.3828,
+      "step": 1919
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.524732229795521e-06,
+      "loss": 0.6062,
+      "step": 1920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5225576111652064e-06,
+      "loss": 0.7294,
+      "step": 1921
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5203829925348913e-06,
+      "loss": 0.1719,
+      "step": 1922
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5182083739045766e-06,
+      "loss": 0.2375,
+      "step": 1923
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.516033755274262e-06,
+      "loss": 0.6041,
+      "step": 1924
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.513859136643947e-06,
+      "loss": 0.5924,
+      "step": 1925
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5116845180136317e-06,
+      "loss": 1.1421,
+      "step": 1926
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.509509899383317e-06,
+      "loss": 0.2505,
+      "step": 1927
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5073352807530023e-06,
+      "loss": 1.1226,
+      "step": 1928
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5051606621226876e-06,
+      "loss": 0.4143,
+      "step": 1929
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5029860434923725e-06,
+      "loss": 0.1432,
+      "step": 1930
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.500811424862058e-06,
+      "loss": 0.929,
+      "step": 1931
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.498636806231743e-06,
+      "loss": 0.8934,
+      "step": 1932
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.496462187601428e-06,
+      "loss": 0.6959,
+      "step": 1933
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4942875689711133e-06,
+      "loss": 0.5192,
+      "step": 1934
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4921129503407986e-06,
+      "loss": 0.5597,
+      "step": 1935
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.489938331710484e-06,
+      "loss": 0.2638,
+      "step": 1936
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.487763713080169e-06,
+      "loss": 0.651,
+      "step": 1937
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.485589094449854e-06,
+      "loss": 0.9596,
+      "step": 1938
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4834144758195394e-06,
+      "loss": 0.389,
+      "step": 1939
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4812398571892243e-06,
+      "loss": 0.1603,
+      "step": 1940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.479065238558909e-06,
+      "loss": 1.0095,
+      "step": 1941
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4768906199285945e-06,
+      "loss": 0.3017,
+      "step": 1942
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.47471600129828e-06,
+      "loss": 0.3679,
+      "step": 1943
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.472541382667965e-06,
+      "loss": 0.5485,
+      "step": 1944
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.47036676403765e-06,
+      "loss": 0.7904,
+      "step": 1945
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4681921454073353e-06,
+      "loss": 0.7985,
+      "step": 1946
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4660175267770206e-06,
+      "loss": 0.4532,
+      "step": 1947
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.463842908146706e-06,
+      "loss": 0.6297,
+      "step": 1948
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.461668289516391e-06,
+      "loss": 0.8628,
+      "step": 1949
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.459493670886076e-06,
+      "loss": 0.3195,
+      "step": 1950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4573190522557614e-06,
+      "loss": 0.5151,
+      "step": 1951
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4551444336254463e-06,
+      "loss": 0.4619,
+      "step": 1952
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4529698149951316e-06,
+      "loss": 0.2785,
+      "step": 1953
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4507951963648165e-06,
+      "loss": 0.8124,
+      "step": 1954
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.448620577734502e-06,
+      "loss": 0.4928,
+      "step": 1955
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.446445959104187e-06,
+      "loss": 0.8572,
+      "step": 1956
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.444271340473872e-06,
+      "loss": 0.5928,
+      "step": 1957
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4420967218435573e-06,
+      "loss": 0.2198,
+      "step": 1958
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4399221032132426e-06,
+      "loss": 0.881,
+      "step": 1959
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4377474845829275e-06,
+      "loss": 0.4606,
+      "step": 1960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.435572865952613e-06,
+      "loss": 0.8621,
+      "step": 1961
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.433398247322298e-06,
+      "loss": 0.2884,
+      "step": 1962
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4312236286919834e-06,
+      "loss": 0.4488,
+      "step": 1963
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4290490100616683e-06,
+      "loss": 0.3036,
+      "step": 1964
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4268743914313536e-06,
+      "loss": 0.6409,
+      "step": 1965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.424699772801039e-06,
+      "loss": 0.9235,
+      "step": 1966
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4225251541707242e-06,
+      "loss": 0.568,
+      "step": 1967
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4203505355404087e-06,
+      "loss": 0.6562,
+      "step": 1968
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.418175916910094e-06,
+      "loss": 0.1732,
+      "step": 1969
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4160012982797793e-06,
+      "loss": 0.1582,
+      "step": 1970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4138266796494646e-06,
+      "loss": 0.6052,
+      "step": 1971
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4116520610191495e-06,
+      "loss": 0.5056,
+      "step": 1972
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.409477442388835e-06,
+      "loss": 0.5721,
+      "step": 1973
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.40730282375852e-06,
+      "loss": 1.0514,
+      "step": 1974
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4051282051282054e-06,
+      "loss": 0.9579,
+      "step": 1975
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4029535864978903e-06,
+      "loss": 0.393,
+      "step": 1976
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4007789678675756e-06,
+      "loss": 0.7544,
+      "step": 1977
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.398604349237261e-06,
+      "loss": 0.2384,
+      "step": 1978
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.396429730606946e-06,
+      "loss": 0.1929,
+      "step": 1979
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.394255111976631e-06,
+      "loss": 1.055,
+      "step": 1980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3920804933463164e-06,
+      "loss": 0.1851,
+      "step": 1981
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3899058747160013e-06,
+      "loss": 0.278,
+      "step": 1982
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3877312560856866e-06,
+      "loss": 0.5294,
+      "step": 1983
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3855566374553715e-06,
+      "loss": 0.4285,
+      "step": 1984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.383382018825057e-06,
+      "loss": 0.1866,
+      "step": 1985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.381207400194742e-06,
+      "loss": 1.3065,
+      "step": 1986
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.379032781564427e-06,
+      "loss": 0.1692,
+      "step": 1987
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3768581629341123e-06,
+      "loss": 0.5638,
+      "step": 1988
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3746835443037976e-06,
+      "loss": 0.6555,
+      "step": 1989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.372508925673483e-06,
+      "loss": 1.0388,
+      "step": 1990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.370334307043168e-06,
+      "loss": 0.2497,
+      "step": 1991
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.368159688412853e-06,
+      "loss": 0.12,
+      "step": 1992
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3659850697825384e-06,
+      "loss": 0.281,
+      "step": 1993
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3638104511522237e-06,
+      "loss": 0.5079,
+      "step": 1994
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3616358325219086e-06,
+      "loss": 0.1697,
+      "step": 1995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3594612138915935e-06,
+      "loss": 0.7386,
+      "step": 1996
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.357286595261279e-06,
+      "loss": 0.52,
+      "step": 1997
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.355111976630964e-06,
+      "loss": 0.1192,
+      "step": 1998
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.352937358000649e-06,
+      "loss": 0.1913,
+      "step": 1999
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3507627393703343e-06,
+      "loss": 0.2775,
+      "step": 2000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3485881207400196e-06,
+      "loss": 0.1938,
+      "step": 2001
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.346413502109705e-06,
+      "loss": 0.1269,
+      "step": 2002
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.34423888347939e-06,
+      "loss": 0.2126,
+      "step": 2003
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.342064264849075e-06,
+      "loss": 0.4066,
+      "step": 2004
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3398896462187604e-06,
+      "loss": 0.1766,
+      "step": 2005
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3377150275884453e-06,
+      "loss": 0.4164,
+      "step": 2006
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3355404089581306e-06,
+      "loss": 0.2275,
+      "step": 2007
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.333365790327816e-06,
+      "loss": 0.7088,
+      "step": 2008
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.331191171697501e-06,
+      "loss": 0.2742,
+      "step": 2009
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3290165530671857e-06,
+      "loss": 1.0023,
+      "step": 2010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.326841934436871e-06,
+      "loss": 2.0295,
+      "step": 2011
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3246673158065563e-06,
+      "loss": 0.3801,
+      "step": 2012
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3224926971762416e-06,
+      "loss": 0.2505,
+      "step": 2013
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3203180785459265e-06,
+      "loss": 0.6502,
+      "step": 2014
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3181434599156118e-06,
+      "loss": 1.5437,
+      "step": 2015
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.315968841285297e-06,
+      "loss": 0.0832,
+      "step": 2016
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3137942226549824e-06,
+      "loss": 0.5971,
+      "step": 2017
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3116196040246673e-06,
+      "loss": 0.5876,
+      "step": 2018
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3094449853943526e-06,
+      "loss": 0.2351,
+      "step": 2019
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.307270366764038e-06,
+      "loss": 0.2984,
+      "step": 2020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.305095748133723e-06,
+      "loss": 1.2763,
+      "step": 2021
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.302921129503408e-06,
+      "loss": 0.1185,
+      "step": 2022
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3007465108730934e-06,
+      "loss": 1.0161,
+      "step": 2023
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2985718922427783e-06,
+      "loss": 0.6553,
+      "step": 2024
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2963972736124636e-06,
+      "loss": 0.4943,
+      "step": 2025
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2942226549821485e-06,
+      "loss": 0.3555,
+      "step": 2026
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2920480363518338e-06,
+      "loss": 0.1113,
+      "step": 2027
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.289873417721519e-06,
+      "loss": 0.2353,
+      "step": 2028
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2876987990912044e-06,
+      "loss": 0.4389,
+      "step": 2029
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2855241804608893e-06,
+      "loss": 0.3702,
+      "step": 2030
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2833495618305746e-06,
+      "loss": 0.948,
+      "step": 2031
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.28117494320026e-06,
+      "loss": 0.2176,
+      "step": 2032
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2790003245699448e-06,
+      "loss": 0.1788,
+      "step": 2033
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.27682570593963e-06,
+      "loss": 0.7485,
+      "step": 2034
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2746510873093154e-06,
+      "loss": 1.3383,
+      "step": 2035
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2724764686790007e-06,
+      "loss": 0.4022,
+      "step": 2036
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2703018500486856e-06,
+      "loss": 0.9805,
+      "step": 2037
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2681272314183705e-06,
+      "loss": 0.2553,
+      "step": 2038
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2659526127880558e-06,
+      "loss": 0.383,
+      "step": 2039
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.263777994157741e-06,
+      "loss": 0.6235,
+      "step": 2040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.261603375527426e-06,
+      "loss": 0.2212,
+      "step": 2041
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2594287568971113e-06,
+      "loss": 0.1208,
+      "step": 2042
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2572541382667966e-06,
+      "loss": 0.0747,
+      "step": 2043
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.255079519636482e-06,
+      "loss": 0.5636,
+      "step": 2044
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2529049010061668e-06,
+      "loss": 2.3037,
+      "step": 2045
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.250730282375852e-06,
+      "loss": 0.151,
+      "step": 2046
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2485556637455374e-06,
+      "loss": 0.7914,
+      "step": 2047
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2463810451152227e-06,
+      "loss": 0.2575,
+      "step": 2048
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.2442064264849076e-06,
+      "loss": 1.4903,
+      "step": 2049
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.242031807854593e-06,
+      "loss": 0.6121,
+      "step": 2050
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.239857189224278e-06,
+      "loss": 2.0126,
+      "step": 2051
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.237682570593963e-06,
+      "loss": 0.992,
+      "step": 2052
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.235507951963648e-06,
+      "loss": 1.1921,
+      "step": 2053
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.2333333333333333e-06,
+      "loss": 0.3329,
+      "step": 2054
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.6652046783625731,
+      "eval_f1": 0.6233552631578947,
+      "eval_loss": 0.7664929628372192,
+      "eval_precision": 0.7124060150375939,
+      "eval_recall": 0.554093567251462,
+      "eval_runtime": 230.8896,
+      "eval_samples_per_second": 5.925,
+      "eval_steps_per_second": 1.481,
+      "step": 2054
+    }
+  ],
+  "max_steps": 3081,
+  "num_train_epochs": 3,
+  "total_flos": 2160667986616320.0,
+  "trial_name": null,
+  "trial_params": null
+}