diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17790 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 2960,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.1007,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 1.932,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 2.2489,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 2.1807,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 2.1022,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993238674780257e-05,
+      "loss": 2.6319,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993238674780257e-05,
+      "loss": 3.0425,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9986477349560515e-05,
+      "loss": 2.1176,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997971602434077e-05,
+      "loss": 1.693,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997295469912103e-05,
+      "loss": 1.8805,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9966193373901284e-05,
+      "loss": 2.1004,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9959432048681543e-05,
+      "loss": 1.9308,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99526707234618e-05,
+      "loss": 1.8611,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9945909398242057e-05,
+      "loss": 1.8331,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9939148073022316e-05,
+      "loss": 1.9879,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993238674780257e-05,
+      "loss": 1.8147,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.992562542258283e-05,
+      "loss": 1.667,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9918864097363085e-05,
+      "loss": 1.8612,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9912102772143343e-05,
+      "loss": 1.9653,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99053414469236e-05,
+      "loss": 2.4434,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9898580121703857e-05,
+      "loss": 1.742,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9891818796484112e-05,
+      "loss": 1.7466,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9885057471264367e-05,
+      "loss": 1.832,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9878296146044626e-05,
+      "loss": 1.8429,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.987153482082488e-05,
+      "loss": 1.7983,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.986477349560514e-05,
+      "loss": 1.7881,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9858012170385395e-05,
+      "loss": 1.7959,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9851250845165654e-05,
+      "loss": 2.2341,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9844489519945912e-05,
+      "loss": 1.5761,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9837728194726168e-05,
+      "loss": 1.6658,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9830966869506426e-05,
+      "loss": 1.6497,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.982420554428668e-05,
+      "loss": 1.4019,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.981744421906694e-05,
+      "loss": 1.5941,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9810682893847195e-05,
+      "loss": 1.9075,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9803921568627454e-05,
+      "loss": 1.7495,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.979716024340771e-05,
+      "loss": 1.667,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9790398918187968e-05,
+      "loss": 1.8206,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9790398918187968e-05,
+      "loss": 2.2,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9783637592968223e-05,
+      "loss": 1.6554,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9776876267748478e-05,
+      "loss": 1.4738,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9770114942528737e-05,
+      "loss": 1.7058,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9763353617308992e-05,
+      "loss": 1.8674,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.975659229208925e-05,
+      "loss": 1.4674,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.974983096686951e-05,
+      "loss": 1.607,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9743069641649764e-05,
+      "loss": 1.7501,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9736308316430023e-05,
+      "loss": 1.5778,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9729546991210278e-05,
+      "loss": 3.189,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9722785665990537e-05,
+      "loss": 1.6319,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9716024340770792e-05,
+      "loss": 2.1078,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.970926301555105e-05,
+      "loss": 2.2133,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9702501690331306e-05,
+      "loss": 1.405,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9695740365111564e-05,
+      "loss": 1.0371,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.968897903989182e-05,
+      "loss": 2.0587,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9682217714672078e-05,
+      "loss": 1.4822,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9682217714672078e-05,
+      "loss": 2.222,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9675456389452333e-05,
+      "loss": 1.5066,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966869506423259e-05,
+      "loss": 1.817,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9661933739012847e-05,
+      "loss": 2.055,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9655172413793106e-05,
+      "loss": 1.9741,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964841108857336e-05,
+      "loss": 1.6344,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964164976335362e-05,
+      "loss": 1.5583,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9634888438133878e-05,
+      "loss": 1.898,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628127112914133e-05,
+      "loss": 1.7718,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962136578769439e-05,
+      "loss": 1.6644,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614604462474647e-05,
+      "loss": 2.0157,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614604462474647e-05,
+      "loss": 4.5153,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614604462474647e-05,
+      "loss": 3.5071,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 1.5587,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960108181203516e-05,
+      "loss": 1.2961,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9594320486815416e-05,
+      "loss": 1.8354,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9587559161595675e-05,
+      "loss": 1.859,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958079783637593e-05,
+      "loss": 1.6066,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957403651115619e-05,
+      "loss": 1.5196,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9567275185936444e-05,
+      "loss": 1.5678,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9560513860716703e-05,
+      "loss": 1.205,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9553752535496958e-05,
+      "loss": 1.3179,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9546991210277216e-05,
+      "loss": 1.6538,
+      "step": 77
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9540229885057475e-05,
+      "loss": 1.5995,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.953346855983773e-05,
+      "loss": 1.9129,
+      "step": 79
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952670723461799e-05,
+      "loss": 1.5146,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952670723461799e-05,
+      "loss": 3.594,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9519945909398244e-05,
+      "loss": 1.8276,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.95131845841785e-05,
+      "loss": 1.7537,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9506423258958758e-05,
+      "loss": 1.8598,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9499661933739013e-05,
+      "loss": 1.7188,
+      "step": 85
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.949290060851927e-05,
+      "loss": 1.5232,
+      "step": 86
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9486139283299527e-05,
+      "loss": 1.3097,
+      "step": 87
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9479377958079786e-05,
+      "loss": 1.6497,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.947261663286004e-05,
+      "loss": 1.6176,
+      "step": 89
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94658553076403e-05,
+      "loss": 1.573,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9459093982420555e-05,
+      "loss": 1.2954,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9452332657200813e-05,
+      "loss": 1.3624,
+      "step": 92
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445571331981072e-05,
+      "loss": 1.4447,
+      "step": 93
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438810006761327e-05,
+      "loss": 1.2505,
+      "step": 94
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432048681541586e-05,
+      "loss": 1.1061,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942528735632184e-05,
+      "loss": 1.9522,
+      "step": 96
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94185260311021e-05,
+      "loss": 1.3407,
+      "step": 97
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9411764705882355e-05,
+      "loss": 1.5548,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.940500338066261e-05,
+      "loss": 1.2321,
+      "step": 99
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939824205544287e-05,
+      "loss": 1.1754,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391480730223124e-05,
+      "loss": 1.4072,
+      "step": 101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384719405003382e-05,
+      "loss": 1.5468,
+      "step": 102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377958079783637e-05,
+      "loss": 2.5829,
+      "step": 103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371196754563896e-05,
+      "loss": 2.9623,
+      "step": 104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936443542934415e-05,
+      "loss": 1.7603,
+      "step": 105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935767410412441e-05,
+      "loss": 1.429,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935091277890467e-05,
+      "loss": 1.4357,
+      "step": 107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9344151453684924e-05,
+      "loss": 1.7017,
+      "step": 108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9337390128465182e-05,
+      "loss": 1.3173,
+      "step": 109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9330628803245438e-05,
+      "loss": 2.2866,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9323867478025696e-05,
+      "loss": 1.4139,
+      "step": 111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931710615280595e-05,
+      "loss": 1.7006,
+      "step": 112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931034482758621e-05,
+      "loss": 1.2744,
+      "step": 113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303583502366465e-05,
+      "loss": 2.2579,
+      "step": 114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.929682217714672e-05,
+      "loss": 1.6803,
+      "step": 115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.929006085192698e-05,
+      "loss": 2.0361,
+      "step": 116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9283299526707234e-05,
+      "loss": 1.2879,
+      "step": 117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9276538201487493e-05,
+      "loss": 1.5641,
+      "step": 118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9269776876267748e-05,
+      "loss": 1.9841,
+      "step": 119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9263015551048007e-05,
+      "loss": 1.7445,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9256254225828265e-05,
+      "loss": 1.5009,
+      "step": 121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.924949290060852e-05,
+      "loss": 2.7139,
+      "step": 122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.924273157538878e-05,
+      "loss": 1.7759,
+      "step": 123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9235970250169034e-05,
+      "loss": 1.6841,
+      "step": 124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9229208924949293e-05,
+      "loss": 2.6352,
+      "step": 125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9222447599729548e-05,
+      "loss": 1.0943,
+      "step": 126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9215686274509807e-05,
+      "loss": 1.7653,
+      "step": 127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9208924949290062e-05,
+      "loss": 1.4519,
+      "step": 128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.920216362407032e-05,
+      "loss": 0.7758,
+      "step": 129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9195402298850576e-05,
+      "loss": 1.0123,
+      "step": 130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.918864097363083e-05,
+      "loss": 1.5934,
+      "step": 131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.918187964841109e-05,
+      "loss": 1.9171,
+      "step": 132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9175118323191345e-05,
+      "loss": 2.4308,
+      "step": 133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9168356997971603e-05,
+      "loss": 2.6888,
+      "step": 134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9161595672751862e-05,
+      "loss": 2.2068,
+      "step": 135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9154834347532117e-05,
+      "loss": 2.2327,
+      "step": 136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9148073022312376e-05,
+      "loss": 1.8085,
+      "step": 137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.914131169709263e-05,
+      "loss": 2.0571,
+      "step": 138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.913455037187289e-05,
+      "loss": 1.1095,
+      "step": 139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9127789046653145e-05,
+      "loss": 1.7265,
+      "step": 140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9121027721433404e-05,
+      "loss": 1.2719,
+      "step": 141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.911426639621366e-05,
+      "loss": 1.6927,
+      "step": 142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9107505070993917e-05,
+      "loss": 1.9839,
+      "step": 143
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9100743745774173e-05,
+      "loss": 1.8372,
+      "step": 144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.909398242055443e-05,
+      "loss": 1.5292,
+      "step": 145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9087221095334686e-05,
+      "loss": 1.4021,
+      "step": 146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.908045977011494e-05,
+      "loss": 1.8594,
+      "step": 147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.90736984448952e-05,
+      "loss": 1.7402,
+      "step": 148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.906693711967546e-05,
+      "loss": 1.8764,
+      "step": 149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9060175794455714e-05,
+      "loss": 1.7828,
+      "step": 150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9053414469235973e-05,
+      "loss": 2.5005,
+      "step": 151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9046653144016228e-05,
+      "loss": 1.7583,
+      "step": 152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9039891818796486e-05,
+      "loss": 1.3033,
+      "step": 153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9033130493576745e-05,
+      "loss": 1.8798,
+      "step": 154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9026369168357e-05,
+      "loss": 1.5722,
+      "step": 155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9019607843137255e-05,
+      "loss": 2.2527,
+      "step": 156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9012846517917514e-05,
+      "loss": 1.2799,
+      "step": 157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.900608519269777e-05,
+      "loss": 1.5943,
+      "step": 158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8999323867478028e-05,
+      "loss": 1.9062,
+      "step": 159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8992562542258283e-05,
+      "loss": 1.533,
+      "step": 160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8985801217038542e-05,
+      "loss": 1.9538,
+      "step": 161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8979039891818797e-05,
+      "loss": 1.571,
+      "step": 162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8972278566599052e-05,
+      "loss": 2.0401,
+      "step": 163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.896551724137931e-05,
+      "loss": 1.8967,
+      "step": 164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.895875591615957e-05,
+      "loss": 0.7138,
+      "step": 165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8951994590939825e-05,
+      "loss": 1.5968,
+      "step": 166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8945233265720083e-05,
+      "loss": 2.6436,
+      "step": 167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8938471940500342e-05,
+      "loss": 2.2791,
+      "step": 168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8931710615280597e-05,
+      "loss": 1.8413,
+      "step": 169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8924949290060856e-05,
+      "loss": 2.2112,
+      "step": 170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.891818796484111e-05,
+      "loss": 1.6033,
+      "step": 171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8911426639621366e-05,
+      "loss": 1.833,
+      "step": 172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8904665314401625e-05,
+      "loss": 1.9761,
+      "step": 173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.889790398918188e-05,
+      "loss": 1.2749,
+      "step": 174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.889114266396214e-05,
+      "loss": 1.7612,
+      "step": 175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8884381338742394e-05,
+      "loss": 1.738,
+      "step": 176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8877620013522652e-05,
+      "loss": 1.4422,
+      "step": 177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8870858688302908e-05,
+      "loss": 1.4651,
+      "step": 178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8864097363083166e-05,
+      "loss": 1.0238,
+      "step": 179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.885733603786342e-05,
+      "loss": 1.9362,
+      "step": 180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.885057471264368e-05,
+      "loss": 0.7507,
+      "step": 181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.884381338742394e-05,
+      "loss": 1.706,
+      "step": 182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8837052062204194e-05,
+      "loss": 2.1557,
+      "step": 183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8830290736984452e-05,
+      "loss": 1.3889,
+      "step": 184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 1.89,
+      "step": 185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8816768086544966e-05,
+      "loss": 1.5032,
+      "step": 186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.881000676132522e-05,
+      "loss": 1.2285,
+      "step": 187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8803245436105477e-05,
+      "loss": 1.9761,
+      "step": 188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8796484110885735e-05,
+      "loss": 0.8688,
+      "step": 189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.878972278566599e-05,
+      "loss": 1.443,
+      "step": 190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.878296146044625e-05,
+      "loss": 1.4064,
+      "step": 191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8776200135226504e-05,
+      "loss": 1.6567,
+      "step": 192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8769438810006763e-05,
+      "loss": 0.8416,
+      "step": 193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8762677484787018e-05,
+      "loss": 1.2989,
+      "step": 194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8755916159567277e-05,
+      "loss": 1.4178,
+      "step": 195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8749154834347535e-05,
+      "loss": 1.594,
+      "step": 196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.874239350912779e-05,
+      "loss": 1.0248,
+      "step": 197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.873563218390805e-05,
+      "loss": 1.4369,
+      "step": 198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8728870858688304e-05,
+      "loss": 2.3154,
+      "step": 199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8722109533468563e-05,
+      "loss": 1.4162,
+      "step": 200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8715348208248818e-05,
+      "loss": 1.1116,
+      "step": 201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8708586883029077e-05,
+      "loss": 1.4054,
+      "step": 202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8701825557809332e-05,
+      "loss": 1.2815,
+      "step": 203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8695064232589587e-05,
+      "loss": 1.5892,
+      "step": 204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8688302907369846e-05,
+      "loss": 1.1375,
+      "step": 205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.86815415821501e-05,
+      "loss": 1.6846,
+      "step": 206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.867478025693036e-05,
+      "loss": 1.535,
+      "step": 207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8668018931710615e-05,
+      "loss": 1.2888,
+      "step": 208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8661257606490873e-05,
+      "loss": 1.3233,
+      "step": 209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8654496281271132e-05,
+      "loss": 1.5425,
+      "step": 210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8647734956051387e-05,
+      "loss": 1.8643,
+      "step": 211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8640973630831646e-05,
+      "loss": 0.7362,
+      "step": 212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.86342123056119e-05,
+      "loss": 1.5906,
+      "step": 213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862745098039216e-05,
+      "loss": 0.9842,
+      "step": 214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8620689655172415e-05,
+      "loss": 1.1466,
+      "step": 215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8613928329952674e-05,
+      "loss": 1.6106,
+      "step": 216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860716700473293e-05,
+      "loss": 1.6942,
+      "step": 217
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8600405679513187e-05,
+      "loss": 1.2757,
+      "step": 218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8593644354293443e-05,
+      "loss": 1.8843,
+      "step": 219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8586883029073698e-05,
+      "loss": 2.1582,
+      "step": 220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8580121703853956e-05,
+      "loss": 1.9189,
+      "step": 221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.857336037863421e-05,
+      "loss": 1.9685,
+      "step": 222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.856659905341447e-05,
+      "loss": 1.4254,
+      "step": 223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.855983772819473e-05,
+      "loss": 1.5133,
+      "step": 224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8553076402974984e-05,
+      "loss": 0.8454,
+      "step": 225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8546315077755243e-05,
+      "loss": 2.0718,
+      "step": 226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8539553752535498e-05,
+      "loss": 1.3267,
+      "step": 227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8532792427315757e-05,
+      "loss": 0.5437,
+      "step": 228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8526031102096012e-05,
+      "loss": 1.802,
+      "step": 229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.851926977687627e-05,
+      "loss": 1.4217,
+      "step": 230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8512508451656526e-05,
+      "loss": 1.9453,
+      "step": 231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8505747126436784e-05,
+      "loss": 1.2207,
+      "step": 232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.849898580121704e-05,
+      "loss": 1.5363,
+      "step": 233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8492224475997298e-05,
+      "loss": 1.5837,
+      "step": 234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8485463150777553e-05,
+      "loss": 1.0207,
+      "step": 235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.847870182555781e-05,
+      "loss": 1.6495,
+      "step": 236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8471940500338067e-05,
+      "loss": 1.6253,
+      "step": 237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8465179175118326e-05,
+      "loss": 0.9104,
+      "step": 238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.845841784989858e-05,
+      "loss": 1.8065,
+      "step": 239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.845165652467884e-05,
+      "loss": 0.8891,
+      "step": 240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8444895199459095e-05,
+      "loss": 0.6385,
+      "step": 241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8438133874239353e-05,
+      "loss": 1.1902,
+      "step": 242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.843137254901961e-05,
+      "loss": 2.1191,
+      "step": 243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8424611223799867e-05,
+      "loss": 1.979,
+      "step": 244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8417849898580122e-05,
+      "loss": 1.4408,
+      "step": 245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.841108857336038e-05,
+      "loss": 1.6497,
+      "step": 246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8404327248140636e-05,
+      "loss": 1.076,
+      "step": 247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8397565922920895e-05,
+      "loss": 1.6376,
+      "step": 248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.839080459770115e-05,
+      "loss": 2.1843,
+      "step": 249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.838404327248141e-05,
+      "loss": 1.5839,
+      "step": 250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8377281947261664e-05,
+      "loss": 1.6687,
+      "step": 251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8370520622041922e-05,
+      "loss": 1.5677,
+      "step": 252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8363759296822178e-05,
+      "loss": 1.1909,
+      "step": 253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8356997971602436e-05,
+      "loss": 1.6484,
+      "step": 254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.835023664638269e-05,
+      "loss": 1.2677,
+      "step": 255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.834347532116295e-05,
+      "loss": 1.1882,
+      "step": 256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.833671399594321e-05,
+      "loss": 1.4911,
+      "step": 257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8329952670723464e-05,
+      "loss": 1.6304,
+      "step": 258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.832319134550372e-05,
+      "loss": 1.083,
+      "step": 259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8316430020283978e-05,
+      "loss": 1.3278,
+      "step": 260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8309668695064233e-05,
+      "loss": 1.4286,
+      "step": 261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.830290736984449e-05,
+      "loss": 2.1976,
+      "step": 262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8296146044624747e-05,
+      "loss": 2.6591,
+      "step": 263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8289384719405005e-05,
+      "loss": 1.6633,
+      "step": 264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.828262339418526e-05,
+      "loss": 1.4494,
+      "step": 265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.827586206896552e-05,
+      "loss": 1.5879,
+      "step": 266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8269100743745774e-05,
+      "loss": 1.3142,
+      "step": 267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8262339418526033e-05,
+      "loss": 1.781,
+      "step": 268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8255578093306288e-05,
+      "loss": 1.421,
+      "step": 269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8248816768086547e-05,
+      "loss": 1.2111,
+      "step": 270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8242055442866805e-05,
+      "loss": 1.488,
+      "step": 271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.823529411764706e-05,
+      "loss": 1.709,
+      "step": 272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.822853279242732e-05,
+      "loss": 2.0163,
+      "step": 273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8221771467207574e-05,
+      "loss": 1.2377,
+      "step": 274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.821501014198783e-05,
+      "loss": 1.9826,
+      "step": 275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8208248816768088e-05,
+      "loss": 1.6849,
+      "step": 276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8201487491548343e-05,
+      "loss": 2.1306,
+      "step": 277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8194726166328602e-05,
+      "loss": 1.9947,
+      "step": 278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8187964841108857e-05,
+      "loss": 1.0959,
+      "step": 279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8181203515889116e-05,
+      "loss": 1.2524,
+      "step": 280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.817444219066937e-05,
+      "loss": 0.9619,
+      "step": 281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.816768086544963e-05,
+      "loss": 1.8078,
+      "step": 282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8160919540229885e-05,
+      "loss": 0.6511,
+      "step": 283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8154158215010144e-05,
+      "loss": 1.345,
+      "step": 284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8147396889790402e-05,
+      "loss": 1.247,
+      "step": 285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8140635564570657e-05,
+      "loss": 1.013,
+      "step": 286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8133874239350916e-05,
+      "loss": 1.1964,
+      "step": 287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.812711291413117e-05,
+      "loss": 1.8923,
+      "step": 288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.812035158891143e-05,
+      "loss": 1.2504,
+      "step": 289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8113590263691685e-05,
+      "loss": 1.601,
+      "step": 290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.810682893847194e-05,
+      "loss": 1.9743,
+      "step": 291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.81000676132522e-05,
+      "loss": 1.783,
+      "step": 292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8093306288032454e-05,
+      "loss": 1.8986,
+      "step": 293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8086544962812713e-05,
+      "loss": 1.3416,
+      "step": 294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8079783637592968e-05,
+      "loss": 2.1022,
+      "step": 295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8073022312373226e-05,
+      "loss": 1.8068,
+      "step": 296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8066260987153482e-05,
+      "loss": 1.5197,
+      "step": 297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.805949966193374e-05,
+      "loss": 1.0507,
+      "step": 298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8052738336714e-05,
+      "loss": 1.3974,
+      "step": 299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8045977011494254e-05,
+      "loss": 2.0972,
+      "step": 300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8039215686274513e-05,
+      "loss": 1.2105,
+      "step": 301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8032454361054768e-05,
+      "loss": 1.9792,
+      "step": 302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8025693035835027e-05,
+      "loss": 1.8938,
+      "step": 303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8018931710615282e-05,
+      "loss": 1.6095,
+      "step": 304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.801217038539554e-05,
+      "loss": 1.6262,
+      "step": 305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8005409060175796e-05,
+      "loss": 0.9142,
+      "step": 306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.799864773495605e-05,
+      "loss": 1.5484,
+      "step": 307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.799188640973631e-05,
+      "loss": 1.5741,
+      "step": 308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7985125084516565e-05,
+      "loss": 0.0,
+      "step": 309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7978363759296823e-05,
+      "loss": 0.5159,
+      "step": 310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.797160243407708e-05,
+      "loss": 2.0734,
+      "step": 311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7964841108857337e-05,
+      "loss": 2.0959,
+      "step": 312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7958079783637596e-05,
+      "loss": 1.4288,
+      "step": 313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.795131845841785e-05,
+      "loss": 1.7256,
+      "step": 314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.794455713319811e-05,
+      "loss": 1.8284,
+      "step": 315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7937795807978365e-05,
+      "loss": 1.0701,
+      "step": 316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7931034482758623e-05,
+      "loss": 1.6857,
+      "step": 317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.792427315753888e-05,
+      "loss": 1.5269,
+      "step": 318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7917511832319137e-05,
+      "loss": 1.5957,
+      "step": 319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7910750507099392e-05,
+      "loss": 1.9908,
+      "step": 320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.790398918187965e-05,
+      "loss": 2.3536,
+      "step": 321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7897227856659906e-05,
+      "loss": 2.0215,
+      "step": 322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.789046653144016e-05,
+      "loss": 1.4783,
+      "step": 323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.788370520622042e-05,
+      "loss": 2.0582,
+      "step": 324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7876943881000675e-05,
+      "loss": 2.4088,
+      "step": 325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7870182555780934e-05,
+      "loss": 1.8167,
+      "step": 326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7863421230561192e-05,
+      "loss": 1.1516,
+      "step": 327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7856659905341448e-05,
+      "loss": 1.2232,
+      "step": 328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7849898580121706e-05,
+      "loss": 1.7428,
+      "step": 329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7843137254901965e-05,
+      "loss": 1.0287,
+      "step": 330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.783637592968222e-05,
+      "loss": 1.5166,
+      "step": 331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7829614604462475e-05,
+      "loss": 1.7046,
+      "step": 332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7822853279242734e-05,
+      "loss": 1.8005,
+      "step": 333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.781609195402299e-05,
+      "loss": 1.6942,
+      "step": 334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7809330628803248e-05,
+      "loss": 1.3771,
+      "step": 335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7802569303583503e-05,
+      "loss": 0.9345,
+      "step": 336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.779580797836376e-05,
+      "loss": 2.0004,
+      "step": 337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7789046653144017e-05,
+      "loss": 1.463,
+      "step": 338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7782285327924272e-05,
+      "loss": 1.4005,
+      "step": 339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.777552400270453e-05,
+      "loss": 1.5952,
+      "step": 340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.776876267748479e-05,
+      "loss": 0.8524,
+      "step": 341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7762001352265044e-05,
+      "loss": 1.8385,
+      "step": 342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7755240027045303e-05,
+      "loss": 0.4893,
+      "step": 343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.774847870182556e-05,
+      "loss": 2.0502,
+      "step": 344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7741717376605817e-05,
+      "loss": 1.1624,
+      "step": 345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7734956051386075e-05,
+      "loss": 1.0935,
+      "step": 346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.772819472616633e-05,
+      "loss": 1.3396,
+      "step": 347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7721433400946586e-05,
+      "loss": 1.1466,
+      "step": 348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7714672075726844e-05,
+      "loss": 1.974,
+      "step": 349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.77079107505071e-05,
+      "loss": 1.7733,
+      "step": 350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.770114942528736e-05,
+      "loss": 1.2569,
+      "step": 351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7694388100067614e-05,
+      "loss": 1.0272,
+      "step": 352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7687626774847872e-05,
+      "loss": 1.8412,
+      "step": 353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7680865449628127e-05,
+      "loss": 1.8982,
+      "step": 354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7674104124408386e-05,
+      "loss": 1.4341,
+      "step": 355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.766734279918864e-05,
+      "loss": 0.9766,
+      "step": 356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.76605814739689e-05,
+      "loss": 1.5498,
+      "step": 357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.765382014874916e-05,
+      "loss": 1.5868,
+      "step": 358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 1.7869,
+      "step": 359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7640297498309672e-05,
+      "loss": 2.0091,
+      "step": 360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7633536173089927e-05,
+      "loss": 1.0428,
+      "step": 361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7626774847870186e-05,
+      "loss": 2.4605,
+      "step": 362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.762001352265044e-05,
+      "loss": 1.3623,
+      "step": 363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7613252197430696e-05,
+      "loss": 1.8887,
+      "step": 364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7606490872210955e-05,
+      "loss": 1.6701,
+      "step": 365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.759972954699121e-05,
+      "loss": 1.18,
+      "step": 366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.759296822177147e-05,
+      "loss": 1.3391,
+      "step": 367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7586206896551724e-05,
+      "loss": 1.772,
+      "step": 368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7579445571331983e-05,
+      "loss": 1.8437,
+      "step": 369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7572684246112238e-05,
+      "loss": 2.0014,
+      "step": 370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7565922920892497e-05,
+      "loss": 1.4436,
+      "step": 371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7559161595672755e-05,
+      "loss": 2.3889,
+      "step": 372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.755240027045301e-05,
+      "loss": 1.0995,
+      "step": 373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.754563894523327e-05,
+      "loss": 1.186,
+      "step": 374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7538877620013524e-05,
+      "loss": 1.7127,
+      "step": 375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7532116294793783e-05,
+      "loss": 1.7249,
+      "step": 376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7525354969574038e-05,
+      "loss": 1.6277,
+      "step": 377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7518593644354297e-05,
+      "loss": 1.2513,
+      "step": 378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7511832319134552e-05,
+      "loss": 1.466,
+      "step": 379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7505070993914807e-05,
+      "loss": 1.8832,
+      "step": 380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7498309668695066e-05,
+      "loss": 1.0401,
+      "step": 381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.749154834347532e-05,
+      "loss": 1.8366,
+      "step": 382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.748478701825558e-05,
+      "loss": 1.095,
+      "step": 383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7478025693035835e-05,
+      "loss": 1.0766,
+      "step": 384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7471264367816093e-05,
+      "loss": 1.365,
+      "step": 385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.746450304259635e-05,
+      "loss": 1.3576,
+      "step": 386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7457741717376607e-05,
+      "loss": 0.9606,
+      "step": 387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7450980392156866e-05,
+      "loss": 1.7632,
+      "step": 388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.744421906693712e-05,
+      "loss": 1.3279,
+      "step": 389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.743745774171738e-05,
+      "loss": 1.0802,
+      "step": 390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7430696416497635e-05,
+      "loss": 1.6383,
+      "step": 391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7423935091277893e-05,
+      "loss": 1.3994,
+      "step": 392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.741717376605815e-05,
+      "loss": 1.3308,
+      "step": 393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7410412440838407e-05,
+      "loss": 1.2662,
+      "step": 394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7403651115618662e-05,
+      "loss": 1.1795,
+      "step": 395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7396889790398918e-05,
+      "loss": 1.523,
+      "step": 396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7390128465179176e-05,
+      "loss": 0.9288,
+      "step": 397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.738336713995943e-05,
+      "loss": 0.86,
+      "step": 398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.737660581473969e-05,
+      "loss": 1.4099,
+      "step": 399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7369844489519945e-05,
+      "loss": 0.9285,
+      "step": 400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7363083164300204e-05,
+      "loss": 0.8961,
+      "step": 401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7356321839080462e-05,
+      "loss": 1.7157,
+      "step": 402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7349560513860718e-05,
+      "loss": 0.9689,
+      "step": 403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7342799188640976e-05,
+      "loss": 1.8976,
+      "step": 404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.733603786342123e-05,
+      "loss": 1.7151,
+      "step": 405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.732927653820149e-05,
+      "loss": 2.1247,
+      "step": 406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7322515212981745e-05,
+      "loss": 1.2534,
+      "step": 407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7315753887762004e-05,
+      "loss": 0.899,
+      "step": 408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.730899256254226e-05,
+      "loss": 1.6634,
+      "step": 409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7302231237322518e-05,
+      "loss": 1.1551,
+      "step": 410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7295469912102773e-05,
+      "loss": 1.175,
+      "step": 411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7288708586883028e-05,
+      "loss": 1.3626,
+      "step": 412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7281947261663287e-05,
+      "loss": 1.5521,
+      "step": 413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7275185936443542e-05,
+      "loss": 1.9822,
+      "step": 414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.72684246112238e-05,
+      "loss": 1.4834,
+      "step": 415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.726166328600406e-05,
+      "loss": 1.3693,
+      "step": 416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7254901960784314e-05,
+      "loss": 1.1337,
+      "step": 417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7248140635564573e-05,
+      "loss": 2.2222,
+      "step": 418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7241379310344828e-05,
+      "loss": 1.6909,
+      "step": 419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7234617985125087e-05,
+      "loss": 1.5321,
+      "step": 420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7227856659905342e-05,
+      "loss": 0.9598,
+      "step": 421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.72210953346856e-05,
+      "loss": 2.001,
+      "step": 422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7214334009465856e-05,
+      "loss": 1.5837,
+      "step": 423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7207572684246115e-05,
+      "loss": 1.6824,
+      "step": 424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.720081135902637e-05,
+      "loss": 1.8823,
+      "step": 425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.719405003380663e-05,
+      "loss": 0.5892,
+      "step": 426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7187288708586884e-05,
+      "loss": 1.124,
+      "step": 427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.718052738336714e-05,
+      "loss": 1.5616,
+      "step": 428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7173766058147397e-05,
+      "loss": 1.3786,
+      "step": 429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7167004732927656e-05,
+      "loss": 1.5554,
+      "step": 430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.716024340770791e-05,
+      "loss": 0.6804,
+      "step": 431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.715348208248817e-05,
+      "loss": 1.6682,
+      "step": 432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.714672075726843e-05,
+      "loss": 1.5858,
+      "step": 433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7139959432048684e-05,
+      "loss": 0.9027,
+      "step": 434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.713319810682894e-05,
+      "loss": 1.5886,
+      "step": 435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7126436781609197e-05,
+      "loss": 1.7926,
+      "step": 436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7119675456389453e-05,
+      "loss": 1.3147,
+      "step": 437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.711291413116971e-05,
+      "loss": 1.6036,
+      "step": 438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7106152805949967e-05,
+      "loss": 1.9749,
+      "step": 439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7099391480730225e-05,
+      "loss": 2.443,
+      "step": 440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.709263015551048e-05,
+      "loss": 1.1031,
+      "step": 441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.708586883029074e-05,
+      "loss": 1.9166,
+      "step": 442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7079107505070994e-05,
+      "loss": 1.3729,
+      "step": 443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7072346179851253e-05,
+      "loss": 1.4014,
+      "step": 444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7065584854631508e-05,
+      "loss": 1.3541,
+      "step": 445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7058823529411767e-05,
+      "loss": 2.6064,
+      "step": 446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7052062204192025e-05,
+      "loss": 1.8873,
+      "step": 447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.704530087897228e-05,
+      "loss": 1.7042,
+      "step": 448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.703853955375254e-05,
+      "loss": 1.9798,
+      "step": 449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7031778228532794e-05,
+      "loss": 1.4438,
+      "step": 450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.702501690331305e-05,
+      "loss": 1.7074,
+      "step": 451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7018255578093308e-05,
+      "loss": 1.0164,
+      "step": 452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7011494252873563e-05,
+      "loss": 1.0322,
+      "step": 453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7004732927653822e-05,
+      "loss": 3.0459,
+      "step": 454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6997971602434077e-05,
+      "loss": 1.6956,
+      "step": 455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6991210277214336e-05,
+      "loss": 1.6528,
+      "step": 456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.698444895199459e-05,
+      "loss": 1.1877,
+      "step": 457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.697768762677485e-05,
+      "loss": 1.5291,
+      "step": 458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6970926301555105e-05,
+      "loss": 2.228,
+      "step": 459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6964164976335363e-05,
+      "loss": 1.5093,
+      "step": 460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6957403651115622e-05,
+      "loss": 1.5263,
+      "step": 461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6950642325895877e-05,
+      "loss": 1.6498,
+      "step": 462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6943881000676136e-05,
+      "loss": 2.0827,
+      "step": 463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.693711967545639e-05,
+      "loss": 0.7065,
+      "step": 464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.693035835023665e-05,
+      "loss": 1.0477,
+      "step": 465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6923597025016905e-05,
+      "loss": 1.5147,
+      "step": 466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.691683569979716e-05,
+      "loss": 1.986,
+      "step": 467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.691007437457742e-05,
+      "loss": 0.5134,
+      "step": 468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6903313049357674e-05,
+      "loss": 2.0999,
+      "step": 469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6896551724137932e-05,
+      "loss": 1.3353,
+      "step": 470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6889790398918188e-05,
+      "loss": 1.6786,
+      "step": 471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6883029073698446e-05,
+      "loss": 1.9746,
+      "step": 472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.68762677484787e-05,
+      "loss": 1.5554,
+      "step": 473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.686950642325896e-05,
+      "loss": 1.3448,
+      "step": 474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.686274509803922e-05,
+      "loss": 1.4008,
+      "step": 475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6855983772819474e-05,
+      "loss": 1.8056,
+      "step": 476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6849222447599733e-05,
+      "loss": 1.3721,
+      "step": 477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6842461122379988e-05,
+      "loss": 2.2674,
+      "step": 478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6835699797160246e-05,
+      "loss": 1.3559,
+      "step": 479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.68289384719405e-05,
+      "loss": 1.5929,
+      "step": 480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.682217714672076e-05,
+      "loss": 1.4193,
+      "step": 481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6815415821501015e-05,
+      "loss": 1.6495,
+      "step": 482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.680865449628127e-05,
+      "loss": 1.4012,
+      "step": 483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.680189317106153e-05,
+      "loss": 1.6109,
+      "step": 484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6795131845841784e-05,
+      "loss": 1.9762,
+      "step": 485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6788370520622043e-05,
+      "loss": 1.3271,
+      "step": 486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6781609195402298e-05,
+      "loss": 1.6459,
+      "step": 487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6774847870182557e-05,
+      "loss": 1.1647,
+      "step": 488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6768086544962815e-05,
+      "loss": 0.5375,
+      "step": 489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.676132521974307e-05,
+      "loss": 1.3489,
+      "step": 490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.675456389452333e-05,
+      "loss": 2.2377,
+      "step": 491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6747802569303585e-05,
+      "loss": 2.0274,
+      "step": 492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6741041244083843e-05,
+      "loss": 0.8628,
+      "step": 493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.67342799188641e-05,
+      "loss": 2.0356,
+      "step": 494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6727518593644357e-05,
+      "loss": 0.7309,
+      "step": 495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6720757268424612e-05,
+      "loss": 0.9742,
+      "step": 496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.671399594320487e-05,
+      "loss": 1.1277,
+      "step": 497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6707234617985126e-05,
+      "loss": 1.1812,
+      "step": 498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.670047329276538e-05,
+      "loss": 1.0347,
+      "step": 499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.669371196754564e-05,
+      "loss": 1.1892,
+      "step": 500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6686950642325895e-05,
+      "loss": 1.5686,
+      "step": 501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6680189317106154e-05,
+      "loss": 2.7968,
+      "step": 502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6673427991886412e-05,
+      "loss": 1.8073,
+      "step": 503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 2.7017,
+      "step": 504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6659905341446926e-05,
+      "loss": 1.7906,
+      "step": 505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.665314401622718e-05,
+      "loss": 1.1333,
+      "step": 506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.664638269100744e-05,
+      "loss": 1.4889,
+      "step": 507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6639621365787695e-05,
+      "loss": 1.5509,
+      "step": 508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6632860040567954e-05,
+      "loss": 1.3289,
+      "step": 509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.662609871534821e-05,
+      "loss": 1.9038,
+      "step": 510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6619337390128468e-05,
+      "loss": 2.747,
+      "step": 511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6612576064908723e-05,
+      "loss": 1.0236,
+      "step": 512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.660581473968898e-05,
+      "loss": 1.4801,
+      "step": 513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6599053414469237e-05,
+      "loss": 1.9984,
+      "step": 514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6592292089249492e-05,
+      "loss": 1.8593,
+      "step": 515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.658553076402975e-05,
+      "loss": 2.0706,
+      "step": 516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.657876943881001e-05,
+      "loss": 2.2519,
+      "step": 517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6572008113590264e-05,
+      "loss": 1.4447,
+      "step": 518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6565246788370523e-05,
+      "loss": 4.8877,
+      "step": 519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6558485463150778e-05,
+      "loss": 2.4604,
+      "step": 520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6551724137931037e-05,
+      "loss": 2.0887,
+      "step": 521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6544962812711295e-05,
+      "loss": 1.7104,
+      "step": 522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.653820148749155e-05,
+      "loss": 1.7704,
+      "step": 523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6531440162271806e-05,
+      "loss": 0.764,
+      "step": 524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6524678837052064e-05,
+      "loss": 1.3044,
+      "step": 525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.651791751183232e-05,
+      "loss": 2.1216,
+      "step": 526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6511156186612578e-05,
+      "loss": 1.3223,
+      "step": 527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6504394861392833e-05,
+      "loss": 1.5024,
+      "step": 528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6497633536173092e-05,
+      "loss": 2.1089,
+      "step": 529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6490872210953347e-05,
+      "loss": 1.6131,
+      "step": 530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6484110885733602e-05,
+      "loss": 1.3869,
+      "step": 531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.647734956051386e-05,
+      "loss": 2.6034,
+      "step": 532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.969,
+      "step": 533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6463826910074375e-05,
+      "loss": 1.4175,
+      "step": 534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6457065584854633e-05,
+      "loss": 1.2031,
+      "step": 535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6450304259634892e-05,
+      "loss": 1.1781,
+      "step": 536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6443542934415147e-05,
+      "loss": 2.1767,
+      "step": 537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6436781609195406e-05,
+      "loss": 2.3127,
+      "step": 538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.643002028397566e-05,
+      "loss": 1.2384,
+      "step": 539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6423258958755916e-05,
+      "loss": 1.5612,
+      "step": 540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6416497633536175e-05,
+      "loss": 2.0691,
+      "step": 541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.640973630831643e-05,
+      "loss": 2.1525,
+      "step": 542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.640297498309669e-05,
+      "loss": 1.7572,
+      "step": 543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6396213657876944e-05,
+      "loss": 1.9971,
+      "step": 544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6389452332657203e-05,
+      "loss": 2.0897,
+      "step": 545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6382691007437458e-05,
+      "loss": 1.6825,
+      "step": 546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6375929682217716e-05,
+      "loss": 1.5327,
+      "step": 547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.636916835699797e-05,
+      "loss": 1.424,
+      "step": 548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.636240703177823e-05,
+      "loss": 1.2404,
+      "step": 549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.635564570655849e-05,
+      "loss": 1.518,
+      "step": 550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6348884381338744e-05,
+      "loss": 2.2082,
+      "step": 551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6342123056119003e-05,
+      "loss": 1.9539,
+      "step": 552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6335361730899258e-05,
+      "loss": 1.6707,
+      "step": 553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6328600405679516e-05,
+      "loss": 1.9389,
+      "step": 554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.632183908045977e-05,
+      "loss": 1.3727,
+      "step": 555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6315077755240027e-05,
+      "loss": 1.759,
+      "step": 556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6308316430020285e-05,
+      "loss": 1.7729,
+      "step": 557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.630155510480054e-05,
+      "loss": 1.5499,
+      "step": 558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.62947937795808e-05,
+      "loss": 1.2365,
+      "step": 559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6288032454361055e-05,
+      "loss": 0.4621,
+      "step": 560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6281271129141313e-05,
+      "loss": 1.0801,
+      "step": 561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.627450980392157e-05,
+      "loss": 1.6172,
+      "step": 562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6267748478701827e-05,
+      "loss": 2.2829,
+      "step": 563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6260987153482086e-05,
+      "loss": 1.586,
+      "step": 564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.625422582826234e-05,
+      "loss": 2.1311,
+      "step": 565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.62474645030426e-05,
+      "loss": 1.4398,
+      "step": 566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6240703177822855e-05,
+      "loss": 1.8494,
+      "step": 567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6233941852603113e-05,
+      "loss": 1.0254,
+      "step": 568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.622718052738337e-05,
+      "loss": 1.3866,
+      "step": 569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6220419202163627e-05,
+      "loss": 1.7235,
+      "step": 570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6213657876943882e-05,
+      "loss": 1.5576,
+      "step": 571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6206896551724137e-05,
+      "loss": 1.7977,
+      "step": 572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6200135226504396e-05,
+      "loss": 0.4826,
+      "step": 573
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.619337390128465e-05,
+      "loss": 1.5074,
+      "step": 574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.618661257606491e-05,
+      "loss": 1.9692,
+      "step": 575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6179851250845165e-05,
+      "loss": 1.1141,
+      "step": 576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6173089925625424e-05,
+      "loss": 2.0325,
+      "step": 577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6166328600405682e-05,
+      "loss": 1.7951,
+      "step": 578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6159567275185938e-05,
+      "loss": 1.0362,
+      "step": 579
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6152805949966196e-05,
+      "loss": 1.5337,
+      "step": 580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.614604462474645e-05,
+      "loss": 1.1632,
+      "step": 581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.613928329952671e-05,
+      "loss": 1.2592,
+      "step": 582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6132521974306965e-05,
+      "loss": 1.5828,
+      "step": 583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6125760649087224e-05,
+      "loss": 2.6294,
+      "step": 584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.611899932386748e-05,
+      "loss": 2.3835,
+      "step": 585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6112237998647738e-05,
+      "loss": 1.585,
+      "step": 586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6105476673427993e-05,
+      "loss": 1.8046,
+      "step": 587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6098715348208248e-05,
+      "loss": 2.3264,
+      "step": 588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6091954022988507e-05,
+      "loss": 1.8357,
+      "step": 589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6085192697768762e-05,
+      "loss": 1.2188,
+      "step": 590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.607843137254902e-05,
+      "loss": 2.7133,
+      "step": 591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.607167004732928e-05,
+      "loss": 1.3834,
+      "step": 592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6064908722109534e-05,
+      "loss": 1.2372,
+      "step": 593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6058147396889793e-05,
+      "loss": 2.9068,
+      "step": 594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6051386071670048e-05,
+      "loss": 1.4611,
+      "step": 595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6044624746450307e-05,
+      "loss": 4.6471,
+      "step": 596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6037863421230562e-05,
+      "loss": 2.8577,
+      "step": 597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.603110209601082e-05,
+      "loss": 2.084,
+      "step": 598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6024340770791076e-05,
+      "loss": 2.8439,
+      "step": 599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6017579445571334e-05,
+      "loss": 1.7366,
+      "step": 600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.601081812035159e-05,
+      "loss": 2.1927,
+      "step": 601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6004056795131848e-05,
+      "loss": 1.5859,
+      "step": 602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5997295469912103e-05,
+      "loss": 3.1694,
+      "step": 603
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.599053414469236e-05,
+      "loss": 2.3848,
+      "step": 604
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5983772819472617e-05,
+      "loss": 2.4531,
+      "step": 605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5977011494252876e-05,
+      "loss": 2.2611,
+      "step": 606
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.597025016903313e-05,
+      "loss": 2.1369,
+      "step": 607
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.596348884381339e-05,
+      "loss": 2.0676,
+      "step": 608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5956727518593645e-05,
+      "loss": 2.2822,
+      "step": 609
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5949966193373903e-05,
+      "loss": 1.0189,
+      "step": 610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.594320486815416e-05,
+      "loss": 1.1214,
+      "step": 611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5936443542934417e-05,
+      "loss": 1.952,
+      "step": 612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5929682217714673e-05,
+      "loss": 1.0556,
+      "step": 613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.592292089249493e-05,
+      "loss": 2.1102,
+      "step": 614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5916159567275186e-05,
+      "loss": 2.0931,
+      "step": 615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5909398242055445e-05,
+      "loss": 1.6915,
+      "step": 616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.59026369168357e-05,
+      "loss": 0.6447,
+      "step": 617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.589587559161596e-05,
+      "loss": 1.0391,
+      "step": 618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5889114266396214e-05,
+      "loss": 0.9371,
+      "step": 619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5882352941176473e-05,
+      "loss": 2.2236,
+      "step": 620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5875591615956728e-05,
+      "loss": 1.5197,
+      "step": 621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5868830290736986e-05,
+      "loss": 1.6166,
+      "step": 622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.586206896551724e-05,
+      "loss": 2.1939,
+      "step": 623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.58553076402975e-05,
+      "loss": 2.1674,
+      "step": 624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.584854631507776e-05,
+      "loss": 1.7331,
+      "step": 625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5841784989858014e-05,
+      "loss": 1.4303,
+      "step": 626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.583502366463827e-05,
+      "loss": 1.3455,
+      "step": 627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5828262339418528e-05,
+      "loss": 1.3253,
+      "step": 628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5821501014198783e-05,
+      "loss": 1.1977,
+      "step": 629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5814739688979042e-05,
+      "loss": 2.1715,
+      "step": 630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5807978363759297e-05,
+      "loss": 1.6682,
+      "step": 631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5801217038539556e-05,
+      "loss": 1.7485,
+      "step": 632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.579445571331981e-05,
+      "loss": 1.7266,
+      "step": 633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.578769438810007e-05,
+      "loss": 1.3007,
+      "step": 634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5780933062880325e-05,
+      "loss": 1.7168,
+      "step": 635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5774171737660583e-05,
+      "loss": 2.271,
+      "step": 636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.576741041244084e-05,
+      "loss": 1.6283,
+      "step": 637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5760649087221097e-05,
+      "loss": 2.1487,
+      "step": 638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.5753887762001356e-05,
+      "loss": 1.3722,
+      "step": 639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.574712643678161e-05,
+      "loss": 2.4187,
+      "step": 640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.574036511156187e-05,
+      "loss": 1.3401,
+      "step": 641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5733603786342125e-05,
+      "loss": 1.2298,
+      "step": 642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.572684246112238e-05,
+      "loss": 2.4407,
+      "step": 643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.572008113590264e-05,
+      "loss": 1.4316,
+      "step": 644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5713319810682894e-05,
+      "loss": 2.3154,
+      "step": 645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5706558485463152e-05,
+      "loss": 1.607,
+      "step": 646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.5699797160243408e-05,
+      "loss": 1.9864,
+      "step": 647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5693035835023666e-05,
+      "loss": 2.1735,
+      "step": 648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.568627450980392e-05,
+      "loss": 1.9867,
+      "step": 649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.567951318458418e-05,
+      "loss": 1.5446,
+      "step": 650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5672751859364435e-05,
+      "loss": 1.4666,
+      "step": 651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5665990534144694e-05,
+      "loss": 2.2551,
+      "step": 652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5659229208924952e-05,
+      "loss": 1.4505,
+      "step": 653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5652467883705208e-05,
+      "loss": 1.4878,
+      "step": 654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5645706558485466e-05,
+      "loss": 2.2949,
+      "step": 655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.563894523326572e-05,
+      "loss": 1.5856,
+      "step": 656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.563218390804598e-05,
+      "loss": 1.467,
+      "step": 657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5625422582826235e-05,
+      "loss": 1.4005,
+      "step": 658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.561866125760649e-05,
+      "loss": 1.8037,
+      "step": 659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.561189993238675e-05,
+      "loss": 2.195,
+      "step": 660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5605138607167004e-05,
+      "loss": 2.0748,
+      "step": 661
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5598377281947263e-05,
+      "loss": 1.9646,
+      "step": 662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5591615956727518e-05,
+      "loss": 2.2082,
+      "step": 663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5584854631507777e-05,
+      "loss": 1.9259,
+      "step": 664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5578093306288032e-05,
+      "loss": 1.0717,
+      "step": 665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.557133198106829e-05,
+      "loss": 1.5453,
+      "step": 666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.556457065584855e-05,
+      "loss": 1.6338,
+      "step": 667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5557809330628804e-05,
+      "loss": 2.3429,
+      "step": 668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5551048005409063e-05,
+      "loss": 1.765,
+      "step": 669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5544286680189318e-05,
+      "loss": 1.3979,
+      "step": 670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5537525354969577e-05,
+      "loss": 1.176,
+      "step": 671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5530764029749832e-05,
+      "loss": 2.1928,
+      "step": 672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.552400270453009e-05,
+      "loss": 1.3461,
+      "step": 673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5517241379310346e-05,
+      "loss": 1.6784,
+      "step": 674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.55104800540906e-05,
+      "loss": 1.6221,
+      "step": 675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.550371872887086e-05,
+      "loss": 1.1023,
+      "step": 676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5496957403651115e-05,
+      "loss": 1.3027,
+      "step": 677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5490196078431373e-05,
+      "loss": 1.5525,
+      "step": 678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.548343475321163e-05,
+      "loss": 1.1876,
+      "step": 679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5476673427991887e-05,
+      "loss": 1.7769,
+      "step": 680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5469912102772146e-05,
+      "loss": 1.3391,
+      "step": 681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.54631507775524e-05,
+      "loss": 1.7948,
+      "step": 682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.545638945233266e-05,
+      "loss": 1.5591,
+      "step": 683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5449628127112915e-05,
+      "loss": 2.4073,
+      "step": 684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5442866801893174e-05,
+      "loss": 2.1989,
+      "step": 685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.543610547667343e-05,
+      "loss": 1.1601,
+      "step": 686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5429344151453687e-05,
+      "loss": 1.8553,
+      "step": 687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5422582826233943e-05,
+      "loss": 1.9692,
+      "step": 688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.54158215010142e-05,
+      "loss": 0.7207,
+      "step": 689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5409060175794456e-05,
+      "loss": 1.2348,
+      "step": 690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.540229885057471e-05,
+      "loss": 1.8967,
+      "step": 691
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.539553752535497e-05,
+      "loss": 2.1095,
+      "step": 692
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5388776200135225e-05,
+      "loss": 1.9602,
+      "step": 693
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5382014874915484e-05,
+      "loss": 1.3271,
+      "step": 694
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5375253549695743e-05,
+      "loss": 2.4121,
+      "step": 695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5368492224475998e-05,
+      "loss": 1.8276,
+      "step": 696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5361730899256256e-05,
+      "loss": 1.7285,
+      "step": 697
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5354969574036515e-05,
+      "loss": 1.8051,
+      "step": 698
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.534820824881677e-05,
+      "loss": 2.3252,
+      "step": 699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5341446923597026e-05,
+      "loss": 0.8626,
+      "step": 700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5334685598377284e-05,
+      "loss": 1.121,
+      "step": 701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.532792427315754e-05,
+      "loss": 1.7585,
+      "step": 702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5321162947937798e-05,
+      "loss": 1.8872,
+      "step": 703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5314401622718053e-05,
+      "loss": 1.9692,
+      "step": 704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5307640297498312e-05,
+      "loss": 1.6365,
+      "step": 705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5300878972278567e-05,
+      "loss": 1.14,
+      "step": 706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5294117647058822e-05,
+      "loss": 1.7152,
+      "step": 707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.528735632183908e-05,
+      "loss": 1.8601,
+      "step": 708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.528059499661934e-05,
+      "loss": 1.089,
+      "step": 709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5273833671399595e-05,
+      "loss": 1.6484,
+      "step": 710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5267072346179853e-05,
+      "loss": 1.4018,
+      "step": 711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5260311020960112e-05,
+      "loss": 0.7714,
+      "step": 712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5253549695740365e-05,
+      "loss": 0.7608,
+      "step": 713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5246788370520624e-05,
+      "loss": 2.2448,
+      "step": 714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5240027045300881e-05,
+      "loss": 2.123,
+      "step": 715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5233265720081136e-05,
+      "loss": 1.8442,
+      "step": 716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5226504394861395e-05,
+      "loss": 1.4663,
+      "step": 717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.521974306964165e-05,
+      "loss": 2.3189,
+      "step": 718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5212981744421909e-05,
+      "loss": 1.2562,
+      "step": 719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5206220419202164e-05,
+      "loss": 1.4882,
+      "step": 720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5199459093982422e-05,
+      "loss": 1.6996,
+      "step": 721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.519269776876268e-05,
+      "loss": 2.0849,
+      "step": 722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5185936443542934e-05,
+      "loss": 2.049,
+      "step": 723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5179175118323193e-05,
+      "loss": 2.5663,
+      "step": 724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5179175118323193e-05,
+      "loss": 3.5387,
+      "step": 725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5172413793103448e-05,
+      "loss": 1.3525,
+      "step": 726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5165652467883707e-05,
+      "loss": 2.2725,
+      "step": 727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5158891142663962e-05,
+      "loss": 1.4887,
+      "step": 728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.515212981744422e-05,
+      "loss": 1.8923,
+      "step": 729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5145368492224478e-05,
+      "loss": 1.6481,
+      "step": 730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5138607167004735e-05,
+      "loss": 1.9452,
+      "step": 731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5131845841784991e-05,
+      "loss": 1.9301,
+      "step": 732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5125084516565247e-05,
+      "loss": 2.4536,
+      "step": 733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5118323191345505e-05,
+      "loss": 1.1097,
+      "step": 734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.511156186612576e-05,
+      "loss": 1.7217,
+      "step": 735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5104800540906019e-05,
+      "loss": 1.6189,
+      "step": 736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5098039215686276e-05,
+      "loss": 1.5671,
+      "step": 737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5091277890466533e-05,
+      "loss": 1.8845,
+      "step": 738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.508451656524679e-05,
+      "loss": 1.7157,
+      "step": 739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5077755240027045e-05,
+      "loss": 1.8018,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5070993914807304e-05,
+      "loss": 1.5253,
+      "step": 741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5064232589587559e-05,
+      "loss": 1.9936,
+      "step": 742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5057471264367817e-05,
+      "loss": 1.4117,
+      "step": 743
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5050709939148074e-05,
+      "loss": 1.9392,
+      "step": 744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5043948613928331e-05,
+      "loss": 1.5718,
+      "step": 745
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5037187288708588e-05,
+      "loss": 1.521,
+      "step": 746
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5030425963488847e-05,
+      "loss": 1.7001,
+      "step": 747
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5023664638269102e-05,
+      "loss": 1.7383,
+      "step": 748
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5016903313049357e-05,
+      "loss": 0.8417,
+      "step": 749
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5010141987829616e-05,
+      "loss": 0.9286,
+      "step": 750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5003380662609873e-05,
+      "loss": 2.2368,
+      "step": 751
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.499661933739013e-05,
+      "loss": 1.372,
+      "step": 752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4989858012170387e-05,
+      "loss": 0.9236,
+      "step": 753
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4983096686950645e-05,
+      "loss": 0.7065,
+      "step": 754
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.49763353617309e-05,
+      "loss": 1.194,
+      "step": 755
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4969574036511156e-05,
+      "loss": 1.3548,
+      "step": 756
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4962812711291414e-05,
+      "loss": 1.5569,
+      "step": 757
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4956051386071671e-05,
+      "loss": 1.7768,
+      "step": 758
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4949290060851928e-05,
+      "loss": 1.0757,
+      "step": 759
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4942528735632185e-05,
+      "loss": 1.0369,
+      "step": 760
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4935767410412444e-05,
+      "loss": 1.046,
+      "step": 761
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4929006085192699e-05,
+      "loss": 1.0622,
+      "step": 762
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4922244759972957e-05,
+      "loss": 0.3932,
+      "step": 763
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4915483434753213e-05,
+      "loss": 1.8611,
+      "step": 764
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.490872210953347e-05,
+      "loss": 1.8844,
+      "step": 765
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4901960784313726e-05,
+      "loss": 1.1678,
+      "step": 766
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4895199459093983e-05,
+      "loss": 0.9868,
+      "step": 767
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4888438133874242e-05,
+      "loss": 0.7262,
+      "step": 768
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4881676808654497e-05,
+      "loss": 1.1194,
+      "step": 769
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4874915483434756e-05,
+      "loss": 1.0573,
+      "step": 770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4868154158215011e-05,
+      "loss": 1.1494,
+      "step": 771
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4861392832995268e-05,
+      "loss": 1.4711,
+      "step": 772
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4854631507775525e-05,
+      "loss": 1.4584,
+      "step": 773
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4847870182555782e-05,
+      "loss": 1.053,
+      "step": 774
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.484110885733604e-05,
+      "loss": 1.5722,
+      "step": 775
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4834347532116296e-05,
+      "loss": 1.5165,
+      "step": 776
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4827586206896554e-05,
+      "loss": 1.2112,
+      "step": 777
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.482082488167681e-05,
+      "loss": 1.7804,
+      "step": 778
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4814063556457068e-05,
+      "loss": 0.7742,
+      "step": 779
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4807302231237323e-05,
+      "loss": 0.6894,
+      "step": 780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.480054090601758e-05,
+      "loss": 1.4917,
+      "step": 781
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4793779580797839e-05,
+      "loss": 1.4558,
+      "step": 782
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4787018255578094e-05,
+      "loss": 1.5638,
+      "step": 783
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4780256930358353e-05,
+      "loss": 1.491,
+      "step": 784
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4773495605138608e-05,
+      "loss": 1.1113,
+      "step": 785
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4766734279918866e-05,
+      "loss": 1.0995,
+      "step": 786
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4759972954699122e-05,
+      "loss": 1.3283,
+      "step": 787
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4753211629479378e-05,
+      "loss": 1.2638,
+      "step": 788
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4746450304259637e-05,
+      "loss": 1.9268,
+      "step": 789
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4739688979039892e-05,
+      "loss": 1.6382,
+      "step": 790
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4732927653820151e-05,
+      "loss": 0.4611,
+      "step": 791
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4726166328600406e-05,
+      "loss": 1.8683,
+      "step": 792
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4719405003380665e-05,
+      "loss": 1.3549,
+      "step": 793
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.471264367816092e-05,
+      "loss": 1.0485,
+      "step": 794
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4705882352941179e-05,
+      "loss": 1.5546,
+      "step": 795
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4699121027721435e-05,
+      "loss": 0.9672,
+      "step": 796
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.469235970250169e-05,
+      "loss": 1.3188,
+      "step": 797
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.468559837728195e-05,
+      "loss": 1.204,
+      "step": 798
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4678837052062205e-05,
+      "loss": 1.7316,
+      "step": 799
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4672075726842463e-05,
+      "loss": 1.8983,
+      "step": 800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4665314401622718e-05,
+      "loss": 1.7458,
+      "step": 801
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4658553076402977e-05,
+      "loss": 1.1672,
+      "step": 802
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4651791751183234e-05,
+      "loss": 1.0615,
+      "step": 803
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4645030425963489e-05,
+      "loss": 0.5914,
+      "step": 804
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4638269100743748e-05,
+      "loss": 2.0877,
+      "step": 805
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4631507775524003e-05,
+      "loss": 0.8406,
+      "step": 806
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4624746450304262e-05,
+      "loss": 0.9182,
+      "step": 807
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4617985125084517e-05,
+      "loss": 0.8732,
+      "step": 808
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4611223799864775e-05,
+      "loss": 1.7811,
+      "step": 809
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4604462474645032e-05,
+      "loss": 1.6807,
+      "step": 810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459770114942529e-05,
+      "loss": 1.3239,
+      "step": 811
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4590939824205546e-05,
+      "loss": 1.4602,
+      "step": 812
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4584178498985801e-05,
+      "loss": 1.2492,
+      "step": 813
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.457741717376606e-05,
+      "loss": 1.4077,
+      "step": 814
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4570655848546315e-05,
+      "loss": 1.5991,
+      "step": 815
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4563894523326574e-05,
+      "loss": 1.1416,
+      "step": 816
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.455713319810683e-05,
+      "loss": 1.042,
+      "step": 817
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550371872887088e-05,
+      "loss": 0.9801,
+      "step": 818
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4543610547667344e-05,
+      "loss": 1.2904,
+      "step": 819
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45368492224476e-05,
+      "loss": 1.0162,
+      "step": 820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4530087897227858e-05,
+      "loss": 1.0022,
+      "step": 821
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4523326572008113e-05,
+      "loss": 2.2613,
+      "step": 822
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4516565246788372e-05,
+      "loss": 0.6859,
+      "step": 823
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4509803921568629e-05,
+      "loss": 1.603,
+      "step": 824
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4503042596348886e-05,
+      "loss": 0.5721,
+      "step": 825
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4496281271129143e-05,
+      "loss": 1.6755,
+      "step": 826
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44895199459094e-05,
+      "loss": 1.0306,
+      "step": 827
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4482758620689657e-05,
+      "loss": 0.7802,
+      "step": 828
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4475997295469912e-05,
+      "loss": 1.3268,
+      "step": 829
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446923597025017e-05,
+      "loss": 1.1148,
+      "step": 830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4462474645030426e-05,
+      "loss": 1.615,
+      "step": 831
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4455713319810684e-05,
+      "loss": 2.3734,
+      "step": 832
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4448951994590941e-05,
+      "loss": 1.1511,
+      "step": 833
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4442190669371198e-05,
+      "loss": 1.7487,
+      "step": 834
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4435429344151455e-05,
+      "loss": 0.61,
+      "step": 835
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.442866801893171e-05,
+      "loss": 1.7435,
+      "step": 836
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4421906693711969e-05,
+      "loss": 1.5092,
+      "step": 837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4415145368492224e-05,
+      "loss": 1.1755,
+      "step": 838
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4408384043272483e-05,
+      "loss": 0.7114,
+      "step": 839
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.440162271805274e-05,
+      "loss": 2.3092,
+      "step": 840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4394861392832996e-05,
+      "loss": 1.4963,
+      "step": 841
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4388100067613253e-05,
+      "loss": 1.1466,
+      "step": 842
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4381338742393512e-05,
+      "loss": 2.1042,
+      "step": 843
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4374577417173767e-05,
+      "loss": 1.4877,
+      "step": 844
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4367816091954022e-05,
+      "loss": 3.1302,
+      "step": 845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4361054766734281e-05,
+      "loss": 2.3891,
+      "step": 846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4354293441514538e-05,
+      "loss": 1.5241,
+      "step": 847
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4347532116294795e-05,
+      "loss": 1.3245,
+      "step": 848
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4340770791075052e-05,
+      "loss": 2.1034,
+      "step": 849
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.433400946585531e-05,
+      "loss": 0.4108,
+      "step": 850
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4327248140635566e-05,
+      "loss": 1.3108,
+      "step": 851
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.432048681541582e-05,
+      "loss": 4.0976,
+      "step": 852
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.431372549019608e-05,
+      "loss": 1.4002,
+      "step": 853
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4306964164976336e-05,
+      "loss": 0.7272,
+      "step": 854
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4300202839756593e-05,
+      "loss": 1.6106,
+      "step": 855
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.429344151453685e-05,
+      "loss": 1.5918,
+      "step": 856
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4286680189317109e-05,
+      "loss": 1.7435,
+      "step": 857
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4279918864097364e-05,
+      "loss": 1.2603,
+      "step": 858
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4273157538877623e-05,
+      "loss": 2.1368,
+      "step": 859
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4266396213657878e-05,
+      "loss": 1.5392,
+      "step": 860
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4259634888438135e-05,
+      "loss": 1.7729,
+      "step": 861
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4252873563218392e-05,
+      "loss": 0.4347,
+      "step": 862
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4246112237998649e-05,
+      "loss": 1.3608,
+      "step": 863
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4239350912778907e-05,
+      "loss": 1.312,
+      "step": 864
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4232589587559162e-05,
+      "loss": 1.604,
+      "step": 865
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4225828262339421e-05,
+      "loss": 1.2876,
+      "step": 866
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4219066937119676e-05,
+      "loss": 1.2131,
+      "step": 867
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4212305611899933e-05,
+      "loss": 1.7553,
+      "step": 868
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.420554428668019e-05,
+      "loss": 1.8521,
+      "step": 869
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4198782961460447e-05,
+      "loss": 1.8591,
+      "step": 870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4192021636240706e-05,
+      "loss": 1.3344,
+      "step": 871
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.418526031102096e-05,
+      "loss": 2.1454,
+      "step": 872
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.417849898580122e-05,
+      "loss": 1.3666,
+      "step": 873
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4171737660581475e-05,
+      "loss": 0.8704,
+      "step": 874
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4164976335361733e-05,
+      "loss": 1.8181,
+      "step": 875
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4158215010141988e-05,
+      "loss": 1.3884,
+      "step": 876
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4151453684922245e-05,
+      "loss": 0.9123,
+      "step": 877
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4144692359702504e-05,
+      "loss": 1.8046,
+      "step": 878
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4137931034482759e-05,
+      "loss": 1.7433,
+      "step": 879
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4131169709263018e-05,
+      "loss": 1.8973,
+      "step": 880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4124408384043273e-05,
+      "loss": 1.436,
+      "step": 881
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.8156,
+      "step": 882
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4110885733603787e-05,
+      "loss": 1.2327,
+      "step": 883
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4104124408384044e-05,
+      "loss": 1.4289,
+      "step": 884
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4097363083164302e-05,
+      "loss": 1.4741,
+      "step": 885
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4090601757944558e-05,
+      "loss": 1.1415,
+      "step": 886
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4083840432724816e-05,
+      "loss": 0.956,
+      "step": 887
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4077079107505071e-05,
+      "loss": 1.4468,
+      "step": 888
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.407031778228533e-05,
+      "loss": 1.1519,
+      "step": 889
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4063556457065585e-05,
+      "loss": 1.5006,
+      "step": 890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4056795131845844e-05,
+      "loss": 1.2558,
+      "step": 891
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.40500338066261e-05,
+      "loss": 0.7549,
+      "step": 892
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4043272481406356e-05,
+      "loss": 1.2231,
+      "step": 893
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4036511156186615e-05,
+      "loss": 1.3661,
+      "step": 894
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.402974983096687e-05,
+      "loss": 0.8829,
+      "step": 895
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4022988505747128e-05,
+      "loss": 1.8678,
+      "step": 896
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4016227180527384e-05,
+      "loss": 0.9866,
+      "step": 897
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4009465855307642e-05,
+      "loss": 1.7451,
+      "step": 898
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4002704530087899e-05,
+      "loss": 1.1273,
+      "step": 899
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3995943204868154e-05,
+      "loss": 0.588,
+      "step": 900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3989181879648413e-05,
+      "loss": 0.5151,
+      "step": 901
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3982420554428668e-05,
+      "loss": 1.4647,
+      "step": 902
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3975659229208927e-05,
+      "loss": 1.6501,
+      "step": 903
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3968897903989182e-05,
+      "loss": 1.6094,
+      "step": 904
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.396213657876944e-05,
+      "loss": 2.9426,
+      "step": 905
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3955375253549697e-05,
+      "loss": 0.7466,
+      "step": 906
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3948613928329954e-05,
+      "loss": 1.1418,
+      "step": 907
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3941852603110211e-05,
+      "loss": 0.7524,
+      "step": 908
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3935091277890466e-05,
+      "loss": 1.8342,
+      "step": 909
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3928329952670725e-05,
+      "loss": 1.0333,
+      "step": 910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.392156862745098e-05,
+      "loss": 0.7567,
+      "step": 911
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3914807302231239e-05,
+      "loss": 0.6909,
+      "step": 912
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3908045977011496e-05,
+      "loss": 0.9408,
+      "step": 913
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3901284651791753e-05,
+      "loss": 1.9746,
+      "step": 914
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.389452332657201e-05,
+      "loss": 1.3582,
+      "step": 915
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3887762001352265e-05,
+      "loss": 1.6783,
+      "step": 916
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3881000676132523e-05,
+      "loss": 0.9482,
+      "step": 917
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3874239350912779e-05,
+      "loss": 1.8208,
+      "step": 918
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3867478025693037e-05,
+      "loss": 1.7632,
+      "step": 919
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3860716700473294e-05,
+      "loss": 0.8158,
+      "step": 920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3853955375253551e-05,
+      "loss": 1.6987,
+      "step": 921
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3847194050033808e-05,
+      "loss": 1.3576,
+      "step": 922
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3840432724814065e-05,
+      "loss": 1.1218,
+      "step": 923
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3833671399594322e-05,
+      "loss": 1.7941,
+      "step": 924
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3826910074374577e-05,
+      "loss": 1.3502,
+      "step": 925
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3820148749154836e-05,
+      "loss": 1.1484,
+      "step": 926
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3813387423935093e-05,
+      "loss": 0.9424,
+      "step": 927
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.380662609871535e-05,
+      "loss": 1.312,
+      "step": 928
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3799864773495606e-05,
+      "loss": 0.8248,
+      "step": 929
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3793103448275863e-05,
+      "loss": 1.774,
+      "step": 930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.378634212305612e-05,
+      "loss": 2.5937,
+      "step": 931
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3779580797836375e-05,
+      "loss": 2.1321,
+      "step": 932
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3772819472616634e-05,
+      "loss": 1.1516,
+      "step": 933
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3766058147396891e-05,
+      "loss": 2.4155,
+      "step": 934
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3759296822177148e-05,
+      "loss": 0.8001,
+      "step": 935
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3752535496957405e-05,
+      "loss": 1.6306,
+      "step": 936
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3745774171737662e-05,
+      "loss": 0.8636,
+      "step": 937
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3739012846517919e-05,
+      "loss": 0.5412,
+      "step": 938
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3732251521298177e-05,
+      "loss": 1.8007,
+      "step": 939
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3725490196078432e-05,
+      "loss": 1.7506,
+      "step": 940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.371872887085869e-05,
+      "loss": 1.6896,
+      "step": 941
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3711967545638946e-05,
+      "loss": 1.2741,
+      "step": 942
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3705206220419203e-05,
+      "loss": 1.3419,
+      "step": 943
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.369844489519946e-05,
+      "loss": 0.8916,
+      "step": 944
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3691683569979717e-05,
+      "loss": 0.6722,
+      "step": 945
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3684922244759976e-05,
+      "loss": 1.0502,
+      "step": 946
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.367816091954023e-05,
+      "loss": 1.141,
+      "step": 947
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3671399594320488e-05,
+      "loss": 1.2495,
+      "step": 948
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3664638269100745e-05,
+      "loss": 1.1402,
+      "step": 949
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3657876943881002e-05,
+      "loss": 1.4758,
+      "step": 950
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3651115618661258e-05,
+      "loss": 1.3699,
+      "step": 951
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3644354293441515e-05,
+      "loss": 0.6146,
+      "step": 952
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3637592968221774e-05,
+      "loss": 1.0364,
+      "step": 953
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.363083164300203e-05,
+      "loss": 0.4451,
+      "step": 954
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3624070317782288e-05,
+      "loss": 1.3729,
+      "step": 955
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3617308992562543e-05,
+      "loss": 0.7743,
+      "step": 956
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.36105476673428e-05,
+      "loss": 1.4092,
+      "step": 957
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3603786342123057e-05,
+      "loss": 0.9928,
+      "step": 958
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3597025016903314e-05,
+      "loss": 1.0226,
+      "step": 959
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3590263691683572e-05,
+      "loss": 0.9312,
+      "step": 960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3583502366463828e-05,
+      "loss": 1.3634,
+      "step": 961
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3576741041244086e-05,
+      "loss": 1.1262,
+      "step": 962
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3569979716024341e-05,
+      "loss": 0.6422,
+      "step": 963
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3563218390804598e-05,
+      "loss": 0.8561,
+      "step": 964
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3556457065584855e-05,
+      "loss": 1.5101,
+      "step": 965
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3549695740365112e-05,
+      "loss": 1.5119,
+      "step": 966
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.354293441514537e-05,
+      "loss": 0.8949,
+      "step": 967
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3536173089925626e-05,
+      "loss": 1.6876,
+      "step": 968
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3529411764705885e-05,
+      "loss": 1.5441,
+      "step": 969
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.352265043948614e-05,
+      "loss": 0.9574,
+      "step": 970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3515889114266398e-05,
+      "loss": 0.7394,
+      "step": 971
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3509127789046654e-05,
+      "loss": 1.9049,
+      "step": 972
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.350236646382691e-05,
+      "loss": 1.5835,
+      "step": 973
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3495605138607169e-05,
+      "loss": 1.2951,
+      "step": 974
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3488843813387424e-05,
+      "loss": 1.3583,
+      "step": 975
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3482082488167683e-05,
+      "loss": 1.3509,
+      "step": 976
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3475321162947938e-05,
+      "loss": 1.2979,
+      "step": 977
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3468559837728197e-05,
+      "loss": 0.9757,
+      "step": 978
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3461798512508452e-05,
+      "loss": 1.4025,
+      "step": 979
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3455037187288709e-05,
+      "loss": 2.0135,
+      "step": 980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3448275862068967e-05,
+      "loss": 1.5932,
+      "step": 981
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3441514536849223e-05,
+      "loss": 1.1537,
+      "step": 982
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3434753211629481e-05,
+      "loss": 1.0217,
+      "step": 983
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3427991886409737e-05,
+      "loss": 2.103,
+      "step": 984
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3421230561189995e-05,
+      "loss": 1.2155,
+      "step": 985
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.341446923597025e-05,
+      "loss": 0.6605,
+      "step": 986
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3407707910750509e-05,
+      "loss": 0.7713,
+      "step": 987
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3400946585530766e-05,
+      "loss": 2.0763,
+      "step": 988
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3394185260311021e-05,
+      "loss": 1.1316,
+      "step": 989
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.338742393509128e-05,
+      "loss": 1.2011,
+      "step": 990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3380662609871535e-05,
+      "loss": 0.8534,
+      "step": 991
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3373901284651794e-05,
+      "loss": 1.3846,
+      "step": 992
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3367139959432049e-05,
+      "loss": 1.332,
+      "step": 993
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3360378634212307e-05,
+      "loss": 1.1373,
+      "step": 994
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3353617308992564e-05,
+      "loss": 0.5123,
+      "step": 995
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.334685598377282e-05,
+      "loss": 1.4421,
+      "step": 996
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3340094658553078e-05,
+      "loss": 1.847,
+      "step": 997
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.214,
+      "step": 998
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3326572008113592e-05,
+      "loss": 1.724,
+      "step": 999
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3319810682893847e-05,
+      "loss": 2.0439,
+      "step": 1000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3313049357674106e-05,
+      "loss": 1.7345,
+      "step": 1001
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3306288032454363e-05,
+      "loss": 1.1802,
+      "step": 1002
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.329952670723462e-05,
+      "loss": 1.6262,
+      "step": 1003
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3292765382014876e-05,
+      "loss": 1.1722,
+      "step": 1004
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3286004056795132e-05,
+      "loss": 1.5292,
+      "step": 1005
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.327924273157539e-05,
+      "loss": 1.1192,
+      "step": 1006
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3272481406355645e-05,
+      "loss": 1.5016,
+      "step": 1007
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3265720081135904e-05,
+      "loss": 1.6346,
+      "step": 1008
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3258958755916161e-05,
+      "loss": 0.26,
+      "step": 1009
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.3252197430696418e-05,
+      "loss": 1.3209,
+      "step": 1010
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3245436105476675e-05,
+      "loss": 1.3543,
+      "step": 1011
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.323867478025693e-05,
+      "loss": 0.8889,
+      "step": 1012
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3231913455037189e-05,
+      "loss": 1.5778,
+      "step": 1013
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3225152129817444e-05,
+      "loss": 1.2576,
+      "step": 1014
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3218390804597702e-05,
+      "loss": 0.3531,
+      "step": 1015
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.321162947937796e-05,
+      "loss": 1.3187,
+      "step": 1016
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.3204868154158216e-05,
+      "loss": 1.0966,
+      "step": 1017
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3198106828938473e-05,
+      "loss": 0.6064,
+      "step": 1018
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3191345503718732e-05,
+      "loss": 0.6013,
+      "step": 1019
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3184584178498987e-05,
+      "loss": 1.5927,
+      "step": 1020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3177822853279242e-05,
+      "loss": 0.726,
+      "step": 1021
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3171061528059501e-05,
+      "loss": 1.1878,
+      "step": 1022
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3164300202839758e-05,
+      "loss": 1.089,
+      "step": 1023
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.3157538877620015e-05,
+      "loss": 0.9471,
+      "step": 1024
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3150777552400272e-05,
+      "loss": 1.3965,
+      "step": 1025
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.314401622718053e-05,
+      "loss": 0.9093,
+      "step": 1026
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3137254901960785e-05,
+      "loss": 1.3893,
+      "step": 1027
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.313049357674104e-05,
+      "loss": 0.6444,
+      "step": 1028
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.31237322515213e-05,
+      "loss": 1.4805,
+      "step": 1029
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3116970926301556e-05,
+      "loss": 1.1096,
+      "step": 1030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.3110209601081813e-05,
+      "loss": 1.3968,
+      "step": 1031
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.310344827586207e-05,
+      "loss": 1.2775,
+      "step": 1032
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3096686950642329e-05,
+      "loss": 1.6956,
+      "step": 1033
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3089925625422584e-05,
+      "loss": 1.3,
+      "step": 1034
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3083164300202842e-05,
+      "loss": 0.7023,
+      "step": 1035
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3076402974983098e-05,
+      "loss": 0.7508,
+      "step": 1036
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3069641649763355e-05,
+      "loss": 1.7604,
+      "step": 1037
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3062880324543611e-05,
+      "loss": 0.7409,
+      "step": 1038
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.3056118999323868e-05,
+      "loss": 1.3865,
+      "step": 1039
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3049357674104127e-05,
+      "loss": 0.9989,
+      "step": 1040
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3042596348884382e-05,
+      "loss": 1.3322,
+      "step": 1041
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.303583502366464e-05,
+      "loss": 0.6654,
+      "step": 1042
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3029073698444896e-05,
+      "loss": 1.2489,
+      "step": 1043
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3022312373225153e-05,
+      "loss": 1.9388,
+      "step": 1044
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.301555104800541e-05,
+      "loss": 2.3571,
+      "step": 1045
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3008789722785667e-05,
+      "loss": 0.9407,
+      "step": 1046
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.3002028397565924e-05,
+      "loss": 1.4646,
+      "step": 1047
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.299526707234618e-05,
+      "loss": 1.6418,
+      "step": 1048
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.298850574712644e-05,
+      "loss": 1.2029,
+      "step": 1049
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2981744421906694e-05,
+      "loss": 1.4225,
+      "step": 1050
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2974983096686953e-05,
+      "loss": 1.7423,
+      "step": 1051
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2968221771467208e-05,
+      "loss": 0.9497,
+      "step": 1052
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2961460446247465e-05,
+      "loss": 1.4331,
+      "step": 1053
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.2954699121027722e-05,
+      "loss": 2.1482,
+      "step": 1054
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2947937795807979e-05,
+      "loss": 0.973,
+      "step": 1055
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 1.1296,
+      "step": 1056
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2934415145368493e-05,
+      "loss": 0.9881,
+      "step": 1057
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2927653820148751e-05,
+      "loss": 1.5184,
+      "step": 1058
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2920892494929007e-05,
+      "loss": 1.3457,
+      "step": 1059
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.2914131169709264e-05,
+      "loss": 1.2312,
+      "step": 1060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.290736984448952e-05,
+      "loss": 1.3788,
+      "step": 1061
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2900608519269777e-05,
+      "loss": 3.1065,
+      "step": 1062
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2893847194050036e-05,
+      "loss": 0.3848,
+      "step": 1063
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2887085868830291e-05,
+      "loss": 0.9718,
+      "step": 1064
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.288032454361055e-05,
+      "loss": 0.6352,
+      "step": 1065
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2873563218390805e-05,
+      "loss": 0.4785,
+      "step": 1066
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2866801893171064e-05,
+      "loss": 1.1391,
+      "step": 1067
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2860040567951319e-05,
+      "loss": 1.8277,
+      "step": 1068
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.2853279242731576e-05,
+      "loss": 1.9056,
+      "step": 1069
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2846517917511834e-05,
+      "loss": 1.1929,
+      "step": 1070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.283975659229209e-05,
+      "loss": 0.9796,
+      "step": 1071
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2832995267072348e-05,
+      "loss": 2.2427,
+      "step": 1072
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2826233941852603e-05,
+      "loss": 1.0603,
+      "step": 1073
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2819472616632862e-05,
+      "loss": 1.6995,
+      "step": 1074
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2812711291413117e-05,
+      "loss": 1.1391,
+      "step": 1075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.2805949966193374e-05,
+      "loss": 1.2813,
+      "step": 1076
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2799188640973633e-05,
+      "loss": 3.8443,
+      "step": 1077
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2792427315753888e-05,
+      "loss": 0.7074,
+      "step": 1078
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2785665990534147e-05,
+      "loss": 1.2703,
+      "step": 1079
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2778904665314402e-05,
+      "loss": 1.4778,
+      "step": 1080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.277214334009466e-05,
+      "loss": 1.0483,
+      "step": 1081
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2765382014874916e-05,
+      "loss": 0.9951,
+      "step": 1082
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2758620689655174e-05,
+      "loss": 0.1963,
+      "step": 1083
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.2751859364435431e-05,
+      "loss": 0.0,
+      "step": 1084
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2745098039215686e-05,
+      "loss": 1.0047,
+      "step": 1085
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2738336713995945e-05,
+      "loss": 1.0446,
+      "step": 1086
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.27315753887762e-05,
+      "loss": 2.0351,
+      "step": 1087
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2724814063556459e-05,
+      "loss": 1.8949,
+      "step": 1088
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2718052738336714e-05,
+      "loss": 1.0983,
+      "step": 1089
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.2711291413116973e-05,
+      "loss": 0.907,
+      "step": 1090
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.270453008789723e-05,
+      "loss": 1.2413,
+      "step": 1091
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2697768762677485e-05,
+      "loss": 0.9077,
+      "step": 1092
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2691007437457743e-05,
+      "loss": 0.8025,
+      "step": 1093
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2684246112237998e-05,
+      "loss": 1.4394,
+      "step": 1094
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2677484787018257e-05,
+      "loss": 1.1018,
+      "step": 1095
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2670723461798512e-05,
+      "loss": 1.8511,
+      "step": 1096
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2663962136578771e-05,
+      "loss": 0.8511,
+      "step": 1097
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.2657200811359028e-05,
+      "loss": 1.6967,
+      "step": 1098
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2650439486139285e-05,
+      "loss": 0.7053,
+      "step": 1099
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2643678160919542e-05,
+      "loss": 1.3023,
+      "step": 1100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2636916835699797e-05,
+      "loss": 0.5415,
+      "step": 1101
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2630155510480055e-05,
+      "loss": 0.8668,
+      "step": 1102
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.262339418526031e-05,
+      "loss": 0.9741,
+      "step": 1103
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.261663286004057e-05,
+      "loss": 1.0109,
+      "step": 1104
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2609871534820826e-05,
+      "loss": 1.2557,
+      "step": 1105
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.2603110209601083e-05,
+      "loss": 0.2722,
+      "step": 1106
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.259634888438134e-05,
+      "loss": 0.9004,
+      "step": 1107
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2589587559161595e-05,
+      "loss": 1.377,
+      "step": 1108
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2582826233941854e-05,
+      "loss": 2.0126,
+      "step": 1109
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2576064908722109e-05,
+      "loss": 1.1118,
+      "step": 1110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2569303583502368e-05,
+      "loss": 0.6805,
+      "step": 1111
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2562542258282625e-05,
+      "loss": 0.6133,
+      "step": 1112
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.2555780933062882e-05,
+      "loss": 1.7789,
+      "step": 1113
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2549019607843138e-05,
+      "loss": 1.2911,
+      "step": 1114
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2542258282623397e-05,
+      "loss": 1.6677,
+      "step": 1115
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2535496957403652e-05,
+      "loss": 1.8582,
+      "step": 1116
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2528735632183907e-05,
+      "loss": 1.2623,
+      "step": 1117
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2521974306964166e-05,
+      "loss": 0.3375,
+      "step": 1118
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2515212981744423e-05,
+      "loss": 1.6088,
+      "step": 1119
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.250845165652468e-05,
+      "loss": 0.9073,
+      "step": 1120
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.2501690331304937e-05,
+      "loss": 0.3548,
+      "step": 1121
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2494929006085195e-05,
+      "loss": 1.4362,
+      "step": 1122
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.248816768086545e-05,
+      "loss": 1.0543,
+      "step": 1123
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2481406355645706e-05,
+      "loss": 1.461,
+      "step": 1124
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2474645030425964e-05,
+      "loss": 0.8389,
+      "step": 1125
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2467883705206221e-05,
+      "loss": 1.5829,
+      "step": 1126
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2461122379986478e-05,
+      "loss": 1.3409,
+      "step": 1127
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.2454361054766735e-05,
+      "loss": 0.6284,
+      "step": 1128
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2447599729546994e-05,
+      "loss": 1.7409,
+      "step": 1129
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2440838404327249e-05,
+      "loss": 1.2976,
+      "step": 1130
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2434077079107508e-05,
+      "loss": 0.5472,
+      "step": 1131
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2427315753887763e-05,
+      "loss": 2.4147,
+      "step": 1132
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.242055442866802e-05,
+      "loss": 1.026,
+      "step": 1133
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2413793103448277e-05,
+      "loss": 0.5381,
+      "step": 1134
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.2407031778228534e-05,
+      "loss": 1.5204,
+      "step": 1135
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2400270453008792e-05,
+      "loss": 5.3318,
+      "step": 1136
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2393509127789047e-05,
+      "loss": 1.7594,
+      "step": 1137
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2386747802569306e-05,
+      "loss": 0.9989,
+      "step": 1138
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2379986477349561e-05,
+      "loss": 0.2974,
+      "step": 1139
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2373225152129818e-05,
+      "loss": 1.8164,
+      "step": 1140
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2366463826910075e-05,
+      "loss": 1.0147,
+      "step": 1141
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.2359702501690332e-05,
+      "loss": 1.5462,
+      "step": 1142
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.235294117647059e-05,
+      "loss": 0.6417,
+      "step": 1143
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2346179851250846e-05,
+      "loss": 0.3342,
+      "step": 1144
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2339418526031104e-05,
+      "loss": 1.8647,
+      "step": 1145
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.233265720081136e-05,
+      "loss": 1.9572,
+      "step": 1146
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2325895875591618e-05,
+      "loss": 0.6584,
+      "step": 1147
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2319134550371873e-05,
+      "loss": 0.4573,
+      "step": 1148
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.231237322515213e-05,
+      "loss": 1.4926,
+      "step": 1149
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.2305611899932389e-05,
+      "loss": 1.3866,
+      "step": 1150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2298850574712644e-05,
+      "loss": 1.0161,
+      "step": 1151
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2292089249492903e-05,
+      "loss": 0.6971,
+      "step": 1152
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2285327924273158e-05,
+      "loss": 0.806,
+      "step": 1153
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2278566599053417e-05,
+      "loss": 1.6153,
+      "step": 1154
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2271805273833672e-05,
+      "loss": 1.2979,
+      "step": 1155
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2265043948613929e-05,
+      "loss": 0.9513,
+      "step": 1156
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2258282623394187e-05,
+      "loss": 1.0648,
+      "step": 1157
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.2251521298174443e-05,
+      "loss": 1.4284,
+      "step": 1158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2244759972954701e-05,
+      "loss": 0.5224,
+      "step": 1159
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2237998647734956e-05,
+      "loss": 2.0116,
+      "step": 1160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2231237322515215e-05,
+      "loss": 1.0965,
+      "step": 1161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.222447599729547e-05,
+      "loss": 1.2249,
+      "step": 1162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2217714672075729e-05,
+      "loss": 1.1023,
+      "step": 1163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2210953346855986e-05,
+      "loss": 1.2884,
+      "step": 1164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.2204192021636241e-05,
+      "loss": 2.1154,
+      "step": 1165
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.21974306964165e-05,
+      "loss": 0.8426,
+      "step": 1166
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2190669371196755e-05,
+      "loss": 0.9194,
+      "step": 1167
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2183908045977013e-05,
+      "loss": 0.7161,
+      "step": 1168
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2177146720757269e-05,
+      "loss": 1.0386,
+      "step": 1169
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2170385395537527e-05,
+      "loss": 1.2246,
+      "step": 1170
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.2163624070317784e-05,
+      "loss": 2.1589,
+      "step": 1171
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.215686274509804e-05,
+      "loss": 1.19,
+      "step": 1172
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2150101419878298e-05,
+      "loss": 0.7341,
+      "step": 1173
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2143340094658553e-05,
+      "loss": 1.3614,
+      "step": 1174
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2136578769438812e-05,
+      "loss": 1.1452,
+      "step": 1175
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2129817444219067e-05,
+      "loss": 0.2143,
+      "step": 1176
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2123056118999326e-05,
+      "loss": 0.8348,
+      "step": 1177
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2116294793779582e-05,
+      "loss": 0.6273,
+      "step": 1178
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.210953346855984e-05,
+      "loss": 0.7591,
+      "step": 1179
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.2102772143340096e-05,
+      "loss": 0.6247,
+      "step": 1180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2096010818120351e-05,
+      "loss": 0.8709,
+      "step": 1181
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.208924949290061e-05,
+      "loss": 1.1725,
+      "step": 1182
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2082488167680865e-05,
+      "loss": 0.616,
+      "step": 1183
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2075726842461124e-05,
+      "loss": 1.5763,
+      "step": 1184
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.206896551724138e-05,
+      "loss": 1.5729,
+      "step": 1185
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2062204192021638e-05,
+      "loss": 1.437,
+      "step": 1186
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2055442866801895e-05,
+      "loss": 1.1272,
+      "step": 1187
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.204868154158215e-05,
+      "loss": 1.5578,
+      "step": 1188
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2041920216362408e-05,
+      "loss": 0.7625,
+      "step": 1189
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2035158891142664e-05,
+      "loss": 1.608,
+      "step": 1190
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2028397565922922e-05,
+      "loss": 1.3188,
+      "step": 1191
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.202163624070318e-05,
+      "loss": 1.7809,
+      "step": 1192
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2014874915483436e-05,
+      "loss": 1.9397,
+      "step": 1193
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2008113590263693e-05,
+      "loss": 1.1338,
+      "step": 1194
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.200135226504395e-05,
+      "loss": 1.2598,
+      "step": 1195
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1994590939824207e-05,
+      "loss": 1.0964,
+      "step": 1196
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1987829614604462e-05,
+      "loss": 1.1513,
+      "step": 1197
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.198106828938472e-05,
+      "loss": 0.7025,
+      "step": 1198
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1974306964164976e-05,
+      "loss": 0.7052,
+      "step": 1199
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1967545638945234e-05,
+      "loss": 1.1071,
+      "step": 1200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1960784313725491e-05,
+      "loss": 1.6684,
+      "step": 1201
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.1954022988505748e-05,
+      "loss": 1.3057,
+      "step": 1202
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1947261663286005e-05,
+      "loss": 0.9129,
+      "step": 1203
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.194050033806626e-05,
+      "loss": 1.3341,
+      "step": 1204
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1933739012846519e-05,
+      "loss": 0.845,
+      "step": 1205
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1926977687626774e-05,
+      "loss": 0.7638,
+      "step": 1206
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1920216362407033e-05,
+      "loss": 1.6056,
+      "step": 1207
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.191345503718729e-05,
+      "loss": 1.368,
+      "step": 1208
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.1906693711967547e-05,
+      "loss": 0.6899,
+      "step": 1209
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1899932386747804e-05,
+      "loss": 0.7784,
+      "step": 1210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1893171061528062e-05,
+      "loss": 1.0131,
+      "step": 1211
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1886409736308317e-05,
+      "loss": 0.9716,
+      "step": 1212
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1879648411088573e-05,
+      "loss": 1.1014,
+      "step": 1213
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1872887085868831e-05,
+      "loss": 1.0,
+      "step": 1214
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1866125760649088e-05,
+      "loss": 0.326,
+      "step": 1215
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1859364435429345e-05,
+      "loss": 1.0353,
+      "step": 1216
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.1852603110209602e-05,
+      "loss": 1.5411,
+      "step": 1217
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.184584178498986e-05,
+      "loss": 0.9009,
+      "step": 1218
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1839080459770116e-05,
+      "loss": 1.3981,
+      "step": 1219
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1832319134550371e-05,
+      "loss": 1.0178,
+      "step": 1220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.182555780933063e-05,
+      "loss": 0.8526,
+      "step": 1221
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1818796484110887e-05,
+      "loss": 1.8224,
+      "step": 1222
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.1812035158891143e-05,
+      "loss": 0.3832,
+      "step": 1223
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.18052738336714e-05,
+      "loss": 1.8509,
+      "step": 1224
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1798512508451659e-05,
+      "loss": 0.6944,
+      "step": 1225
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1791751183231914e-05,
+      "loss": 0.8936,
+      "step": 1226
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1784989858012173e-05,
+      "loss": 1.6488,
+      "step": 1227
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1778228532792428e-05,
+      "loss": 1.8236,
+      "step": 1228
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1771467207572685e-05,
+      "loss": 0.5466,
+      "step": 1229
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 1.205,
+      "step": 1230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1757944557133199e-05,
+      "loss": 1.1125,
+      "step": 1231
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.1751183231913457e-05,
+      "loss": 1.792,
+      "step": 1232
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1744421906693713e-05,
+      "loss": 1.0698,
+      "step": 1233
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1737660581473971e-05,
+      "loss": 1.6623,
+      "step": 1234
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1730899256254226e-05,
+      "loss": 1.4277,
+      "step": 1235
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1724137931034483e-05,
+      "loss": 0.6463,
+      "step": 1236
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.171737660581474e-05,
+      "loss": 0.2027,
+      "step": 1237
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1710615280594997e-05,
+      "loss": 2.2035,
+      "step": 1238
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.1703853955375256e-05,
+      "loss": 2.1366,
+      "step": 1239
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1697092630155511e-05,
+      "loss": 1.2791,
+      "step": 1240
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.169033130493577e-05,
+      "loss": 0.2587,
+      "step": 1241
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1683569979716025e-05,
+      "loss": 1.1183,
+      "step": 1242
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1676808654496283e-05,
+      "loss": 1.4235,
+      "step": 1243
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1670047329276539e-05,
+      "loss": 1.0843,
+      "step": 1244
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1663286004056796e-05,
+      "loss": 1.0538,
+      "step": 1245
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1656524678837054e-05,
+      "loss": 0.9113,
+      "step": 1246
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.164976335361731e-05,
+      "loss": 1.7452,
+      "step": 1247
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1643002028397568e-05,
+      "loss": 0.8763,
+      "step": 1248
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1636240703177823e-05,
+      "loss": 0.2519,
+      "step": 1249
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1629479377958082e-05,
+      "loss": 0.7923,
+      "step": 1250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1622718052738337e-05,
+      "loss": 1.5426,
+      "step": 1251
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1615956727518594e-05,
+      "loss": 0.3512,
+      "step": 1252
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1609195402298852e-05,
+      "loss": 0.9216,
+      "step": 1253
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1602434077079108e-05,
+      "loss": 2.7005,
+      "step": 1254
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1595672751859366e-05,
+      "loss": 1.2295,
+      "step": 1255
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1588911426639622e-05,
+      "loss": 1.4179,
+      "step": 1256
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.158215010141988e-05,
+      "loss": 0.9155,
+      "step": 1257
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1575388776200135e-05,
+      "loss": 0.4519,
+      "step": 1258
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1568627450980394e-05,
+      "loss": 1.1371,
+      "step": 1259
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1561866125760651e-05,
+      "loss": 1.258,
+      "step": 1260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1555104800540906e-05,
+      "loss": 0.3495,
+      "step": 1261
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1548343475321165e-05,
+      "loss": 0.8472,
+      "step": 1262
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.154158215010142e-05,
+      "loss": 1.7322,
+      "step": 1263
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1534820824881679e-05,
+      "loss": 0.2681,
+      "step": 1264
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1528059499661934e-05,
+      "loss": 0.5418,
+      "step": 1265
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1521298174442192e-05,
+      "loss": 1.4484,
+      "step": 1266
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.151453684922245e-05,
+      "loss": 0.9196,
+      "step": 1267
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1507775524002704e-05,
+      "loss": 0.2011,
+      "step": 1268
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1501014198782963e-05,
+      "loss": 0.4247,
+      "step": 1269
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1494252873563218e-05,
+      "loss": 1.5099,
+      "step": 1270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1487491548343477e-05,
+      "loss": 1.6875,
+      "step": 1271
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1480730223123732e-05,
+      "loss": 0.6929,
+      "step": 1272
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.147396889790399e-05,
+      "loss": 0.2921,
+      "step": 1273
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1467207572684248e-05,
+      "loss": 1.1371,
+      "step": 1274
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1460446247464505e-05,
+      "loss": 1.5035,
+      "step": 1275
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.1453684922244761e-05,
+      "loss": 1.243,
+      "step": 1276
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1446923597025017e-05,
+      "loss": 1.4348,
+      "step": 1277
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1440162271805275e-05,
+      "loss": 1.108,
+      "step": 1278
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.143340094658553e-05,
+      "loss": 0.9901,
+      "step": 1279
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1426639621365789e-05,
+      "loss": 1.4512,
+      "step": 1280
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1419878296146046e-05,
+      "loss": 1.2957,
+      "step": 1281
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1413116970926303e-05,
+      "loss": 1.566,
+      "step": 1282
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.140635564570656e-05,
+      "loss": 1.1572,
+      "step": 1283
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1399594320486815e-05,
+      "loss": 1.028,
+      "step": 1284
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1392832995267074e-05,
+      "loss": 1.2674,
+      "step": 1285
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1386071670047329e-05,
+      "loss": 0.9303,
+      "step": 1286
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1379310344827587e-05,
+      "loss": 0.5449,
+      "step": 1287
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1372549019607844e-05,
+      "loss": 1.1604,
+      "step": 1288
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1365787694388101e-05,
+      "loss": 0.6872,
+      "step": 1289
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1359026369168358e-05,
+      "loss": 1.352,
+      "step": 1290
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1352265043948615e-05,
+      "loss": 1.5398,
+      "step": 1291
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1345503718728872e-05,
+      "loss": 1.5131,
+      "step": 1292
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1338742393509127e-05,
+      "loss": 1.0487,
+      "step": 1293
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1331981068289386e-05,
+      "loss": 0.652,
+      "step": 1294
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1325219743069643e-05,
+      "loss": 1.2205,
+      "step": 1295
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.13184584178499e-05,
+      "loss": 0.8233,
+      "step": 1296
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1311697092630157e-05,
+      "loss": 1.1672,
+      "step": 1297
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1304935767410414e-05,
+      "loss": 1.916,
+      "step": 1298
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.129817444219067e-05,
+      "loss": 0.6937,
+      "step": 1299
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1291413116970926e-05,
+      "loss": 1.0831,
+      "step": 1300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1284651791751184e-05,
+      "loss": 1.9853,
+      "step": 1301
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1277890466531441e-05,
+      "loss": 1.6617,
+      "step": 1302
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1271129141311698e-05,
+      "loss": 0.8116,
+      "step": 1303
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1264367816091955e-05,
+      "loss": 0.9155,
+      "step": 1304
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1257606490872212e-05,
+      "loss": 0.9887,
+      "step": 1305
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1250845165652469e-05,
+      "loss": 0.9805,
+      "step": 1306
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1244083840432727e-05,
+      "loss": 0.7297,
+      "step": 1307
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1237322515212983e-05,
+      "loss": 0.8264,
+      "step": 1308
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.123056118999324e-05,
+      "loss": 1.4784,
+      "step": 1309
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1223799864773496e-05,
+      "loss": 0.903,
+      "step": 1310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1217038539553753e-05,
+      "loss": 0.4492,
+      "step": 1311
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.121027721433401e-05,
+      "loss": 0.492,
+      "step": 1312
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1203515889114267e-05,
+      "loss": 1.173,
+      "step": 1313
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1196754563894526e-05,
+      "loss": 0.2272,
+      "step": 1314
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1189993238674781e-05,
+      "loss": 0.4815,
+      "step": 1315
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1183231913455038e-05,
+      "loss": 1.1507,
+      "step": 1316
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1176470588235295e-05,
+      "loss": 0.5358,
+      "step": 1317
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1169709263015552e-05,
+      "loss": 0.561,
+      "step": 1318
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1162947937795809e-05,
+      "loss": 1.1718,
+      "step": 1319
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1156186612576066e-05,
+      "loss": 1.687,
+      "step": 1320
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1149425287356324e-05,
+      "loss": 1.7124,
+      "step": 1321
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.114266396213658e-05,
+      "loss": 0.5849,
+      "step": 1322
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1135902636916838e-05,
+      "loss": 0.8265,
+      "step": 1323
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1129141311697093e-05,
+      "loss": 0.7763,
+      "step": 1324
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.112237998647735e-05,
+      "loss": 0.8224,
+      "step": 1325
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1115618661257607e-05,
+      "loss": 0.6241,
+      "step": 1326
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1108857336037864e-05,
+      "loss": 1.2162,
+      "step": 1327
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1102096010818123e-05,
+      "loss": 0.6498,
+      "step": 1328
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1095334685598378e-05,
+      "loss": 0.778,
+      "step": 1329
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1088573360378636e-05,
+      "loss": 0.9136,
+      "step": 1330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1081812035158892e-05,
+      "loss": 1.3413,
+      "step": 1331
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1075050709939149e-05,
+      "loss": 1.1782,
+      "step": 1332
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1068289384719405e-05,
+      "loss": 0.9905,
+      "step": 1333
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1061528059499662e-05,
+      "loss": 1.9638,
+      "step": 1334
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1054766734279921e-05,
+      "loss": 0.489,
+      "step": 1335
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1048005409060176e-05,
+      "loss": 0.9386,
+      "step": 1336
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1041244083840435e-05,
+      "loss": 0.3841,
+      "step": 1337
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.103448275862069e-05,
+      "loss": 1.2497,
+      "step": 1338
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1027721433400949e-05,
+      "loss": 1.8799,
+      "step": 1339
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1020960108181204e-05,
+      "loss": 1.9151,
+      "step": 1340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.101419878296146e-05,
+      "loss": 0.5857,
+      "step": 1341
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.100743745774172e-05,
+      "loss": 3.0695,
+      "step": 1342
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1000676132521975e-05,
+      "loss": 0.6458,
+      "step": 1343
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0993914807302233e-05,
+      "loss": 0.792,
+      "step": 1344
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0987153482082488e-05,
+      "loss": 3.417,
+      "step": 1345
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0980392156862747e-05,
+      "loss": 1.9139,
+      "step": 1346
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0973630831643002e-05,
+      "loss": 1.3008,
+      "step": 1347
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0966869506423259e-05,
+      "loss": 2.7971,
+      "step": 1348
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0960108181203518e-05,
+      "loss": 1.0037,
+      "step": 1349
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0953346855983773e-05,
+      "loss": 1.3116,
+      "step": 1350
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0946585530764032e-05,
+      "loss": 0.8784,
+      "step": 1351
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0939824205544287e-05,
+      "loss": 0.3156,
+      "step": 1352
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0933062880324545e-05,
+      "loss": 0.6775,
+      "step": 1353
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.09263015551048e-05,
+      "loss": 0.9792,
+      "step": 1354
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.091954022988506e-05,
+      "loss": 1.5311,
+      "step": 1355
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0912778904665316e-05,
+      "loss": 1.4844,
+      "step": 1356
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.0906017579445571e-05,
+      "loss": 0.9198,
+      "step": 1357
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.089925625422583e-05,
+      "loss": 1.275,
+      "step": 1358
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0892494929006085e-05,
+      "loss": 0.2187,
+      "step": 1359
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0885733603786344e-05,
+      "loss": 0.552,
+      "step": 1360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0878972278566599e-05,
+      "loss": 1.5368,
+      "step": 1361
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0872210953346858e-05,
+      "loss": 1.2785,
+      "step": 1362
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0865449628127114e-05,
+      "loss": 0.8987,
+      "step": 1363
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.085868830290737e-05,
+      "loss": 1.6315,
+      "step": 1364
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.0851926977687628e-05,
+      "loss": 0.5914,
+      "step": 1365
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0845165652467883e-05,
+      "loss": 1.048,
+      "step": 1366
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0838404327248142e-05,
+      "loss": 0.6059,
+      "step": 1367
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0831643002028397e-05,
+      "loss": 0.7959,
+      "step": 1368
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0824881676808656e-05,
+      "loss": 1.041,
+      "step": 1369
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0818120351588913e-05,
+      "loss": 0.9777,
+      "step": 1370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.081135902636917e-05,
+      "loss": 0.7285,
+      "step": 1371
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.0804597701149427e-05,
+      "loss": 0.6275,
+      "step": 1372
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0797836375929682e-05,
+      "loss": 1.1929,
+      "step": 1373
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.079107505070994e-05,
+      "loss": 0.9544,
+      "step": 1374
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0784313725490196e-05,
+      "loss": 1.1468,
+      "step": 1375
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0777552400270454e-05,
+      "loss": 0.7814,
+      "step": 1376
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0770791075050711e-05,
+      "loss": 0.2193,
+      "step": 1377
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0764029749830968e-05,
+      "loss": 1.3109,
+      "step": 1378
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0757268424611225e-05,
+      "loss": 1.7463,
+      "step": 1379
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.075050709939148e-05,
+      "loss": 1.186,
+      "step": 1380
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0743745774171739e-05,
+      "loss": 1.324,
+      "step": 1381
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0736984448951994e-05,
+      "loss": 1.3078,
+      "step": 1382
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0730223123732253e-05,
+      "loss": 0.7904,
+      "step": 1383
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.072346179851251e-05,
+      "loss": 1.3155,
+      "step": 1384
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0716700473292767e-05,
+      "loss": 0.4275,
+      "step": 1385
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0709939148073023e-05,
+      "loss": 1.3453,
+      "step": 1386
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0703177822853282e-05,
+      "loss": 1.0574,
+      "step": 1387
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0696416497633537e-05,
+      "loss": 0.6466,
+      "step": 1388
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0689655172413792e-05,
+      "loss": 1.2077,
+      "step": 1389
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0682893847194051e-05,
+      "loss": 0.4759,
+      "step": 1390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0676132521974308e-05,
+      "loss": 0.5142,
+      "step": 1391
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0669371196754565e-05,
+      "loss": 1.4383,
+      "step": 1392
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0662609871534822e-05,
+      "loss": 0.4835,
+      "step": 1393
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.065584854631508e-05,
+      "loss": 0.4518,
+      "step": 1394
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0649087221095336e-05,
+      "loss": 1.1027,
+      "step": 1395
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.064232589587559e-05,
+      "loss": 1.6866,
+      "step": 1396
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.063556457065585e-05,
+      "loss": 0.9105,
+      "step": 1397
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0628803245436106e-05,
+      "loss": 0.863,
+      "step": 1398
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0622041920216363e-05,
+      "loss": 0.4719,
+      "step": 1399
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.061528059499662e-05,
+      "loss": 1.0842,
+      "step": 1400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0608519269776879e-05,
+      "loss": 0.1429,
+      "step": 1401
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0601757944557134e-05,
+      "loss": 0.6055,
+      "step": 1402
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0594996619337393e-05,
+      "loss": 2.0232,
+      "step": 1403
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.3788,
+      "step": 1404
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0581473968897905e-05,
+      "loss": 0.5079,
+      "step": 1405
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0574712643678162e-05,
+      "loss": 1.1344,
+      "step": 1406
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0567951318458419e-05,
+      "loss": 0.6525,
+      "step": 1407
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0561189993238677e-05,
+      "loss": 1.6188,
+      "step": 1408
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0554428668018932e-05,
+      "loss": 0.9842,
+      "step": 1409
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0547667342799191e-05,
+      "loss": 1.1054,
+      "step": 1410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0540906017579446e-05,
+      "loss": 1.1978,
+      "step": 1411
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0534144692359703e-05,
+      "loss": 0.2405,
+      "step": 1412
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.052738336713996e-05,
+      "loss": 0.6766,
+      "step": 1413
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0520622041920217e-05,
+      "loss": 0.741,
+      "step": 1414
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0513860716700476e-05,
+      "loss": 0.2201,
+      "step": 1415
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.050709939148073e-05,
+      "loss": 0.4589,
+      "step": 1416
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.050033806626099e-05,
+      "loss": 0.645,
+      "step": 1417
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0493576741041245e-05,
+      "loss": 1.1159,
+      "step": 1418
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0486815415821503e-05,
+      "loss": 0.4521,
+      "step": 1419
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0480054090601758e-05,
+      "loss": 0.844,
+      "step": 1420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0473292765382015e-05,
+      "loss": 0.8715,
+      "step": 1421
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0466531440162272e-05,
+      "loss": 1.1242,
+      "step": 1422
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0459770114942529e-05,
+      "loss": 0.2137,
+      "step": 1423
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0453008789722788e-05,
+      "loss": 1.405,
+      "step": 1424
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0446247464503043e-05,
+      "loss": 0.8538,
+      "step": 1425
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0439486139283302e-05,
+      "loss": 0.8066,
+      "step": 1426
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0432724814063557e-05,
+      "loss": 0.7797,
+      "step": 1427
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0425963488843814e-05,
+      "loss": 1.2241,
+      "step": 1428
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.041920216362407e-05,
+      "loss": 1.4155,
+      "step": 1429
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0412440838404328e-05,
+      "loss": 0.7269,
+      "step": 1430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0405679513184586e-05,
+      "loss": 1.0485,
+      "step": 1431
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0398918187964841e-05,
+      "loss": 1.7019,
+      "step": 1432
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.03921568627451e-05,
+      "loss": 1.6052,
+      "step": 1433
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0385395537525355e-05,
+      "loss": 0.8631,
+      "step": 1434
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0378634212305614e-05,
+      "loss": 0.9874,
+      "step": 1435
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0371872887085869e-05,
+      "loss": 0.9306,
+      "step": 1436
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0365111561866126e-05,
+      "loss": 1.3352,
+      "step": 1437
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0358350236646385e-05,
+      "loss": 2.1567,
+      "step": 1438
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.035158891142664e-05,
+      "loss": 0.7349,
+      "step": 1439
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0344827586206898e-05,
+      "loss": 0.3906,
+      "step": 1440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0338066260987154e-05,
+      "loss": 1.1797,
+      "step": 1441
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0331304935767412e-05,
+      "loss": 0.9033,
+      "step": 1442
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0324543610547667e-05,
+      "loss": 1.9248,
+      "step": 1443
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0317782285327924e-05,
+      "loss": 1.5378,
+      "step": 1444
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0311020960108183e-05,
+      "loss": 0.5644,
+      "step": 1445
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0304259634888438e-05,
+      "loss": 0.9894,
+      "step": 1446
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0297498309668697e-05,
+      "loss": 0.9518,
+      "step": 1447
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0290736984448952e-05,
+      "loss": 1.0687,
+      "step": 1448
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.028397565922921e-05,
+      "loss": 1.4276,
+      "step": 1449
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0277214334009466e-05,
+      "loss": 1.2222,
+      "step": 1450
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0270453008789724e-05,
+      "loss": 0.9787,
+      "step": 1451
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0263691683569981e-05,
+      "loss": 1.0112,
+      "step": 1452
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0256930358350236e-05,
+      "loss": 0.8983,
+      "step": 1453
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0250169033130495e-05,
+      "loss": 1.6428,
+      "step": 1454
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.024340770791075e-05,
+      "loss": 1.6104,
+      "step": 1455
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0236646382691009e-05,
+      "loss": 0.8364,
+      "step": 1456
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0229885057471264e-05,
+      "loss": 0.5095,
+      "step": 1457
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0223123732251523e-05,
+      "loss": 1.2192,
+      "step": 1458
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.021636240703178e-05,
+      "loss": 1.4937,
+      "step": 1459
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0209601081812035e-05,
+      "loss": 1.1206,
+      "step": 1460
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0202839756592293e-05,
+      "loss": 0.6077,
+      "step": 1461
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0196078431372549e-05,
+      "loss": 0.2392,
+      "step": 1462
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0189317106152807e-05,
+      "loss": 1.1027,
+      "step": 1463
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0182555780933063e-05,
+      "loss": 1.3158,
+      "step": 1464
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0175794455713321e-05,
+      "loss": 0.6179,
+      "step": 1465
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0169033130493578e-05,
+      "loss": 2.079,
+      "step": 1466
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0162271805273835e-05,
+      "loss": 0.8314,
+      "step": 1467
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0155510480054092e-05,
+      "loss": 1.4057,
+      "step": 1468
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0148749154834347e-05,
+      "loss": 1.2091,
+      "step": 1469
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0141987829614606e-05,
+      "loss": 0.172,
+      "step": 1470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0135226504394861e-05,
+      "loss": 0.7872,
+      "step": 1471
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.012846517917512e-05,
+      "loss": 1.6877,
+      "step": 1472
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0121703853955376e-05,
+      "loss": 0.7751,
+      "step": 1473
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0114942528735633e-05,
+      "loss": 0.9094,
+      "step": 1474
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.010818120351589e-05,
+      "loss": 0.9875,
+      "step": 1475
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0101419878296145e-05,
+      "loss": 0.9045,
+      "step": 1476
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0094658553076404e-05,
+      "loss": 0.6508,
+      "step": 1477
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.008789722785666e-05,
+      "loss": 0.161,
+      "step": 1478
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0081135902636918e-05,
+      "loss": 0.8232,
+      "step": 1479
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0074374577417175e-05,
+      "loss": 1.6111,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0067613252197432e-05,
+      "loss": 0.8024,
+      "step": 1481
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0060851926977689e-05,
+      "loss": 0.749,
+      "step": 1482
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0054090601757947e-05,
+      "loss": 0.5078,
+      "step": 1483
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0047329276538202e-05,
+      "loss": 1.0107,
+      "step": 1484
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0040567951318458e-05,
+      "loss": 1.0314,
+      "step": 1485
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0033806626098716e-05,
+      "loss": 0.3119,
+      "step": 1486
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0027045300878973e-05,
+      "loss": 0.8978,
+      "step": 1487
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.002028397565923e-05,
+      "loss": 0.698,
+      "step": 1488
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0013522650439487e-05,
+      "loss": 0.3373,
+      "step": 1489
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0006761325219746e-05,
+      "loss": 0.3774,
+      "step": 1490
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1e-05,
+      "loss": 0.5626,
+      "step": 1491
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.993238674780258e-06,
+      "loss": 0.3701,
+      "step": 1492
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.986477349560515e-06,
+      "loss": 0.7972,
+      "step": 1493
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.979716024340772e-06,
+      "loss": 0.211,
+      "step": 1494
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.972954699121028e-06,
+      "loss": 0.5233,
+      "step": 1495
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.966193373901285e-06,
+      "loss": 0.872,
+      "step": 1496
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.959432048681542e-06,
+      "loss": 0.9572,
+      "step": 1497
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.9526707234618e-06,
+      "loss": 0.4601,
+      "step": 1498
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.945909398242056e-06,
+      "loss": 0.7683,
+      "step": 1499
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.939148073022313e-06,
+      "loss": 0.5382,
+      "step": 1500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.93238674780257e-06,
+      "loss": 0.8832,
+      "step": 1501
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.925625422582827e-06,
+      "loss": 0.5312,
+      "step": 1502
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.918864097363084e-06,
+      "loss": 0.5637,
+      "step": 1503
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.91210277214334e-06,
+      "loss": 2.671,
+      "step": 1504
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.905341446923598e-06,
+      "loss": 0.4623,
+      "step": 1505
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.898580121703854e-06,
+      "loss": 0.6044,
+      "step": 1506
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.891818796484111e-06,
+      "loss": 0.7576,
+      "step": 1507
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.885057471264368e-06,
+      "loss": 0.7561,
+      "step": 1508
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.878296146044625e-06,
+      "loss": 0.1808,
+      "step": 1509
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.871534820824882e-06,
+      "loss": 0.4482,
+      "step": 1510
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.864773495605139e-06,
+      "loss": 0.3099,
+      "step": 1511
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.858012170385396e-06,
+      "loss": 0.1814,
+      "step": 1512
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.851250845165653e-06,
+      "loss": 0.2045,
+      "step": 1513
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.84448951994591e-06,
+      "loss": 0.584,
+      "step": 1514
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.837728194726167e-06,
+      "loss": 0.6864,
+      "step": 1515
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.830966869506424e-06,
+      "loss": 0.7463,
+      "step": 1516
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.82420554428668e-06,
+      "loss": 1.1149,
+      "step": 1517
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.817444219066939e-06,
+      "loss": 0.5235,
+      "step": 1518
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.810682893847194e-06,
+      "loss": 0.679,
+      "step": 1519
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 0.4443,
+      "step": 1520
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.797160243407708e-06,
+      "loss": 0.5425,
+      "step": 1521
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.790398918187965e-06,
+      "loss": 0.5504,
+      "step": 1522
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.783637592968222e-06,
+      "loss": 0.5385,
+      "step": 1523
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.776876267748479e-06,
+      "loss": 1.13,
+      "step": 1524
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.770114942528738e-06,
+      "loss": 0.5767,
+      "step": 1525
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.763353617308994e-06,
+      "loss": 0.8084,
+      "step": 1526
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.75659229208925e-06,
+      "loss": 0.6125,
+      "step": 1527
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.749830966869507e-06,
+      "loss": 0.3697,
+      "step": 1528
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.743069641649763e-06,
+      "loss": 0.4523,
+      "step": 1529
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.73630831643002e-06,
+      "loss": 0.5258,
+      "step": 1530
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.729546991210277e-06,
+      "loss": 0.65,
+      "step": 1531
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.722785665990536e-06,
+      "loss": 0.7155,
+      "step": 1532
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.716024340770793e-06,
+      "loss": 0.6186,
+      "step": 1533
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.70926301555105e-06,
+      "loss": 0.6566,
+      "step": 1534
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.702501690331305e-06,
+      "loss": 0.4876,
+      "step": 1535
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.695740365111562e-06,
+      "loss": 0.3706,
+      "step": 1536
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.688979039891819e-06,
+      "loss": 0.4438,
+      "step": 1537
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.682217714672076e-06,
+      "loss": 0.2373,
+      "step": 1538
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.675456389452334e-06,
+      "loss": 1.1616,
+      "step": 1539
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.668695064232591e-06,
+      "loss": 1.0402,
+      "step": 1540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.661933739012848e-06,
+      "loss": 0.6054,
+      "step": 1541
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.655172413793105e-06,
+      "loss": 0.5324,
+      "step": 1542
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.64841108857336e-06,
+      "loss": 1.5676,
+      "step": 1543
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.641649763353617e-06,
+      "loss": 0.6546,
+      "step": 1544
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.634888438133874e-06,
+      "loss": 0.7707,
+      "step": 1545
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.628127112914133e-06,
+      "loss": 0.7948,
+      "step": 1546
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.62136578769439e-06,
+      "loss": 0.4801,
+      "step": 1547
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.614604462474646e-06,
+      "loss": 0.1423,
+      "step": 1548
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.607843137254903e-06,
+      "loss": 0.3337,
+      "step": 1549
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.60108181203516e-06,
+      "loss": 0.0569,
+      "step": 1550
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.594320486815416e-06,
+      "loss": 0.6925,
+      "step": 1551
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.587559161595672e-06,
+      "loss": 0.282,
+      "step": 1552
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580797836375931e-06,
+      "loss": 1.2124,
+      "step": 1553
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.574036511156188e-06,
+      "loss": 0.5237,
+      "step": 1554
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.567275185936445e-06,
+      "loss": 0.6081,
+      "step": 1555
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.560513860716702e-06,
+      "loss": 1.4845,
+      "step": 1556
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.553752535496959e-06,
+      "loss": 0.6682,
+      "step": 1557
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.546991210277216e-06,
+      "loss": 0.7874,
+      "step": 1558
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.54022988505747e-06,
+      "loss": 0.5972,
+      "step": 1559
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.53346855983773e-06,
+      "loss": 1.2085,
+      "step": 1560
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.526707234617986e-06,
+      "loss": 0.5025,
+      "step": 1561
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.519945909398243e-06,
+      "loss": 0.2948,
+      "step": 1562
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.5131845841785e-06,
+      "loss": 0.166,
+      "step": 1563
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.506423258958757e-06,
+      "loss": 0.9009,
+      "step": 1564
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.499661933739014e-06,
+      "loss": 0.2773,
+      "step": 1565
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.492900608519271e-06,
+      "loss": 0.9972,
+      "step": 1566
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.486139283299526e-06,
+      "loss": 0.7336,
+      "step": 1567
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.479377958079785e-06,
+      "loss": 0.4192,
+      "step": 1568
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.472616632860042e-06,
+      "loss": 0.6386,
+      "step": 1569
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.465855307640299e-06,
+      "loss": 0.8922,
+      "step": 1570
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.459093982420555e-06,
+      "loss": 0.614,
+      "step": 1571
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.452332657200812e-06,
+      "loss": 0.6769,
+      "step": 1572
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.44557133198107e-06,
+      "loss": 0.413,
+      "step": 1573
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.438810006761326e-06,
+      "loss": 0.5351,
+      "step": 1574
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.432048681541583e-06,
+      "loss": 0.4692,
+      "step": 1575
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.42528735632184e-06,
+      "loss": 1.1011,
+      "step": 1576
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.418526031102097e-06,
+      "loss": 0.7505,
+      "step": 1577
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.9667,
+      "step": 1578
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.40500338066261e-06,
+      "loss": 0.6953,
+      "step": 1579
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.398242055442868e-06,
+      "loss": 0.9609,
+      "step": 1580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.391480730223125e-06,
+      "loss": 0.4182,
+      "step": 1581
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.384719405003381e-06,
+      "loss": 0.4127,
+      "step": 1582
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.377958079783638e-06,
+      "loss": 1.2465,
+      "step": 1583
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.371196754563895e-06,
+      "loss": 0.7562,
+      "step": 1584
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.364435429344152e-06,
+      "loss": 0.5132,
+      "step": 1585
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.357674104124409e-06,
+      "loss": 0.9858,
+      "step": 1586
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.350912778904666e-06,
+      "loss": 0.3883,
+      "step": 1587
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.344151453684923e-06,
+      "loss": 0.4958,
+      "step": 1588
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.33739012846518e-06,
+      "loss": 0.2286,
+      "step": 1589
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.330628803245437e-06,
+      "loss": 0.4581,
+      "step": 1590
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.323867478025694e-06,
+      "loss": 0.3855,
+      "step": 1591
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.31710615280595e-06,
+      "loss": 0.2908,
+      "step": 1592
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.310344827586207e-06,
+      "loss": 0.5176,
+      "step": 1593
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.303583502366464e-06,
+      "loss": 0.4547,
+      "step": 1594
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.296822177146721e-06,
+      "loss": 0.4963,
+      "step": 1595
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.290060851926978e-06,
+      "loss": 0.5809,
+      "step": 1596
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.283299526707235e-06,
+      "loss": 1.0031,
+      "step": 1597
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.276538201487492e-06,
+      "loss": 0.4331,
+      "step": 1598
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.269776876267749e-06,
+      "loss": 0.3911,
+      "step": 1599
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.263015551048006e-06,
+      "loss": 0.523,
+      "step": 1600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.256254225828263e-06,
+      "loss": 1.108,
+      "step": 1601
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.24949290060852e-06,
+      "loss": 0.6517,
+      "step": 1602
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.242731575388777e-06,
+      "loss": 0.8096,
+      "step": 1603
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.235970250169034e-06,
+      "loss": 1.4282,
+      "step": 1604
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.22920892494929e-06,
+      "loss": 0.9614,
+      "step": 1605
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.222447599729547e-06,
+      "loss": 0.5364,
+      "step": 1606
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.215686274509804e-06,
+      "loss": 0.6286,
+      "step": 1607
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.208924949290061e-06,
+      "loss": 0.6194,
+      "step": 1608
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.202163624070318e-06,
+      "loss": 0.4022,
+      "step": 1609
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.195402298850575e-06,
+      "loss": 0.436,
+      "step": 1610
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.188640973630832e-06,
+      "loss": 0.4933,
+      "step": 1611
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.181879648411089e-06,
+      "loss": 0.6159,
+      "step": 1612
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.175118323191346e-06,
+      "loss": 0.451,
+      "step": 1613
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.168356997971604e-06,
+      "loss": 0.2853,
+      "step": 1614
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.16159567275186e-06,
+      "loss": 0.3411,
+      "step": 1615
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.154834347532116e-06,
+      "loss": 0.7086,
+      "step": 1616
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.148073022312373e-06,
+      "loss": 0.5537,
+      "step": 1617
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.14131169709263e-06,
+      "loss": 0.3773,
+      "step": 1618
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.134550371872887e-06,
+      "loss": 0.3254,
+      "step": 1619
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.127789046653144e-06,
+      "loss": 0.6021,
+      "step": 1620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.121027721433403e-06,
+      "loss": 0.4092,
+      "step": 1621
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.11426639621366e-06,
+      "loss": 1.0266,
+      "step": 1622
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.107505070993915e-06,
+      "loss": 0.496,
+      "step": 1623
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.100743745774172e-06,
+      "loss": 0.8993,
+      "step": 1624
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.093982420554429e-06,
+      "loss": 1.392,
+      "step": 1625
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.087221095334686e-06,
+      "loss": 1.142,
+      "step": 1626
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.080459770114942e-06,
+      "loss": 0.885,
+      "step": 1627
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.073698444895201e-06,
+      "loss": 0.151,
+      "step": 1628
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.066937119675458e-06,
+      "loss": 0.621,
+      "step": 1629
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.060175794455715e-06,
+      "loss": 0.5456,
+      "step": 1630
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.05341446923597e-06,
+      "loss": 0.7199,
+      "step": 1631
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.046653144016227e-06,
+      "loss": 1.029,
+      "step": 1632
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.039891818796484e-06,
+      "loss": 1.4984,
+      "step": 1633
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.033130493576741e-06,
+      "loss": 0.1781,
+      "step": 1634
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.026369168357e-06,
+      "loss": 0.7907,
+      "step": 1635
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.019607843137256e-06,
+      "loss": 1.1752,
+      "step": 1636
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.012846517917513e-06,
+      "loss": 1.3145,
+      "step": 1637
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.00608519269777e-06,
+      "loss": 0.3095,
+      "step": 1638
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.999323867478025e-06,
+      "loss": 0.4205,
+      "step": 1639
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.992562542258282e-06,
+      "loss": 0.5151,
+      "step": 1640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.98580121703854e-06,
+      "loss": 1.0477,
+      "step": 1641
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.979039891818798e-06,
+      "loss": 0.0293,
+      "step": 1642
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.972278566599055e-06,
+      "loss": 1.1403,
+      "step": 1643
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.965517241379312e-06,
+      "loss": 0.4188,
+      "step": 1644
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.958755916159569e-06,
+      "loss": 1.1594,
+      "step": 1645
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.951994590939825e-06,
+      "loss": 0.2285,
+      "step": 1646
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.94523326572008e-06,
+      "loss": 0.2757,
+      "step": 1647
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.938471940500338e-06,
+      "loss": 1.4073,
+      "step": 1648
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.931710615280596e-06,
+      "loss": 1.3083,
+      "step": 1649
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.924949290060853e-06,
+      "loss": 0.6682,
+      "step": 1650
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.91818796484111e-06,
+      "loss": 0.3647,
+      "step": 1651
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.911426639621367e-06,
+      "loss": 1.051,
+      "step": 1652
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.904665314401624e-06,
+      "loss": 0.6024,
+      "step": 1653
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.89790398918188e-06,
+      "loss": 0.4324,
+      "step": 1654
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.891142663962136e-06,
+      "loss": 0.3802,
+      "step": 1655
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.884381338742395e-06,
+      "loss": 1.2744,
+      "step": 1656
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.877620013522652e-06,
+      "loss": 0.7094,
+      "step": 1657
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.870858688302908e-06,
+      "loss": 0.7318,
+      "step": 1658
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.864097363083165e-06,
+      "loss": 0.3725,
+      "step": 1659
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.857336037863422e-06,
+      "loss": 0.7365,
+      "step": 1660
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.85057471264368e-06,
+      "loss": 0.2596,
+      "step": 1661
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.843813387423936e-06,
+      "loss": 0.6272,
+      "step": 1662
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.837052062204193e-06,
+      "loss": 0.9933,
+      "step": 1663
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.83029073698445e-06,
+      "loss": 0.2309,
+      "step": 1664
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.2123,
+      "step": 1665
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.816768086544964e-06,
+      "loss": 1.2099,
+      "step": 1666
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.81000676132522e-06,
+      "loss": 0.6377,
+      "step": 1667
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.803245436105478e-06,
+      "loss": 0.8327,
+      "step": 1668
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.796484110885734e-06,
+      "loss": 0.2864,
+      "step": 1669
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.789722785665991e-06,
+      "loss": 0.2858,
+      "step": 1670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.782961460446248e-06,
+      "loss": 0.2375,
+      "step": 1671
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.776200135226505e-06,
+      "loss": 0.2461,
+      "step": 1672
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.769438810006762e-06,
+      "loss": 0.1762,
+      "step": 1673
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.762677484787019e-06,
+      "loss": 0.6748,
+      "step": 1674
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.755916159567276e-06,
+      "loss": 0.3271,
+      "step": 1675
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.749154834347533e-06,
+      "loss": 1.3975,
+      "step": 1676
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.74239350912779e-06,
+      "loss": 1.0924,
+      "step": 1677
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.735632183908047e-06,
+      "loss": 1.0314,
+      "step": 1678
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.728870858688304e-06,
+      "loss": 0.6573,
+      "step": 1679
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.72210953346856e-06,
+      "loss": 0.666,
+      "step": 1680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.715348208248817e-06,
+      "loss": 0.453,
+      "step": 1681
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.708586883029074e-06,
+      "loss": 0.4384,
+      "step": 1682
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.701825557809331e-06,
+      "loss": 0.2657,
+      "step": 1683
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.695064232589588e-06,
+      "loss": 1.3774,
+      "step": 1684
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.688302907369845e-06,
+      "loss": 1.4619,
+      "step": 1685
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.681541582150102e-06,
+      "loss": 0.672,
+      "step": 1686
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.674780256930359e-06,
+      "loss": 0.3201,
+      "step": 1687
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.668018931710616e-06,
+      "loss": 1.3035,
+      "step": 1688
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.661257606490873e-06,
+      "loss": 0.6873,
+      "step": 1689
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.65449628127113e-06,
+      "loss": 0.8285,
+      "step": 1690
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.647734956051387e-06,
+      "loss": 1.1522,
+      "step": 1691
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.640973630831643e-06,
+      "loss": 0.4597,
+      "step": 1692
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.6342123056119e-06,
+      "loss": 0.2859,
+      "step": 1693
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.627450980392157e-06,
+      "loss": 0.4083,
+      "step": 1694
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.620689655172414e-06,
+      "loss": 0.9214,
+      "step": 1695
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.613928329952671e-06,
+      "loss": 1.4438,
+      "step": 1696
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.607167004732928e-06,
+      "loss": 0.495,
+      "step": 1697
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.600405679513185e-06,
+      "loss": 0.9573,
+      "step": 1698
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.593644354293442e-06,
+      "loss": 0.7962,
+      "step": 1699
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.586883029073699e-06,
+      "loss": 1.0386,
+      "step": 1700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.580121703853956e-06,
+      "loss": 0.6111,
+      "step": 1701
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.573360378634214e-06,
+      "loss": 0.402,
+      "step": 1702
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.56659905341447e-06,
+      "loss": 0.5425,
+      "step": 1703
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.559837728194726e-06,
+      "loss": 0.3238,
+      "step": 1704
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.553076402974983e-06,
+      "loss": 0.7394,
+      "step": 1705
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.54631507775524e-06,
+      "loss": 0.8659,
+      "step": 1706
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.539553752535497e-06,
+      "loss": 0.6707,
+      "step": 1707
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.532792427315754e-06,
+      "loss": 1.568,
+      "step": 1708
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.526031102096013e-06,
+      "loss": 0.4711,
+      "step": 1709
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.51926977687627e-06,
+      "loss": 0.5563,
+      "step": 1710
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.512508451656525e-06,
+      "loss": 0.1869,
+      "step": 1711
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.505747126436782e-06,
+      "loss": 0.2924,
+      "step": 1712
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.498985801217039e-06,
+      "loss": 0.8093,
+      "step": 1713
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.492224475997295e-06,
+      "loss": 0.6859,
+      "step": 1714
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.485463150777552e-06,
+      "loss": 0.2996,
+      "step": 1715
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.478701825557811e-06,
+      "loss": 0.1523,
+      "step": 1716
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.471940500338068e-06,
+      "loss": 0.4322,
+      "step": 1717
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.465179175118325e-06,
+      "loss": 0.8751,
+      "step": 1718
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.45841784989858e-06,
+      "loss": 0.2859,
+      "step": 1719
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.451656524678837e-06,
+      "loss": 1.1722,
+      "step": 1720
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.444895199459094e-06,
+      "loss": 0.204,
+      "step": 1721
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.43813387423935e-06,
+      "loss": 2.7082,
+      "step": 1722
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.43137254901961e-06,
+      "loss": 0.3239,
+      "step": 1723
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.424611223799866e-06,
+      "loss": 0.8542,
+      "step": 1724
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.417849898580123e-06,
+      "loss": 0.5842,
+      "step": 1725
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.41108857336038e-06,
+      "loss": 0.8551,
+      "step": 1726
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.404327248140635e-06,
+      "loss": 1.1223,
+      "step": 1727
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.397565922920892e-06,
+      "loss": 0.2063,
+      "step": 1728
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.390804597701149e-06,
+      "loss": 0.4795,
+      "step": 1729
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.384043272481408e-06,
+      "loss": 0.9881,
+      "step": 1730
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.377281947261665e-06,
+      "loss": 0.6295,
+      "step": 1731
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.370520622041922e-06,
+      "loss": 0.7417,
+      "step": 1732
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.363759296822178e-06,
+      "loss": 1.1187,
+      "step": 1733
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.356997971602435e-06,
+      "loss": 1.0599,
+      "step": 1734
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.35023664638269e-06,
+      "loss": 0.8072,
+      "step": 1735
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.343475321162948e-06,
+      "loss": 0.3023,
+      "step": 1736
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.336713995943206e-06,
+      "loss": 0.7848,
+      "step": 1737
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.329952670723463e-06,
+      "loss": 0.9126,
+      "step": 1738
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.32319134550372e-06,
+      "loss": 0.4583,
+      "step": 1739
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.316430020283977e-06,
+      "loss": 0.2901,
+      "step": 1740
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.309668695064234e-06,
+      "loss": 0.8706,
+      "step": 1741
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 8.30290736984449e-06,
+      "loss": 0.5399,
+      "step": 1742
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.296146044624746e-06,
+      "loss": 0.3923,
+      "step": 1743
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.289384719405005e-06,
+      "loss": 0.4079,
+      "step": 1744
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.282623394185261e-06,
+      "loss": 0.3705,
+      "step": 1745
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.275862068965518e-06,
+      "loss": 0.8559,
+      "step": 1746
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.269100743745775e-06,
+      "loss": 0.967,
+      "step": 1747
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.262339418526032e-06,
+      "loss": 0.7779,
+      "step": 1748
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.255578093306289e-06,
+      "loss": 0.2946,
+      "step": 1749
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.248816768086546e-06,
+      "loss": 4.1767,
+      "step": 1750
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.242055442866801e-06,
+      "loss": 0.6053,
+      "step": 1751
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.9661,
+      "step": 1752
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.228532792427317e-06,
+      "loss": 0.6559,
+      "step": 1753
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.221771467207574e-06,
+      "loss": 0.9873,
+      "step": 1754
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.21501014198783e-06,
+      "loss": 0.9971,
+      "step": 1755
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.208248816768087e-06,
+      "loss": 1.047,
+      "step": 1756
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.201487491548344e-06,
+      "loss": 1.0186,
+      "step": 1757
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.194726166328601e-06,
+      "loss": 0.1316,
+      "step": 1758
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.187964841108858e-06,
+      "loss": 0.7791,
+      "step": 1759
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.181203515889115e-06,
+      "loss": 0.5622,
+      "step": 1760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.174442190669372e-06,
+      "loss": 0.4689,
+      "step": 1761
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.167680865449629e-06,
+      "loss": 0.4288,
+      "step": 1762
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.160919540229886e-06,
+      "loss": 0.5401,
+      "step": 1763
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.154158215010143e-06,
+      "loss": 0.5634,
+      "step": 1764
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.1473968897904e-06,
+      "loss": 1.0383,
+      "step": 1765
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.140635564570657e-06,
+      "loss": 1.001,
+      "step": 1766
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.133874239350913e-06,
+      "loss": 0.3271,
+      "step": 1767
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.12711291413117e-06,
+      "loss": 0.4179,
+      "step": 1768
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.120351588911427e-06,
+      "loss": 0.3213,
+      "step": 1769
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.113590263691684e-06,
+      "loss": 0.3562,
+      "step": 1770
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.106828938471941e-06,
+      "loss": 0.1582,
+      "step": 1771
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.100067613252198e-06,
+      "loss": 1.4764,
+      "step": 1772
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.093306288032455e-06,
+      "loss": 0.6034,
+      "step": 1773
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.086544962812712e-06,
+      "loss": 0.7866,
+      "step": 1774
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.079783637592969e-06,
+      "loss": 0.1967,
+      "step": 1775
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.073022312373226e-06,
+      "loss": 0.2682,
+      "step": 1776
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.066260987153483e-06,
+      "loss": 0.8323,
+      "step": 1777
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.05949966193374e-06,
+      "loss": 0.592,
+      "step": 1778
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.052738336713996e-06,
+      "loss": 0.4011,
+      "step": 1779
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.045977011494253e-06,
+      "loss": 0.2064,
+      "step": 1780
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.03921568627451e-06,
+      "loss": 0.6434,
+      "step": 1781
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.032454361054767e-06,
+      "loss": 1.1192,
+      "step": 1782
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.025693035835024e-06,
+      "loss": 0.3932,
+      "step": 1783
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.018931710615281e-06,
+      "loss": 0.6639,
+      "step": 1784
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.012170385395538e-06,
+      "loss": 1.2056,
+      "step": 1785
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.005409060175795e-06,
+      "loss": 0.2578,
+      "step": 1786
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.998647734956052e-06,
+      "loss": 0.3453,
+      "step": 1787
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.991886409736309e-06,
+      "loss": 0.9464,
+      "step": 1788
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.985125084516566e-06,
+      "loss": 0.2135,
+      "step": 1789
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.978363759296822e-06,
+      "loss": 0.5935,
+      "step": 1790
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.97160243407708e-06,
+      "loss": 0.4211,
+      "step": 1791
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.964841108857336e-06,
+      "loss": 0.4109,
+      "step": 1792
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.958079783637593e-06,
+      "loss": 0.8017,
+      "step": 1793
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.95131845841785e-06,
+      "loss": 0.1625,
+      "step": 1794
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.944557133198107e-06,
+      "loss": 0.3601,
+      "step": 1795
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.937795807978364e-06,
+      "loss": 0.7947,
+      "step": 1796
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.93103448275862e-06,
+      "loss": 0.9902,
+      "step": 1797
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.92427315753888e-06,
+      "loss": 0.8061,
+      "step": 1798
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.917511832319135e-06,
+      "loss": 0.6717,
+      "step": 1799
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.910750507099392e-06,
+      "loss": 1.092,
+      "step": 1800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.903989181879648e-06,
+      "loss": 1.2554,
+      "step": 1801
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.897227856659905e-06,
+      "loss": 0.6051,
+      "step": 1802
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.890466531440162e-06,
+      "loss": 0.6749,
+      "step": 1803
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.88370520622042e-06,
+      "loss": 0.6888,
+      "step": 1804
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.876943881000678e-06,
+      "loss": 2.627,
+      "step": 1805
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.870182555780935e-06,
+      "loss": 1.1305,
+      "step": 1806
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.86342123056119e-06,
+      "loss": 0.9826,
+      "step": 1807
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.856659905341447e-06,
+      "loss": 0.5667,
+      "step": 1808
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.849898580121704e-06,
+      "loss": 0.5727,
+      "step": 1809
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 0.4516,
+      "step": 1810
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.836375929682218e-06,
+      "loss": 0.7795,
+      "step": 1811
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.829614604462476e-06,
+      "loss": 1.0251,
+      "step": 1812
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.822853279242733e-06,
+      "loss": 0.7304,
+      "step": 1813
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.81609195402299e-06,
+      "loss": 0.6727,
+      "step": 1814
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.809330628803245e-06,
+      "loss": 0.4183,
+      "step": 1815
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.802569303583502e-06,
+      "loss": 0.4996,
+      "step": 1816
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.795807978363759e-06,
+      "loss": 0.8276,
+      "step": 1817
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.789046653144016e-06,
+      "loss": 0.9045,
+      "step": 1818
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.782285327924275e-06,
+      "loss": 0.6917,
+      "step": 1819
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.775524002704531e-06,
+      "loss": 1.1951,
+      "step": 1820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.768762677484788e-06,
+      "loss": 0.8411,
+      "step": 1821
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.762001352265045e-06,
+      "loss": 1.0657,
+      "step": 1822
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.7552400270453e-06,
+      "loss": 0.8639,
+      "step": 1823
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.748478701825557e-06,
+      "loss": 0.2591,
+      "step": 1824
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.741717376605814e-06,
+      "loss": 0.3652,
+      "step": 1825
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.734956051386073e-06,
+      "loss": 0.2678,
+      "step": 1826
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.72819472616633e-06,
+      "loss": 0.4208,
+      "step": 1827
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.721433400946587e-06,
+      "loss": 0.6246,
+      "step": 1828
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.714672075726844e-06,
+      "loss": 0.418,
+      "step": 1829
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.7079107505071e-06,
+      "loss": 1.9015,
+      "step": 1830
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.701149425287356e-06,
+      "loss": 0.6286,
+      "step": 1831
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.694388100067613e-06,
+      "loss": 0.9065,
+      "step": 1832
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.687626774847871e-06,
+      "loss": 0.7429,
+      "step": 1833
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.680865449628128e-06,
+      "loss": 0.3807,
+      "step": 1834
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.674104124408385e-06,
+      "loss": 0.4922,
+      "step": 1835
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.667342799188642e-06,
+      "loss": 0.3355,
+      "step": 1836
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.660581473968899e-06,
+      "loss": 0.6257,
+      "step": 1837
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.653820148749156e-06,
+      "loss": 0.6463,
+      "step": 1838
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.647058823529411e-06,
+      "loss": 0.9817,
+      "step": 1839
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.64029749830967e-06,
+      "loss": 1.1543,
+      "step": 1840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.633536173089927e-06,
+      "loss": 1.0571,
+      "step": 1841
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.626774847870183e-06,
+      "loss": 1.5587,
+      "step": 1842
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.6200135226504404e-06,
+      "loss": 0.604,
+      "step": 1843
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.613252197430697e-06,
+      "loss": 1.6693,
+      "step": 1844
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.606490872210954e-06,
+      "loss": 1.0671,
+      "step": 1845
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.599729546991211e-06,
+      "loss": 0.4081,
+      "step": 1846
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.592968221771467e-06,
+      "loss": 0.4917,
+      "step": 1847
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.586206896551724e-06,
+      "loss": 0.2226,
+      "step": 1848
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.579445571331981e-06,
+      "loss": 0.4951,
+      "step": 1849
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.572684246112239e-06,
+      "loss": 1.1385,
+      "step": 1850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.565922920892496e-06,
+      "loss": 0.6592,
+      "step": 1851
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.559161595672753e-06,
+      "loss": 0.4883,
+      "step": 1852
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.5524002704530095e-06,
+      "loss": 0.5125,
+      "step": 1853
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.5456389452332665e-06,
+      "loss": 1.3376,
+      "step": 1854
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.5388776200135225e-06,
+      "loss": 0.7388,
+      "step": 1855
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.5321162947937794e-06,
+      "loss": 0.8396,
+      "step": 1856
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.525354969574037e-06,
+      "loss": 0.9252,
+      "step": 1857
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.518593644354294e-06,
+      "loss": 0.5133,
+      "step": 1858
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.511832319134551e-06,
+      "loss": 0.3121,
+      "step": 1859
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.505070993914808e-06,
+      "loss": 0.9429,
+      "step": 1860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.498309668695065e-06,
+      "loss": 0.6513,
+      "step": 1861
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.491548343475323e-06,
+      "loss": 0.6698,
+      "step": 1862
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.484787018255578e-06,
+      "loss": 0.5433,
+      "step": 1863
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.4780256930358356e-06,
+      "loss": 0.3354,
+      "step": 1864
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.4712643678160925e-06,
+      "loss": 0.2195,
+      "step": 1865
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.464503042596349e-06,
+      "loss": 0.4876,
+      "step": 1866
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.457741717376606e-06,
+      "loss": 0.8261,
+      "step": 1867
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.450980392156863e-06,
+      "loss": 0.4057,
+      "step": 1868
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.444219066937121e-06,
+      "loss": 0.0896,
+      "step": 1869
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.437457741717378e-06,
+      "loss": 1.1603,
+      "step": 1870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.430696416497634e-06,
+      "loss": 1.1611,
+      "step": 1871
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.423935091277891e-06,
+      "loss": 0.409,
+      "step": 1872
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.417173766058148e-06,
+      "loss": 1.3407,
+      "step": 1873
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.410412440838405e-06,
+      "loss": 0.6318,
+      "step": 1874
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.403651115618662e-06,
+      "loss": 0.7599,
+      "step": 1875
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.396889790398919e-06,
+      "loss": 0.3272,
+      "step": 1876
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.390128465179176e-06,
+      "loss": 0.2183,
+      "step": 1877
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.383367139959433e-06,
+      "loss": 0.7601,
+      "step": 1878
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.376605814739689e-06,
+      "loss": 0.9864,
+      "step": 1879
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.369844489519946e-06,
+      "loss": 0.8018,
+      "step": 1880
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.363083164300203e-06,
+      "loss": 0.9069,
+      "step": 1881
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.35632183908046e-06,
+      "loss": 0.4485,
+      "step": 1882
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.349560513860718e-06,
+      "loss": 0.2167,
+      "step": 1883
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.342799188640975e-06,
+      "loss": 1.0602,
+      "step": 1884
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.3360378634212316e-06,
+      "loss": 0.8253,
+      "step": 1885
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.3292765382014885e-06,
+      "loss": 0.4562,
+      "step": 1886
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.3225152129817445e-06,
+      "loss": 0.3885,
+      "step": 1887
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.3157538877620015e-06,
+      "loss": 0.5549,
+      "step": 1888
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.308992562542258e-06,
+      "loss": 0.4858,
+      "step": 1889
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.302231237322516e-06,
+      "loss": 0.4947,
+      "step": 1890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.295469912102773e-06,
+      "loss": 0.5256,
+      "step": 1891
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.28870858688303e-06,
+      "loss": 0.7135,
+      "step": 1892
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.281947261663287e-06,
+      "loss": 0.7841,
+      "step": 1893
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.275185936443544e-06,
+      "loss": 1.2975,
+      "step": 1894
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.2684246112238e-06,
+      "loss": 0.3322,
+      "step": 1895
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.261663286004057e-06,
+      "loss": 0.7039,
+      "step": 1896
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.2549019607843145e-06,
+      "loss": 0.5227,
+      "step": 1897
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.2549019607843145e-06,
+      "loss": 2.2341,
+      "step": 1898
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.248140635564571e-06,
+      "loss": 0.8716,
+      "step": 1899
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.241379310344828e-06,
+      "loss": 0.204,
+      "step": 1900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.234617985125085e-06,
+      "loss": 0.6823,
+      "step": 1901
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.227856659905342e-06,
+      "loss": 0.401,
+      "step": 1902
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.221095334685599e-06,
+      "loss": 7.7008,
+      "step": 1903
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.214334009465855e-06,
+      "loss": 0.4368,
+      "step": 1904
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.207572684246112e-06,
+      "loss": 0.5458,
+      "step": 1905
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.20081135902637e-06,
+      "loss": 0.1697,
+      "step": 1906
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.194050033806627e-06,
+      "loss": 0.5864,
+      "step": 1907
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.187288708586884e-06,
+      "loss": 0.6223,
+      "step": 1908
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.1805273833671405e-06,
+      "loss": 0.943,
+      "step": 1909
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.1737660581473974e-06,
+      "loss": 0.5715,
+      "step": 1910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.167004732927655e-06,
+      "loss": 0.3969,
+      "step": 1911
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.16024340770791e-06,
+      "loss": 0.1373,
+      "step": 1912
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.153482082488168e-06,
+      "loss": 0.2243,
+      "step": 1913
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.146720757268425e-06,
+      "loss": 0.5686,
+      "step": 1914
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.139959432048682e-06,
+      "loss": 0.1047,
+      "step": 1915
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.133198106828939e-06,
+      "loss": 0.7205,
+      "step": 1916
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.126436781609196e-06,
+      "loss": 3.7923,
+      "step": 1917
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.119675456389454e-06,
+      "loss": 0.6724,
+      "step": 1918
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.1129141311697105e-06,
+      "loss": 1.1049,
+      "step": 1919
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.1061528059499666e-06,
+      "loss": 0.5241,
+      "step": 1920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.0993914807302235e-06,
+      "loss": 0.5039,
+      "step": 1921
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.09263015551048e-06,
+      "loss": 0.6941,
+      "step": 1922
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.085868830290737e-06,
+      "loss": 0.7946,
+      "step": 1923
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.079107505070994e-06,
+      "loss": 0.1608,
+      "step": 1924
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.072346179851252e-06,
+      "loss": 0.2414,
+      "step": 1925
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.065584854631509e-06,
+      "loss": 0.269,
+      "step": 1926
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 1.0744,
+      "step": 1927
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.052062204192022e-06,
+      "loss": 0.4935,
+      "step": 1928
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.045300878972279e-06,
+      "loss": 0.6893,
+      "step": 1929
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.038539553752536e-06,
+      "loss": 0.5312,
+      "step": 1930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.031778228532793e-06,
+      "loss": 0.1989,
+      "step": 1931
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.02501690331305e-06,
+      "loss": 0.446,
+      "step": 1932
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.018255578093307e-06,
+      "loss": 0.2938,
+      "step": 1933
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.011494252873564e-06,
+      "loss": 0.247,
+      "step": 1934
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.004732927653821e-06,
+      "loss": 0.5267,
+      "step": 1935
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.997971602434077e-06,
+      "loss": 0.2244,
+      "step": 1936
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.991210277214334e-06,
+      "loss": 0.2423,
+      "step": 1937
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.984448951994591e-06,
+      "loss": 0.2578,
+      "step": 1938
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.977687626774849e-06,
+      "loss": 1.3694,
+      "step": 1939
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.970926301555106e-06,
+      "loss": 0.2999,
+      "step": 1940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.9641649763353625e-06,
+      "loss": 0.4406,
+      "step": 1941
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.9574036511156195e-06,
+      "loss": 0.4703,
+      "step": 1942
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.950642325895876e-06,
+      "loss": 0.3377,
+      "step": 1943
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.9438810006761324e-06,
+      "loss": 0.5063,
+      "step": 1944
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.937119675456389e-06,
+      "loss": 0.7548,
+      "step": 1945
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.930358350236647e-06,
+      "loss": 0.7429,
+      "step": 1946
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.923597025016904e-06,
+      "loss": 0.9771,
+      "step": 1947
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.916835699797161e-06,
+      "loss": 0.5546,
+      "step": 1948
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.910074374577418e-06,
+      "loss": 0.8239,
+      "step": 1949
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.903313049357675e-06,
+      "loss": 0.296,
+      "step": 1950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.896551724137932e-06,
+      "loss": 0.3048,
+      "step": 1951
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.889790398918188e-06,
+      "loss": 0.4553,
+      "step": 1952
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.8830290736984455e-06,
+      "loss": 0.925,
+      "step": 1953
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.876267748478702e-06,
+      "loss": 1.073,
+      "step": 1954
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.869506423258959e-06,
+      "loss": 0.254,
+      "step": 1955
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.862745098039216e-06,
+      "loss": 0.4591,
+      "step": 1956
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.855983772819473e-06,
+      "loss": 0.8786,
+      "step": 1957
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.84922244759973e-06,
+      "loss": 0.8155,
+      "step": 1958
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.842461122379988e-06,
+      "loss": 0.6435,
+      "step": 1959
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.835699797160244e-06,
+      "loss": 1.0479,
+      "step": 1960
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.828938471940501e-06,
+      "loss": 0.4768,
+      "step": 1961
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.822177146720758e-06,
+      "loss": 0.3764,
+      "step": 1962
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.815415821501015e-06,
+      "loss": 0.2468,
+      "step": 1963
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.8086544962812715e-06,
+      "loss": 0.4674,
+      "step": 1964
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.801893171061528e-06,
+      "loss": 0.3185,
+      "step": 1965
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.795131845841786e-06,
+      "loss": 0.7363,
+      "step": 1966
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.788370520622043e-06,
+      "loss": 0.1495,
+      "step": 1967
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.781609195402299e-06,
+      "loss": 1.3347,
+      "step": 1968
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.774847870182556e-06,
+      "loss": 0.3922,
+      "step": 1969
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.768086544962813e-06,
+      "loss": 0.3257,
+      "step": 1970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.76132521974307e-06,
+      "loss": 0.393,
+      "step": 1971
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.754563894523327e-06,
+      "loss": 0.4779,
+      "step": 1972
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.7478025693035846e-06,
+      "loss": 0.1888,
+      "step": 1973
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.7410412440838415e-06,
+      "loss": 0.6366,
+      "step": 1974
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.734279918864098e-06,
+      "loss": 1.116,
+      "step": 1975
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.7275185936443544e-06,
+      "loss": 0.3943,
+      "step": 1976
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.720757268424611e-06,
+      "loss": 0.2254,
+      "step": 1977
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.713995943204868e-06,
+      "loss": 1.1666,
+      "step": 1978
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.707234617985125e-06,
+      "loss": 0.0,
+      "step": 1979
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.700473292765383e-06,
+      "loss": 0.7887,
+      "step": 1980
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.69371196754564e-06,
+      "loss": 0.2802,
+      "step": 1981
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.686950642325897e-06,
+      "loss": 0.8082,
+      "step": 1982
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.680189317106154e-06,
+      "loss": 0.3272,
+      "step": 1983
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.67342799188641e-06,
+      "loss": 0.1945,
+      "step": 1984
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.8212,
+      "step": 1985
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.6599053414469236e-06,
+      "loss": 1.1065,
+      "step": 1986
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.653144016227181e-06,
+      "loss": 0.1671,
+      "step": 1987
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.646382691007438e-06,
+      "loss": 1.2554,
+      "step": 1988
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.639621365787695e-06,
+      "loss": 0.5639,
+      "step": 1989
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.632860040567952e-06,
+      "loss": 0.9741,
+      "step": 1990
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.626098715348209e-06,
+      "loss": 0.6914,
+      "step": 1991
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.619337390128465e-06,
+      "loss": 0.5048,
+      "step": 1992
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.612576064908722e-06,
+      "loss": 1.4156,
+      "step": 1993
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.60581473968898e-06,
+      "loss": 0.1008,
+      "step": 1994
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.599053414469237e-06,
+      "loss": 0.4516,
+      "step": 1995
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.5922920892494935e-06,
+      "loss": 0.3102,
+      "step": 1996
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.5855307640297504e-06,
+      "loss": 1.3474,
+      "step": 1997
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.578769438810007e-06,
+      "loss": 0.4541,
+      "step": 1998
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.572008113590265e-06,
+      "loss": 0.319,
+      "step": 1999
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.56524678837052e-06,
+      "loss": 0.8174,
+      "step": 2000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.558485463150778e-06,
+      "loss": 0.3817,
+      "step": 2001
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.551724137931035e-06,
+      "loss": 0.5354,
+      "step": 2002
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.544962812711292e-06,
+      "loss": 1.5903,
+      "step": 2003
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.538201487491549e-06,
+      "loss": 1.2723,
+      "step": 2004
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.531440162271806e-06,
+      "loss": 2.0227,
+      "step": 2005
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.5246788370520635e-06,
+      "loss": 0.4595,
+      "step": 2006
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.51791751183232e-06,
+      "loss": 0.5452,
+      "step": 2007
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.5111561866125765e-06,
+      "loss": 0.5973,
+      "step": 2008
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.504394861392833e-06,
+      "loss": 0.3705,
+      "step": 2009
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.49763353617309e-06,
+      "loss": 0.4812,
+      "step": 2010
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.490872210953347e-06,
+      "loss": 0.4486,
+      "step": 2011
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.484110885733604e-06,
+      "loss": 0.5134,
+      "step": 2012
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.477349560513861e-06,
+      "loss": 0.1695,
+      "step": 2013
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 0.372,
+      "step": 2014
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.463826910074376e-06,
+      "loss": 0.3014,
+      "step": 2015
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.457065584854632e-06,
+      "loss": 0.194,
+      "step": 2016
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.450304259634889e-06,
+      "loss": 0.2012,
+      "step": 2017
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.4435429344151456e-06,
+      "loss": 0.3968,
+      "step": 2018
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.4367816091954025e-06,
+      "loss": 0.5171,
+      "step": 2019
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.430020283975659e-06,
+      "loss": 0.4068,
+      "step": 2020
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.423258958755917e-06,
+      "loss": 0.7277,
+      "step": 2021
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.416497633536174e-06,
+      "loss": 0.712,
+      "step": 2022
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.409736308316431e-06,
+      "loss": 0.1488,
+      "step": 2023
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.402974983096687e-06,
+      "loss": 1.3497,
+      "step": 2024
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.396213657876944e-06,
+      "loss": 1.2563,
+      "step": 2025
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.389452332657201e-06,
+      "loss": 0.5364,
+      "step": 2026
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.382691007437458e-06,
+      "loss": 0.4087,
+      "step": 2027
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.3759296822177155e-06,
+      "loss": 0.5142,
+      "step": 2028
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.3691683569979724e-06,
+      "loss": 0.5401,
+      "step": 2029
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.362407031778229e-06,
+      "loss": 1.5168,
+      "step": 2030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.355645706558486e-06,
+      "loss": 0.6052,
+      "step": 2031
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.348884381338742e-06,
+      "loss": 0.539,
+      "step": 2032
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.342123056118999e-06,
+      "loss": 0.5163,
+      "step": 2033
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.335361730899256e-06,
+      "loss": 0.4933,
+      "step": 2034
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.328600405679514e-06,
+      "loss": 0.3566,
+      "step": 2035
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.321839080459771e-06,
+      "loss": 0.5381,
+      "step": 2036
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.315077755240028e-06,
+      "loss": 0.4789,
+      "step": 2037
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.308316430020285e-06,
+      "loss": 0.5051,
+      "step": 2038
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.3015551048005416e-06,
+      "loss": 0.1448,
+      "step": 2039
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.294793779580798e-06,
+      "loss": 0.1457,
+      "step": 2040
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.2880324543610545e-06,
+      "loss": 0.5169,
+      "step": 2041
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.281271129141312e-06,
+      "loss": 0.1496,
+      "step": 2042
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.274509803921569e-06,
+      "loss": 0.3345,
+      "step": 2043
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.267748478701826e-06,
+      "loss": 0.2372,
+      "step": 2044
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.260987153482083e-06,
+      "loss": 0.2955,
+      "step": 2045
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.25422582826234e-06,
+      "loss": 0.9627,
+      "step": 2046
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.247464503042598e-06,
+      "loss": 0.0926,
+      "step": 2047
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.240703177822853e-06,
+      "loss": 0.271,
+      "step": 2048
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.233941852603111e-06,
+      "loss": 0.2322,
+      "step": 2049
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.227180527383368e-06,
+      "loss": 0.7502,
+      "step": 2050
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.2204192021636245e-06,
+      "loss": 0.1156,
+      "step": 2051
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.213657876943881e-06,
+      "loss": 0.2348,
+      "step": 2052
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.206896551724138e-06,
+      "loss": 0.3944,
+      "step": 2053
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.200135226504396e-06,
+      "loss": 0.401,
+      "step": 2054
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.193373901284653e-06,
+      "loss": 0.2387,
+      "step": 2055
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.186612576064909e-06,
+      "loss": 1.543,
+      "step": 2056
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.179851250845166e-06,
+      "loss": 0.3577,
+      "step": 2057
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.173089925625423e-06,
+      "loss": 1.2939,
+      "step": 2058
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.16632860040568e-06,
+      "loss": 0.5839,
+      "step": 2059
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.159567275185937e-06,
+      "loss": 0.4358,
+      "step": 2060
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.1528059499661945e-06,
+      "loss": 0.7513,
+      "step": 2061
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.146044624746451e-06,
+      "loss": 0.3157,
+      "step": 2062
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.139283299526708e-06,
+      "loss": 0.194,
+      "step": 2063
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.132521974306964e-06,
+      "loss": 0.2253,
+      "step": 2064
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.125760649087221e-06,
+      "loss": 0.4677,
+      "step": 2065
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.118999323867478e-06,
+      "loss": 0.2239,
+      "step": 2066
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.112237998647735e-06,
+      "loss": 0.157,
+      "step": 2067
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.105476673427993e-06,
+      "loss": 0.5959,
+      "step": 2068
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.09871534820825e-06,
+      "loss": 1.2587,
+      "step": 2069
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.091954022988507e-06,
+      "loss": 0.3619,
+      "step": 2070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.0851926977687636e-06,
+      "loss": 1.1883,
+      "step": 2071
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.07843137254902e-06,
+      "loss": 0.4912,
+      "step": 2072
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.0716700473292766e-06,
+      "loss": 0.4014,
+      "step": 2073
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.0649087221095335e-06,
+      "loss": 0.3333,
+      "step": 2074
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.058147396889791e-06,
+      "loss": 0.6812,
+      "step": 2075
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.051386071670048e-06,
+      "loss": 0.1509,
+      "step": 2076
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.044624746450305e-06,
+      "loss": 1.1136,
+      "step": 2077
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.037863421230562e-06,
+      "loss": 0.3999,
+      "step": 2078
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.031102096010819e-06,
+      "loss": 0.2317,
+      "step": 2079
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.024340770791075e-06,
+      "loss": 0.7905,
+      "step": 2080
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.017579445571332e-06,
+      "loss": 0.0624,
+      "step": 2081
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.01081812035159e-06,
+      "loss": 0.3697,
+      "step": 2082
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.0040567951318465e-06,
+      "loss": 0.072,
+      "step": 2083
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.9972954699121034e-06,
+      "loss": 0.1521,
+      "step": 2084
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.99053414469236e-06,
+      "loss": 0.0582,
+      "step": 2085
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.983772819472617e-06,
+      "loss": 0.5336,
+      "step": 2086
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.977011494252874e-06,
+      "loss": 0.1524,
+      "step": 2087
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.97025016903313e-06,
+      "loss": 0.7106,
+      "step": 2088
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.963488843813387e-06,
+      "loss": 1.5859,
+      "step": 2089
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.956727518593645e-06,
+      "loss": 0.0997,
+      "step": 2090
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.949966193373902e-06,
+      "loss": 0.6522,
+      "step": 2091
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.943204868154159e-06,
+      "loss": 0.277,
+      "step": 2092
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.936443542934416e-06,
+      "loss": 0.8385,
+      "step": 2093
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.9296822177146725e-06,
+      "loss": 0.2037,
+      "step": 2094
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.92292089249493e-06,
+      "loss": 0.0347,
+      "step": 2095
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.9161595672751855e-06,
+      "loss": 0.4153,
+      "step": 2096
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.909398242055443e-06,
+      "loss": 0.2647,
+      "step": 2097
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.9026369168357e-06,
+      "loss": 0.1414,
+      "step": 2098
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.895875591615957e-06,
+      "loss": 0.476,
+      "step": 2099
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.889114266396214e-06,
+      "loss": 0.1477,
+      "step": 2100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 1.0273,
+      "step": 2101
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.875591615956729e-06,
+      "loss": 0.0857,
+      "step": 2102
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.868830290736986e-06,
+      "loss": 0.1313,
+      "step": 2103
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.862068965517242e-06,
+      "loss": 0.4226,
+      "step": 2104
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.8553076402974986e-06,
+      "loss": 0.409,
+      "step": 2105
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.8485463150777555e-06,
+      "loss": 0.6915,
+      "step": 2106
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.841784989858012e-06,
+      "loss": 0.3137,
+      "step": 2107
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.835023664638269e-06,
+      "loss": 0.2225,
+      "step": 2108
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.828262339418527e-06,
+      "loss": 0.3279,
+      "step": 2109
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.821501014198784e-06,
+      "loss": 0.4007,
+      "step": 2110
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.814739688979041e-06,
+      "loss": 0.6022,
+      "step": 2111
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.807978363759297e-06,
+      "loss": 0.6542,
+      "step": 2112
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.801217038539554e-06,
+      "loss": 0.3424,
+      "step": 2113
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.794455713319811e-06,
+      "loss": 0.8618,
+      "step": 2114
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.787694388100068e-06,
+      "loss": 0.3919,
+      "step": 2115
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.7809330628803254e-06,
+      "loss": 0.5303,
+      "step": 2116
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.774171737660582e-06,
+      "loss": 0.4463,
+      "step": 2117
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.767410412440839e-06,
+      "loss": 1.3051,
+      "step": 2118
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.760649087221096e-06,
+      "loss": 0.87,
+      "step": 2119
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.753887762001352e-06,
+      "loss": 0.3587,
+      "step": 2120
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.747126436781609e-06,
+      "loss": 1.2605,
+      "step": 2121
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.740365111561866e-06,
+      "loss": 0.1743,
+      "step": 2122
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.733603786342124e-06,
+      "loss": 0.4376,
+      "step": 2123
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.726842461122381e-06,
+      "loss": 0.5396,
+      "step": 2124
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.720081135902638e-06,
+      "loss": 0.5163,
+      "step": 2125
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.7133198106828946e-06,
+      "loss": 0.1816,
+      "step": 2126
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.7065584854631515e-06,
+      "loss": 0.0923,
+      "step": 2127
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6997971602434075e-06,
+      "loss": 0.8091,
+      "step": 2128
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6930358350236644e-06,
+      "loss": 0.9699,
+      "step": 2129
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.686274509803922e-06,
+      "loss": 0.4752,
+      "step": 2130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.679513184584179e-06,
+      "loss": 0.8712,
+      "step": 2131
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.672751859364436e-06,
+      "loss": 0.2345,
+      "step": 2132
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.665990534144693e-06,
+      "loss": 0.4026,
+      "step": 2133
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.65922920892495e-06,
+      "loss": 1.1986,
+      "step": 2134
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.652467883705207e-06,
+      "loss": 0.2682,
+      "step": 2135
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.645706558485463e-06,
+      "loss": 0.3657,
+      "step": 2136
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.638945233265721e-06,
+      "loss": 0.4735,
+      "step": 2137
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.6321839080459775e-06,
+      "loss": 1.2221,
+      "step": 2138
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.625422582826234e-06,
+      "loss": 0.5978,
+      "step": 2139
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.618661257606491e-06,
+      "loss": 0.3018,
+      "step": 2140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.611899932386748e-06,
+      "loss": 0.5564,
+      "step": 2141
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.605138607167005e-06,
+      "loss": 0.1663,
+      "step": 2142
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.598377281947263e-06,
+      "loss": 0.5304,
+      "step": 2143
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.591615956727519e-06,
+      "loss": 1.1263,
+      "step": 2144
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.584854631507776e-06,
+      "loss": 0.4043,
+      "step": 2145
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.578093306288033e-06,
+      "loss": 1.4786,
+      "step": 2146
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.57133198106829e-06,
+      "loss": 0.2782,
+      "step": 2147
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.564570655848547e-06,
+      "loss": 1.2658,
+      "step": 2148
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5578093306288035e-06,
+      "loss": 0.3177,
+      "step": 2149
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.551048005409061e-06,
+      "loss": 0.5002,
+      "step": 2150
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.544286680189318e-06,
+      "loss": 0.588,
+      "step": 2151
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.537525354969574e-06,
+      "loss": 0.3172,
+      "step": 2152
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.530764029749831e-06,
+      "loss": 0.3415,
+      "step": 2153
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.524002704530088e-06,
+      "loss": 0.5266,
+      "step": 2154
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.517241379310345e-06,
+      "loss": 0.3604,
+      "step": 2155
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.510480054090602e-06,
+      "loss": 1.1845,
+      "step": 2156
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.50371872887086e-06,
+      "loss": 0.1001,
+      "step": 2157
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.4969574036511166e-06,
+      "loss": 0.1222,
+      "step": 2158
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.4901960784313735e-06,
+      "loss": 0.9819,
+      "step": 2159
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.4834347532116295e-06,
+      "loss": 1.1506,
+      "step": 2160
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.4766734279918865e-06,
+      "loss": 0.4868,
+      "step": 2161
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.469912102772143e-06,
+      "loss": 0.19,
+      "step": 2162
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.4631507775524e-06,
+      "loss": 0.4149,
+      "step": 2163
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.456389452332658e-06,
+      "loss": 0.2603,
+      "step": 2164
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.449628127112915e-06,
+      "loss": 0.3861,
+      "step": 2165
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.442866801893172e-06,
+      "loss": 0.1135,
+      "step": 2166
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.436105476673429e-06,
+      "loss": 1.0232,
+      "step": 2167
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.429344151453685e-06,
+      "loss": 1.1334,
+      "step": 2168
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.422582826233942e-06,
+      "loss": 0.1971,
+      "step": 2169
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.415821501014199e-06,
+      "loss": 0.3795,
+      "step": 2170
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.409060175794456e-06,
+      "loss": 0.3926,
+      "step": 2171
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.402298850574713e-06,
+      "loss": 0.2922,
+      "step": 2172
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.39553752535497e-06,
+      "loss": 0.5037,
+      "step": 2173
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.388776200135227e-06,
+      "loss": 0.389,
+      "step": 2174
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.382014874915484e-06,
+      "loss": 0.3572,
+      "step": 2175
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.37525354969574e-06,
+      "loss": 0.47,
+      "step": 2176
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.368492224475997e-06,
+      "loss": 0.9512,
+      "step": 2177
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.361730899256255e-06,
+      "loss": 0.4953,
+      "step": 2178
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.354969574036512e-06,
+      "loss": 0.3143,
+      "step": 2179
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.348208248816769e-06,
+      "loss": 0.1605,
+      "step": 2180
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.3414469235970255e-06,
+      "loss": 1.1512,
+      "step": 2181
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.3346855983772824e-06,
+      "loss": 0.4039,
+      "step": 2182
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.32792427315754e-06,
+      "loss": 0.321,
+      "step": 2183
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.321162947937795e-06,
+      "loss": 1.0315,
+      "step": 2184
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.314401622718053e-06,
+      "loss": 0.6871,
+      "step": 2185
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.30764029749831e-06,
+      "loss": 0.5243,
+      "step": 2186
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.300878972278567e-06,
+      "loss": 0.2675,
+      "step": 2187
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.3877,
+      "step": 2188
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.287356321839081e-06,
+      "loss": 0.2833,
+      "step": 2189
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.280594996619339e-06,
+      "loss": 0.5672,
+      "step": 2190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.2738336713995955e-06,
+      "loss": 0.4989,
+      "step": 2191
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.2670723461798516e-06,
+      "loss": 0.1748,
+      "step": 2192
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.2603110209601085e-06,
+      "loss": 0.2672,
+      "step": 2193
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.253549695740365e-06,
+      "loss": 0.182,
+      "step": 2194
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.246788370520622e-06,
+      "loss": 0.1769,
+      "step": 2195
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.240027045300879e-06,
+      "loss": 0.2543,
+      "step": 2196
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.233265720081136e-06,
+      "loss": 0.3407,
+      "step": 2197
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.226504394861394e-06,
+      "loss": 0.2521,
+      "step": 2198
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.219743069641651e-06,
+      "loss": 1.9726,
+      "step": 2199
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.212981744421907e-06,
+      "loss": 0.6016,
+      "step": 2200
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.206220419202164e-06,
+      "loss": 0.2089,
+      "step": 2201
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.199459093982421e-06,
+      "loss": 0.4687,
+      "step": 2202
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.192697768762678e-06,
+      "loss": 0.2964,
+      "step": 2203
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.1859364435429345e-06,
+      "loss": 0.053,
+      "step": 2204
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.179175118323192e-06,
+      "loss": 0.3938,
+      "step": 2205
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.172413793103449e-06,
+      "loss": 0.6534,
+      "step": 2206
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.165652467883706e-06,
+      "loss": 0.5256,
+      "step": 2207
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.158891142663962e-06,
+      "loss": 0.6876,
+      "step": 2208
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.152129817444219e-06,
+      "loss": 0.8966,
+      "step": 2209
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.145368492224476e-06,
+      "loss": 0.3756,
+      "step": 2210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.138607167004733e-06,
+      "loss": 0.6095,
+      "step": 2211
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.131845841784991e-06,
+      "loss": 1.1057,
+      "step": 2212
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.1250845165652475e-06,
+      "loss": 0.7235,
+      "step": 2213
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.1183231913455045e-06,
+      "loss": 0.391,
+      "step": 2214
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.111561866125761e-06,
+      "loss": 0.3359,
+      "step": 2215
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.1048005409060174e-06,
+      "loss": 0.264,
+      "step": 2216
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.098039215686274e-06,
+      "loss": 0.2618,
+      "step": 2217
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.091277890466531e-06,
+      "loss": 0.2835,
+      "step": 2218
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.084516565246789e-06,
+      "loss": 0.2062,
+      "step": 2219
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.077755240027046e-06,
+      "loss": 0.7516,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.070993914807303e-06,
+      "loss": 0.2357,
+      "step": 2221
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.06423258958756e-06,
+      "loss": 0.115,
+      "step": 2222
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.057471264367817e-06,
+      "loss": 0.0676,
+      "step": 2223
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.050709939148073e-06,
+      "loss": 0.2036,
+      "step": 2224
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.04394861392833e-06,
+      "loss": 0.1653,
+      "step": 2225
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.037187288708587e-06,
+      "loss": 0.1362,
+      "step": 2226
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.030425963488844e-06,
+      "loss": 0.2119,
+      "step": 2227
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.023664638269101e-06,
+      "loss": 0.5783,
+      "step": 2228
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.016903313049358e-06,
+      "loss": 0.2476,
+      "step": 2229
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.010141987829615e-06,
+      "loss": 0.3121,
+      "step": 2230
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.003380662609873e-06,
+      "loss": 0.1611,
+      "step": 2231
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.996619337390129e-06,
+      "loss": 0.2922,
+      "step": 2232
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.989858012170386e-06,
+      "loss": 0.5937,
+      "step": 2233
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.983096686950643e-06,
+      "loss": 0.2276,
+      "step": 2234
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.9763353617309e-06,
+      "loss": 0.6098,
+      "step": 2235
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.9695740365111565e-06,
+      "loss": 0.2346,
+      "step": 2236
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.962812711291413e-06,
+      "loss": 0.421,
+      "step": 2237
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.95605138607167e-06,
+      "loss": 0.6877,
+      "step": 2238
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.949290060851927e-06,
+      "loss": 0.1877,
+      "step": 2239
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.942528735632184e-06,
+      "loss": 0.0733,
+      "step": 2240
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.935767410412441e-06,
+      "loss": 0.1271,
+      "step": 2241
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.929006085192698e-06,
+      "loss": 0.0843,
+      "step": 2242
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.922244759972955e-06,
+      "loss": 0.0281,
+      "step": 2243
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.915483434753212e-06,
+      "loss": 0.125,
+      "step": 2244
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.9087221095334696e-06,
+      "loss": 0.1927,
+      "step": 2245
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.901960784313726e-06,
+      "loss": 0.314,
+      "step": 2246
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8951994590939825e-06,
+      "loss": 0.0379,
+      "step": 2247
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8884381338742394e-06,
+      "loss": 0.2003,
+      "step": 2248
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.881676808654497e-06,
+      "loss": 0.2131,
+      "step": 2249
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.874915483434753e-06,
+      "loss": 0.4485,
+      "step": 2250
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.86815415821501e-06,
+      "loss": 0.4532,
+      "step": 2251
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.861392832995268e-06,
+      "loss": 0.1358,
+      "step": 2252
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.854631507775525e-06,
+      "loss": 0.2506,
+      "step": 2253
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.847870182555781e-06,
+      "loss": 0.048,
+      "step": 2254
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.841108857336038e-06,
+      "loss": 0.0643,
+      "step": 2255
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.834347532116296e-06,
+      "loss": 0.0916,
+      "step": 2256
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8275862068965525e-06,
+      "loss": 0.1416,
+      "step": 2257
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8208248816768086e-06,
+      "loss": 0.358,
+      "step": 2258
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.814063556457066e-06,
+      "loss": 0.1884,
+      "step": 2259
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.807302231237323e-06,
+      "loss": 0.3095,
+      "step": 2260
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.80054090601758e-06,
+      "loss": 0.1719,
+      "step": 2261
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.793779580797836e-06,
+      "loss": 0.1965,
+      "step": 2262
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.787018255578094e-06,
+      "loss": 0.1603,
+      "step": 2263
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.780256930358351e-06,
+      "loss": 0.2732,
+      "step": 2264
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.773495605138608e-06,
+      "loss": 0.1921,
+      "step": 2265
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.766734279918865e-06,
+      "loss": 0.1847,
+      "step": 2266
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.759972954699122e-06,
+      "loss": 0.0828,
+      "step": 2267
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.7532116294793785e-06,
+      "loss": 0.2426,
+      "step": 2268
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.7464503042596354e-06,
+      "loss": 0.224,
+      "step": 2269
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.739688979039892e-06,
+      "loss": 0.371,
+      "step": 2270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.732927653820149e-06,
+      "loss": 0.2727,
+      "step": 2271
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.726166328600406e-06,
+      "loss": 0.613,
+      "step": 2272
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.719405003380663e-06,
+      "loss": 0.0794,
+      "step": 2273
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.71264367816092e-06,
+      "loss": 0.2396,
+      "step": 2274
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.3027,
+      "step": 2275
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.699121027721434e-06,
+      "loss": 0.0609,
+      "step": 2276
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.692359702501691e-06,
+      "loss": 0.2401,
+      "step": 2277
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.685598377281948e-06,
+      "loss": 0.6839,
+      "step": 2278
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.6788370520622046e-06,
+      "loss": 0.1701,
+      "step": 2279
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.6720757268424615e-06,
+      "loss": 0.4448,
+      "step": 2280
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.665314401622718e-06,
+      "loss": 0.1406,
+      "step": 2281
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.658553076402975e-06,
+      "loss": 0.2917,
+      "step": 2282
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.651791751183232e-06,
+      "loss": 0.1498,
+      "step": 2283
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.645030425963489e-06,
+      "loss": 0.0283,
+      "step": 2284
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.638269100743746e-06,
+      "loss": 0.3051,
+      "step": 2285
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.631507775524003e-06,
+      "loss": 0.2308,
+      "step": 2286
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.62474645030426e-06,
+      "loss": 0.2972,
+      "step": 2287
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.617985125084517e-06,
+      "loss": 0.2736,
+      "step": 2288
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.611223799864774e-06,
+      "loss": 0.1388,
+      "step": 2289
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.604462474645031e-06,
+      "loss": 0.1813,
+      "step": 2290
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.5977011494252875e-06,
+      "loss": 0.075,
+      "step": 2291
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.590939824205544e-06,
+      "loss": 0.3001,
+      "step": 2292
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.584178498985802e-06,
+      "loss": 0.1161,
+      "step": 2293
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.577417173766058e-06,
+      "loss": 0.217,
+      "step": 2294
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.570655848546315e-06,
+      "loss": 0.2011,
+      "step": 2295
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.563894523326572e-06,
+      "loss": 0.0883,
+      "step": 2296
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.55713319810683e-06,
+      "loss": 0.1111,
+      "step": 2297
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.550371872887086e-06,
+      "loss": 0.1828,
+      "step": 2298
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.543610547667343e-06,
+      "loss": 0.4594,
+      "step": 2299
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.5368492224476005e-06,
+      "loss": 0.4897,
+      "step": 2300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.5300878972278575e-06,
+      "loss": 0.4207,
+      "step": 2301
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.5233265720081135e-06,
+      "loss": 0.1146,
+      "step": 2302
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.5165652467883704e-06,
+      "loss": 0.0777,
+      "step": 2303
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.509803921568628e-06,
+      "loss": 0.1808,
+      "step": 2304
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.503042596348885e-06,
+      "loss": 0.1708,
+      "step": 2305
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.496281271129141e-06,
+      "loss": 0.0798,
+      "step": 2306
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.489519945909399e-06,
+      "loss": 0.2748,
+      "step": 2307
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.482758620689656e-06,
+      "loss": 0.2192,
+      "step": 2308
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.475997295469913e-06,
+      "loss": 0.0959,
+      "step": 2309
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.469235970250169e-06,
+      "loss": 0.205,
+      "step": 2310
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.4624746450304266e-06,
+      "loss": 0.7251,
+      "step": 2311
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.4557133198106835e-06,
+      "loss": 0.0756,
+      "step": 2312
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.44895199459094e-06,
+      "loss": 0.1918,
+      "step": 2313
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.442190669371197e-06,
+      "loss": 0.1583,
+      "step": 2314
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.435429344151454e-06,
+      "loss": 0.3797,
+      "step": 2315
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.428668018931711e-06,
+      "loss": 0.1989,
+      "step": 2316
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.421906693711968e-06,
+      "loss": 0.1515,
+      "step": 2317
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.415145368492225e-06,
+      "loss": 0.4145,
+      "step": 2318
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.408384043272482e-06,
+      "loss": 0.2046,
+      "step": 2319
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.401622718052739e-06,
+      "loss": 0.1874,
+      "step": 2320
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.394861392832996e-06,
+      "loss": 0.1271,
+      "step": 2321
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.388100067613253e-06,
+      "loss": 0.3626,
+      "step": 2322
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.3813387423935095e-06,
+      "loss": 0.3188,
+      "step": 2323
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.374577417173766e-06,
+      "loss": 0.1576,
+      "step": 2324
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.367816091954023e-06,
+      "loss": 0.2083,
+      "step": 2325
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.36105476673428e-06,
+      "loss": 0.4876,
+      "step": 2326
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.354293441514537e-06,
+      "loss": 0.2382,
+      "step": 2327
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.347532116294794e-06,
+      "loss": 0.1043,
+      "step": 2328
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.340770791075051e-06,
+      "loss": 0.142,
+      "step": 2329
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.334009465855308e-06,
+      "loss": 0.0585,
+      "step": 2330
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.327248140635565e-06,
+      "loss": 0.2454,
+      "step": 2331
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.320486815415822e-06,
+      "loss": 0.2267,
+      "step": 2332
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.313725490196079e-06,
+      "loss": 0.3781,
+      "step": 2333
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.3069641649763355e-06,
+      "loss": 0.3391,
+      "step": 2334
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.3002028397565924e-06,
+      "loss": 0.0454,
+      "step": 2335
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.293441514536849e-06,
+      "loss": 0.1723,
+      "step": 2336
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.286680189317107e-06,
+      "loss": 0.2538,
+      "step": 2337
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.279918864097363e-06,
+      "loss": 0.083,
+      "step": 2338
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.27315753887762e-06,
+      "loss": 0.3081,
+      "step": 2339
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.266396213657877e-06,
+      "loss": 0.1667,
+      "step": 2340
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.259634888438135e-06,
+      "loss": 0.1483,
+      "step": 2341
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.252873563218391e-06,
+      "loss": 0.0702,
+      "step": 2342
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.246112237998648e-06,
+      "loss": 0.1037,
+      "step": 2343
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.2393509127789055e-06,
+      "loss": 0.2336,
+      "step": 2344
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.232589587559162e-06,
+      "loss": 0.2844,
+      "step": 2345
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.2258282623394185e-06,
+      "loss": 0.2006,
+      "step": 2346
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.219066937119675e-06,
+      "loss": 0.2401,
+      "step": 2347
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.212305611899933e-06,
+      "loss": 0.4512,
+      "step": 2348
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.20554428668019e-06,
+      "loss": 0.1403,
+      "step": 2349
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.198782961460446e-06,
+      "loss": 0.1969,
+      "step": 2350
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.192021636240704e-06,
+      "loss": 0.0692,
+      "step": 2351
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.185260311020961e-06,
+      "loss": 0.122,
+      "step": 2352
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.178498985801218e-06,
+      "loss": 0.1468,
+      "step": 2353
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.171737660581474e-06,
+      "loss": 0.2333,
+      "step": 2354
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.1649763353617315e-06,
+      "loss": 0.1906,
+      "step": 2355
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.1582150101419884e-06,
+      "loss": 0.0695,
+      "step": 2356
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.151453684922245e-06,
+      "loss": 0.1909,
+      "step": 2357
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.144692359702502e-06,
+      "loss": 0.2834,
+      "step": 2358
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.137931034482759e-06,
+      "loss": 0.1766,
+      "step": 2359
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.131169709263016e-06,
+      "loss": 0.0426,
+      "step": 2360
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.124408384043273e-06,
+      "loss": 0.0442,
+      "step": 2361
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 0.1011,
+      "step": 2362
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.110885733603787e-06,
+      "loss": 0.1171,
+      "step": 2363
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.104124408384044e-06,
+      "loss": 0.1088,
+      "step": 2364
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.097363083164301e-06,
+      "loss": 0.1372,
+      "step": 2365
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.0906017579445575e-06,
+      "loss": 0.0926,
+      "step": 2366
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.0838404327248145e-06,
+      "loss": 0.2514,
+      "step": 2367
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.077079107505071e-06,
+      "loss": 0.8352,
+      "step": 2368
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.070317782285328e-06,
+      "loss": 0.1411,
+      "step": 2369
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.063556457065585e-06,
+      "loss": 0.1878,
+      "step": 2370
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.056795131845842e-06,
+      "loss": 0.0767,
+      "step": 2371
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.050033806626099e-06,
+      "loss": 0.2523,
+      "step": 2372
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.043272481406356e-06,
+      "loss": 0.3118,
+      "step": 2373
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.036511156186613e-06,
+      "loss": 0.0784,
+      "step": 2374
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.02974983096687e-06,
+      "loss": 0.1228,
+      "step": 2375
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.022988505747127e-06,
+      "loss": 0.3168,
+      "step": 2376
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.0162271805273836e-06,
+      "loss": 0.2071,
+      "step": 2377
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.0094658553076405e-06,
+      "loss": 0.0906,
+      "step": 2378
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.002704530087897e-06,
+      "loss": 0.5069,
+      "step": 2379
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.995943204868154e-06,
+      "loss": 0.6009,
+      "step": 2380
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.989181879648411e-06,
+      "loss": 0.4841,
+      "step": 2381
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.982420554428668e-06,
+      "loss": 0.08,
+      "step": 2382
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.975659229208925e-06,
+      "loss": 0.22,
+      "step": 2383
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.968897903989182e-06,
+      "loss": 0.1206,
+      "step": 2384
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.96213657876944e-06,
+      "loss": 0.2225,
+      "step": 2385
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.955375253549696e-06,
+      "loss": 0.2327,
+      "step": 2386
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.948613928329953e-06,
+      "loss": 0.1727,
+      "step": 2387
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.94185260311021e-06,
+      "loss": 0.252,
+      "step": 2388
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.935091277890467e-06,
+      "loss": 0.3803,
+      "step": 2389
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.928329952670723e-06,
+      "loss": 0.18,
+      "step": 2390
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 0.7951,
+      "step": 2391
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.914807302231238e-06,
+      "loss": 0.1112,
+      "step": 2392
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.908045977011495e-06,
+      "loss": 0.3329,
+      "step": 2393
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.901284651791751e-06,
+      "loss": 0.1118,
+      "step": 2394
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.894523326572008e-06,
+      "loss": 0.4398,
+      "step": 2395
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.887762001352266e-06,
+      "loss": 0.4788,
+      "step": 2396
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.881000676132523e-06,
+      "loss": 0.1853,
+      "step": 2397
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.874239350912779e-06,
+      "loss": 0.5145,
+      "step": 2398
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.8674780256930365e-06,
+      "loss": 0.1849,
+      "step": 2399
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.860716700473293e-06,
+      "loss": 0.1659,
+      "step": 2400
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.85395537525355e-06,
+      "loss": 0.2644,
+      "step": 2401
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.847194050033806e-06,
+      "loss": 0.5856,
+      "step": 2402
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.840432724814064e-06,
+      "loss": 0.0528,
+      "step": 2403
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.833671399594321e-06,
+      "loss": 0.3417,
+      "step": 2404
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.826910074374578e-06,
+      "loss": 0.1348,
+      "step": 2405
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.820148749154835e-06,
+      "loss": 0.1187,
+      "step": 2406
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.8133874239350913e-06,
+      "loss": 0.558,
+      "step": 2407
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.8066260987153487e-06,
+      "loss": 0.1865,
+      "step": 2408
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.7998647734956056e-06,
+      "loss": 0.2209,
+      "step": 2409
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.793103448275862e-06,
+      "loss": 0.3909,
+      "step": 2410
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.7863421230561194e-06,
+      "loss": 0.161,
+      "step": 2411
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.7795807978363763e-06,
+      "loss": 0.2018,
+      "step": 2412
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.7728194726166332e-06,
+      "loss": 0.085,
+      "step": 2413
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.7660581473968897e-06,
+      "loss": 0.1028,
+      "step": 2414
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.759296822177147e-06,
+      "loss": 0.4048,
+      "step": 2415
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.752535496957404e-06,
+      "loss": 0.3075,
+      "step": 2416
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.7457741717376613e-06,
+      "loss": 0.7911,
+      "step": 2417
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.7390128465179178e-06,
+      "loss": 0.0511,
+      "step": 2418
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.7322515212981747e-06,
+      "loss": 0.1382,
+      "step": 2419
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.7254901960784316e-06,
+      "loss": 0.0961,
+      "step": 2420
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.718728870858689e-06,
+      "loss": 0.2896,
+      "step": 2421
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.7119675456389454e-06,
+      "loss": 0.2641,
+      "step": 2422
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.7052062204192023e-06,
+      "loss": 0.3371,
+      "step": 2423
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.6984448951994597e-06,
+      "loss": 0.2133,
+      "step": 2424
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.6916835699797166e-06,
+      "loss": 0.0938,
+      "step": 2425
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.684922244759973e-06,
+      "loss": 0.4987,
+      "step": 2426
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.67816091954023e-06,
+      "loss": 0.2357,
+      "step": 2427
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.6713995943204873e-06,
+      "loss": 0.4892,
+      "step": 2428
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.6646382691007442e-06,
+      "loss": 0.1222,
+      "step": 2429
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.6578769438810007e-06,
+      "loss": 0.2003,
+      "step": 2430
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.651115618661258e-06,
+      "loss": 0.5327,
+      "step": 2431
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.644354293441515e-06,
+      "loss": 0.0739,
+      "step": 2432
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.637592968221772e-06,
+      "loss": 0.1823,
+      "step": 2433
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.6308316430020284e-06,
+      "loss": 0.2254,
+      "step": 2434
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.6240703177822857e-06,
+      "loss": 0.0954,
+      "step": 2435
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.6173089925625426e-06,
+      "loss": 0.0694,
+      "step": 2436
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.6105476673427995e-06,
+      "loss": 0.1965,
+      "step": 2437
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.603786342123056e-06,
+      "loss": 0.2137,
+      "step": 2438
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.5970250169033134e-06,
+      "loss": 0.2018,
+      "step": 2439
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.5902636916835703e-06,
+      "loss": 0.1658,
+      "step": 2440
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.5835023664638276e-06,
+      "loss": 0.1195,
+      "step": 2441
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.576741041244084e-06,
+      "loss": 0.0618,
+      "step": 2442
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.569979716024341e-06,
+      "loss": 0.0693,
+      "step": 2443
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.563218390804598e-06,
+      "loss": 0.0808,
+      "step": 2444
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.5564570655848552e-06,
+      "loss": 0.1641,
+      "step": 2445
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.5496957403651117e-06,
+      "loss": 0.1775,
+      "step": 2446
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.5429344151453686e-06,
+      "loss": 0.1813,
+      "step": 2447
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.536173089925626e-06,
+      "loss": 0.2647,
+      "step": 2448
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.2132,
+      "step": 2449
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.5226504394861394e-06,
+      "loss": 0.4781,
+      "step": 2450
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.5158891142663963e-06,
+      "loss": 0.1199,
+      "step": 2451
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.5091277890466536e-06,
+      "loss": 0.146,
+      "step": 2452
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.5023664638269105e-06,
+      "loss": 0.2095,
+      "step": 2453
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.495605138607167e-06,
+      "loss": 0.1344,
+      "step": 2454
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.4888438133874244e-06,
+      "loss": 0.1599,
+      "step": 2455
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.4820824881676813e-06,
+      "loss": 0.1456,
+      "step": 2456
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.475321162947938e-06,
+      "loss": 0.0223,
+      "step": 2457
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.4685598377281947e-06,
+      "loss": 0.2043,
+      "step": 2458
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.461798512508452e-06,
+      "loss": 0.2113,
+      "step": 2459
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.455037187288709e-06,
+      "loss": 0.1625,
+      "step": 2460
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.448275862068966e-06,
+      "loss": 0.0897,
+      "step": 2461
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.4415145368492227e-06,
+      "loss": 0.1754,
+      "step": 2462
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.4347532116294797e-06,
+      "loss": 0.446,
+      "step": 2463
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.4279918864097366e-06,
+      "loss": 0.3264,
+      "step": 2464
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.421230561189994e-06,
+      "loss": 0.011,
+      "step": 2465
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.4144692359702504e-06,
+      "loss": 0.0461,
+      "step": 2466
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.4077079107505073e-06,
+      "loss": 0.1676,
+      "step": 2467
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.400946585530764e-06,
+      "loss": 0.2085,
+      "step": 2468
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3941852603110215e-06,
+      "loss": 0.2035,
+      "step": 2469
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.387423935091278e-06,
+      "loss": 0.2973,
+      "step": 2470
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.380662609871535e-06,
+      "loss": 0.22,
+      "step": 2471
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3739012846517923e-06,
+      "loss": 0.1113,
+      "step": 2472
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.367139959432049e-06,
+      "loss": 0.2452,
+      "step": 2473
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3603786342123057e-06,
+      "loss": 0.5109,
+      "step": 2474
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3536173089925626e-06,
+      "loss": 0.0875,
+      "step": 2475
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.34685598377282e-06,
+      "loss": 0.2075,
+      "step": 2476
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.340094658553077e-06,
+      "loss": 0.299,
+      "step": 2477
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.4789,
+      "step": 2478
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3265720081135907e-06,
+      "loss": 0.3075,
+      "step": 2479
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3198106828938476e-06,
+      "loss": 0.2019,
+      "step": 2480
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3130493576741045e-06,
+      "loss": 0.0684,
+      "step": 2481
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.306288032454361e-06,
+      "loss": 0.1592,
+      "step": 2482
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2995267072346183e-06,
+      "loss": 0.2045,
+      "step": 2483
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2927653820148752e-06,
+      "loss": 0.0719,
+      "step": 2484
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2860040567951326e-06,
+      "loss": 0.14,
+      "step": 2485
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.279242731575389e-06,
+      "loss": 0.0402,
+      "step": 2486
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.272481406355646e-06,
+      "loss": 0.3172,
+      "step": 2487
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.265720081135903e-06,
+      "loss": 0.4953,
+      "step": 2488
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.25895875591616e-06,
+      "loss": 0.0333,
+      "step": 2489
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.2521974306964167e-06,
+      "loss": 0.702,
+      "step": 2490
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2454361054766736e-06,
+      "loss": 0.0581,
+      "step": 2491
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2386747802569305e-06,
+      "loss": 0.2138,
+      "step": 2492
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.231913455037188e-06,
+      "loss": 0.2558,
+      "step": 2493
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2251521298174443e-06,
+      "loss": 0.0204,
+      "step": 2494
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2183908045977012e-06,
+      "loss": 0.2319,
+      "step": 2495
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2116294793779586e-06,
+      "loss": 0.1251,
+      "step": 2496
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2048681541582155e-06,
+      "loss": 0.5158,
+      "step": 2497
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.198106828938472e-06,
+      "loss": 0.3084,
+      "step": 2498
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.191345503718729e-06,
+      "loss": 0.2398,
+      "step": 2499
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.1845841784989862e-06,
+      "loss": 0.0844,
+      "step": 2500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.177822853279243e-06,
+      "loss": 0.514,
+      "step": 2501
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.1710615280594996e-06,
+      "loss": 0.2674,
+      "step": 2502
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.164300202839757e-06,
+      "loss": 0.1681,
+      "step": 2503
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.157538877620014e-06,
+      "loss": 0.7377,
+      "step": 2504
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.1507775524002708e-06,
+      "loss": 0.1492,
+      "step": 2505
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.1440162271805273e-06,
+      "loss": 0.2495,
+      "step": 2506
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.1372549019607846e-06,
+      "loss": 0.1549,
+      "step": 2507
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.1304935767410415e-06,
+      "loss": 0.2691,
+      "step": 2508
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.123732251521299e-06,
+      "loss": 0.0859,
+      "step": 2509
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.1169709263015553e-06,
+      "loss": 0.1873,
+      "step": 2510
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.1102096010818122e-06,
+      "loss": 0.1484,
+      "step": 2511
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.103448275862069e-06,
+      "loss": 0.2294,
+      "step": 2512
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0966869506423265e-06,
+      "loss": 0.1834,
+      "step": 2513
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.089925625422583e-06,
+      "loss": 0.1006,
+      "step": 2514
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.08316430020284e-06,
+      "loss": 0.0416,
+      "step": 2515
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0764029749830972e-06,
+      "loss": 0.1928,
+      "step": 2516
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.069641649763354e-06,
+      "loss": 0.1145,
+      "step": 2517
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0628803245436106e-06,
+      "loss": 0.1563,
+      "step": 2518
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0561189993238675e-06,
+      "loss": 0.3089,
+      "step": 2519
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.049357674104125e-06,
+      "loss": 0.1003,
+      "step": 2520
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0425963488843818e-06,
+      "loss": 0.1388,
+      "step": 2521
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0358350236646383e-06,
+      "loss": 0.1834,
+      "step": 2522
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0290736984448956e-06,
+      "loss": 0.7297,
+      "step": 2523
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0223123732251525e-06,
+      "loss": 0.3564,
+      "step": 2524
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0155510480054094e-06,
+      "loss": 0.1683,
+      "step": 2525
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.008789722785666e-06,
+      "loss": 0.1234,
+      "step": 2526
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0020283975659233e-06,
+      "loss": 0.6587,
+      "step": 2527
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.99526707234618e-06,
+      "loss": 0.3971,
+      "step": 2528
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.988505747126437e-06,
+      "loss": 0.1115,
+      "step": 2529
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.9817444219066936e-06,
+      "loss": 0.0507,
+      "step": 2530
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.974983096686951e-06,
+      "loss": 0.229,
+      "step": 2531
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.968221771467208e-06,
+      "loss": 0.0609,
+      "step": 2532
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.961460446247465e-06,
+      "loss": 0.3545,
+      "step": 2533
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.9546991210277216e-06,
+      "loss": 0.2643,
+      "step": 2534
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.9479377958079785e-06,
+      "loss": 0.2478,
+      "step": 2535
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 0.2415,
+      "step": 2536
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.934415145368493e-06,
+      "loss": 0.5994,
+      "step": 2537
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.9276538201487493e-06,
+      "loss": 0.2946,
+      "step": 2538
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.920892494929006e-06,
+      "loss": 1.938,
+      "step": 2539
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.9141311697092635e-06,
+      "loss": 0.5161,
+      "step": 2540
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.9073698444895204e-06,
+      "loss": 0.136,
+      "step": 2541
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.900608519269777e-06,
+      "loss": 0.2331,
+      "step": 2542
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.893847194050034e-06,
+      "loss": 0.1365,
+      "step": 2543
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.887085868830291e-06,
+      "loss": 0.188,
+      "step": 2544
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.880324543610548e-06,
+      "loss": 0.3726,
+      "step": 2545
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.8735632183908046e-06,
+      "loss": 0.0471,
+      "step": 2546
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.866801893171062e-06,
+      "loss": 0.1048,
+      "step": 2547
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.860040567951319e-06,
+      "loss": 0.0821,
+      "step": 2548
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.8532792427315757e-06,
+      "loss": 0.068,
+      "step": 2549
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.8465179175118322e-06,
+      "loss": 0.1573,
+      "step": 2550
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.8397565922920896e-06,
+      "loss": 0.0866,
+      "step": 2551
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.8329952670723465e-06,
+      "loss": 0.1971,
+      "step": 2552
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.8262339418526034e-06,
+      "loss": 0.14,
+      "step": 2553
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.8194726166328603e-06,
+      "loss": 0.3023,
+      "step": 2554
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.812711291413117e-06,
+      "loss": 0.1916,
+      "step": 2555
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.805949966193374e-06,
+      "loss": 0.0778,
+      "step": 2556
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.7991886409736314e-06,
+      "loss": 0.187,
+      "step": 2557
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.792427315753888e-06,
+      "loss": 0.2451,
+      "step": 2558
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.785665990534145e-06,
+      "loss": 0.1579,
+      "step": 2559
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.7789046653144018e-06,
+      "loss": 0.0934,
+      "step": 2560
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.772143340094659e-06,
+      "loss": 0.1398,
+      "step": 2561
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.7653820148749156e-06,
+      "loss": 0.1232,
+      "step": 2562
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.7586206896551725e-06,
+      "loss": 0.1874,
+      "step": 2563
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.75185936443543e-06,
+      "loss": 0.285,
+      "step": 2564
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.7450980392156867e-06,
+      "loss": 0.39,
+      "step": 2565
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.7383367139959432e-06,
+      "loss": 0.5646,
+      "step": 2566
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.7315753887762e-06,
+      "loss": 0.0847,
+      "step": 2567
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.7248140635564575e-06,
+      "loss": 0.0845,
+      "step": 2568
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.7180527383367144e-06,
+      "loss": 0.1331,
+      "step": 2569
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.711291413116971e-06,
+      "loss": 0.0663,
+      "step": 2570
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.704530087897228e-06,
+      "loss": 0.2318,
+      "step": 2571
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.697768762677485e-06,
+      "loss": 0.069,
+      "step": 2572
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.691007437457742e-06,
+      "loss": 0.2944,
+      "step": 2573
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.6842461122379985e-06,
+      "loss": 0.19,
+      "step": 2574
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.677484787018256e-06,
+      "loss": 0.2301,
+      "step": 2575
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.6707234617985128e-06,
+      "loss": 0.2933,
+      "step": 2576
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.66396213657877e-06,
+      "loss": 0.2313,
+      "step": 2577
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.6572008113590266e-06,
+      "loss": 0.1262,
+      "step": 2578
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.6504394861392835e-06,
+      "loss": 0.3871,
+      "step": 2579
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.6436781609195404e-06,
+      "loss": 0.465,
+      "step": 2580
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.6369168356997977e-06,
+      "loss": 0.1104,
+      "step": 2581
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.6301555104800542e-06,
+      "loss": 0.9148,
+      "step": 2582
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.623394185260311e-06,
+      "loss": 0.176,
+      "step": 2583
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.616632860040568e-06,
+      "loss": 0.2213,
+      "step": 2584
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.6098715348208254e-06,
+      "loss": 0.1726,
+      "step": 2585
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.603110209601082e-06,
+      "loss": 0.2572,
+      "step": 2586
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.596348884381339e-06,
+      "loss": 0.3024,
+      "step": 2587
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.589587559161596e-06,
+      "loss": 0.1246,
+      "step": 2588
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.582826233941853e-06,
+      "loss": 0.1358,
+      "step": 2589
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.5760649087221095e-06,
+      "loss": 0.0801,
+      "step": 2590
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.5693035835023664e-06,
+      "loss": 0.1354,
+      "step": 2591
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.5625422582826238e-06,
+      "loss": 0.1996,
+      "step": 2592
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.5557809330628807e-06,
+      "loss": 0.2828,
+      "step": 2593
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.549019607843137e-06,
+      "loss": 0.1328,
+      "step": 2594
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.5422582826233945e-06,
+      "loss": 0.1037,
+      "step": 2595
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.5354969574036514e-06,
+      "loss": 0.1132,
+      "step": 2596
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.5287356321839083e-06,
+      "loss": 0.0329,
+      "step": 2597
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.521974306964165e-06,
+      "loss": 0.2301,
+      "step": 2598
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.515212981744422e-06,
+      "loss": 0.1087,
+      "step": 2599
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.508451656524679e-06,
+      "loss": 0.1384,
+      "step": 2600
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.5016903313049364e-06,
+      "loss": 0.0341,
+      "step": 2601
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.494929006085193e-06,
+      "loss": 0.2041,
+      "step": 2602
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.48816768086545e-06,
+      "loss": 0.0889,
+      "step": 2603
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.4814063556457067e-06,
+      "loss": 0.1949,
+      "step": 2604
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.4746450304259636e-06,
+      "loss": 0.313,
+      "step": 2605
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.4678837052062205e-06,
+      "loss": 0.0368,
+      "step": 2606
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.4611223799864774e-06,
+      "loss": 0.209,
+      "step": 2607
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.4543610547667348e-06,
+      "loss": 0.0841,
+      "step": 2608
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.4475997295469913e-06,
+      "loss": 0.4289,
+      "step": 2609
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.4408384043272486e-06,
+      "loss": 0.5404,
+      "step": 2610
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.434077079107505e-06,
+      "loss": 0.1333,
+      "step": 2611
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.4273157538877624e-06,
+      "loss": 0.1919,
+      "step": 2612
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.420554428668019e-06,
+      "loss": 0.2493,
+      "step": 2613
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.4137931034482762e-06,
+      "loss": 0.0566,
+      "step": 2614
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.407031778228533e-06,
+      "loss": 0.041,
+      "step": 2615
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.40027045300879e-06,
+      "loss": 0.1644,
+      "step": 2616
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.393509127789047e-06,
+      "loss": 0.1791,
+      "step": 2617
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.386747802569304e-06,
+      "loss": 0.0958,
+      "step": 2618
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.379986477349561e-06,
+      "loss": 0.2367,
+      "step": 2619
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.3732251521298177e-06,
+      "loss": 0.0132,
+      "step": 2620
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.3664638269100746e-06,
+      "loss": 0.2665,
+      "step": 2621
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.3597025016903315e-06,
+      "loss": 0.5098,
+      "step": 2622
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.6287,
+      "step": 2623
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.3461798512508454e-06,
+      "loss": 0.4891,
+      "step": 2624
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.3394185260311023e-06,
+      "loss": 0.4314,
+      "step": 2625
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.332657200811359e-06,
+      "loss": 0.2402,
+      "step": 2626
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.325895875591616e-06,
+      "loss": 0.1516,
+      "step": 2627
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.319134550371873e-06,
+      "loss": 0.1503,
+      "step": 2628
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.31237322515213e-06,
+      "loss": 0.3556,
+      "step": 2629
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.305611899932387e-06,
+      "loss": 0.1381,
+      "step": 2630
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.2988505747126437e-06,
+      "loss": 0.211,
+      "step": 2631
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.292089249492901e-06,
+      "loss": 0.426,
+      "step": 2632
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.2853279242731576e-06,
+      "loss": 0.0597,
+      "step": 2633
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.278566599053415e-06,
+      "loss": 0.059,
+      "step": 2634
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.2718052738336714e-06,
+      "loss": 1.5977,
+      "step": 2635
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.2650439486139287e-06,
+      "loss": 0.3281,
+      "step": 2636
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.2582826233941852e-06,
+      "loss": 0.1169,
+      "step": 2637
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.2515212981744425e-06,
+      "loss": 0.0496,
+      "step": 2638
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.2447599729546995e-06,
+      "loss": 0.0409,
+      "step": 2639
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.2379986477349564e-06,
+      "loss": 0.2367,
+      "step": 2640
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.2312373225152133e-06,
+      "loss": 0.0995,
+      "step": 2641
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.22447599729547e-06,
+      "loss": 0.1342,
+      "step": 2642
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.217714672075727e-06,
+      "loss": 0.2568,
+      "step": 2643
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.210953346855984e-06,
+      "loss": 0.5336,
+      "step": 2644
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.204192021636241e-06,
+      "loss": 0.1947,
+      "step": 2645
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.197430696416498e-06,
+      "loss": 0.5152,
+      "step": 2646
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.1906693711967548e-06,
+      "loss": 0.2156,
+      "step": 2647
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.1839080459770117e-06,
+      "loss": 0.1499,
+      "step": 2648
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.1771467207572686e-06,
+      "loss": 0.0872,
+      "step": 2649
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.1703853955375255e-06,
+      "loss": 0.1228,
+      "step": 2650
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.1636240703177824e-06,
+      "loss": 0.9983,
+      "step": 2651
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.1568627450980393e-06,
+      "loss": 0.0787,
+      "step": 2652
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.1501014198782962e-06,
+      "loss": 0.1091,
+      "step": 2653
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.1433400946585536e-06,
+      "loss": 0.0804,
+      "step": 2654
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.13657876943881e-06,
+      "loss": 0.0607,
+      "step": 2655
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.1298174442190674e-06,
+      "loss": 0.4617,
+      "step": 2656
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.123056118999324e-06,
+      "loss": 0.0136,
+      "step": 2657
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.116294793779581e-06,
+      "loss": 0.6834,
+      "step": 2658
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.1095334685598377e-06,
+      "loss": 0.1484,
+      "step": 2659
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.102772143340095e-06,
+      "loss": 0.4205,
+      "step": 2660
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.096010818120352e-06,
+      "loss": 1.3661,
+      "step": 2661
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.089249492900609e-06,
+      "loss": 0.1402,
+      "step": 2662
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.0824881676808658e-06,
+      "loss": 0.5285,
+      "step": 2663
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.0757268424611227e-06,
+      "loss": 0.1287,
+      "step": 2664
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.0689655172413796e-06,
+      "loss": 0.0619,
+      "step": 2665
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.0622041920216365e-06,
+      "loss": 0.0923,
+      "step": 2666
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.0554428668018934e-06,
+      "loss": 0.1489,
+      "step": 2667
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.0486815415821503e-06,
+      "loss": 0.3031,
+      "step": 2668
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.0419202163624072e-06,
+      "loss": 0.1496,
+      "step": 2669
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.035158891142664e-06,
+      "loss": 0.2982,
+      "step": 2670
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.028397565922921e-06,
+      "loss": 0.0504,
+      "step": 2671
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.021636240703178e-06,
+      "loss": 0.2897,
+      "step": 2672
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.014874915483435e-06,
+      "loss": 0.1833,
+      "step": 2673
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.0081135902636918e-06,
+      "loss": 0.1145,
+      "step": 2674
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.0013522650439487e-06,
+      "loss": 0.1555,
+      "step": 2675
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9945909398242056e-06,
+      "loss": 0.1054,
+      "step": 2676
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9878296146044625e-06,
+      "loss": 0.0745,
+      "step": 2677
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.98106828938472e-06,
+      "loss": 0.515,
+      "step": 2678
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9743069641649763e-06,
+      "loss": 0.0979,
+      "step": 2679
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9675456389452337e-06,
+      "loss": 0.1546,
+      "step": 2680
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.96078431372549e-06,
+      "loss": 0.3031,
+      "step": 2681
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9540229885057475e-06,
+      "loss": 0.0237,
+      "step": 2682
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.947261663286004e-06,
+      "loss": 0.0659,
+      "step": 2683
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9405003380662613e-06,
+      "loss": 0.1363,
+      "step": 2684
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9337390128465182e-06,
+      "loss": 0.6543,
+      "step": 2685
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.926977687626775e-06,
+      "loss": 0.268,
+      "step": 2686
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.920216362407032e-06,
+      "loss": 0.0799,
+      "step": 2687
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.913455037187289e-06,
+      "loss": 0.2842,
+      "step": 2688
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9066937119675457e-06,
+      "loss": 0.167,
+      "step": 2689
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8999323867478028e-06,
+      "loss": 0.5386,
+      "step": 2690
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8931710615280597e-06,
+      "loss": 0.1114,
+      "step": 2691
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8864097363083166e-06,
+      "loss": 0.1515,
+      "step": 2692
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8796484110885735e-06,
+      "loss": 0.0547,
+      "step": 2693
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8728870858688306e-06,
+      "loss": 0.4493,
+      "step": 2694
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8661257606490873e-06,
+      "loss": 0.0544,
+      "step": 2695
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8593644354293445e-06,
+      "loss": 0.1666,
+      "step": 2696
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8526031102096012e-06,
+      "loss": 0.0592,
+      "step": 2697
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.8458417849898583e-06,
+      "loss": 0.714,
+      "step": 2698
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.839080459770115e-06,
+      "loss": 0.1027,
+      "step": 2699
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.8323191345503721e-06,
+      "loss": 0.0914,
+      "step": 2700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.825557809330629e-06,
+      "loss": 0.2015,
+      "step": 2701
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.818796484110886e-06,
+      "loss": 0.1151,
+      "step": 2702
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.8120351588911429e-06,
+      "loss": 0.1114,
+      "step": 2703
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.8052738336713998e-06,
+      "loss": 0.0166,
+      "step": 2704
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7985125084516567e-06,
+      "loss": 0.0222,
+      "step": 2705
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7917511832319138e-06,
+      "loss": 0.122,
+      "step": 2706
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7849898580121705e-06,
+      "loss": 0.1821,
+      "step": 2707
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7782285327924276e-06,
+      "loss": 0.1822,
+      "step": 2708
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7714672075726843e-06,
+      "loss": 0.1185,
+      "step": 2709
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7647058823529414e-06,
+      "loss": 0.0912,
+      "step": 2710
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7579445571331981e-06,
+      "loss": 0.0677,
+      "step": 2711
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7511832319134553e-06,
+      "loss": 0.1542,
+      "step": 2712
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7444219066937122e-06,
+      "loss": 0.0605,
+      "step": 2713
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.737660581473969e-06,
+      "loss": 0.0662,
+      "step": 2714
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.730899256254226e-06,
+      "loss": 0.0513,
+      "step": 2715
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.724137931034483e-06,
+      "loss": 0.1314,
+      "step": 2716
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7173766058147398e-06,
+      "loss": 1.6669,
+      "step": 2717
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.710615280594997e-06,
+      "loss": 0.118,
+      "step": 2718
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7038539553752536e-06,
+      "loss": 0.2583,
+      "step": 2719
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6970926301555108e-06,
+      "loss": 0.0754,
+      "step": 2720
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6903313049357675e-06,
+      "loss": 0.2993,
+      "step": 2721
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6835699797160246e-06,
+      "loss": 0.3076,
+      "step": 2722
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6768086544962813e-06,
+      "loss": 0.267,
+      "step": 2723
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6700473292765384e-06,
+      "loss": 0.4542,
+      "step": 2724
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6632860040567953e-06,
+      "loss": 0.1519,
+      "step": 2725
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6565246788370522e-06,
+      "loss": 0.0649,
+      "step": 2726
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6497633536173092e-06,
+      "loss": 0.0445,
+      "step": 2727
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6430020283975663e-06,
+      "loss": 0.1501,
+      "step": 2728
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.636240703177823e-06,
+      "loss": 0.1025,
+      "step": 2729
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.62947937795808e-06,
+      "loss": 0.0752,
+      "step": 2730
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6227180527383368e-06,
+      "loss": 0.1127,
+      "step": 2731
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.615956727518594e-06,
+      "loss": 0.0604,
+      "step": 2732
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6091954022988506e-06,
+      "loss": 0.0,
+      "step": 2733
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6024340770791077e-06,
+      "loss": 0.1722,
+      "step": 2734
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5956727518593644e-06,
+      "loss": 0.7458,
+      "step": 2735
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5889114266396216e-06,
+      "loss": 0.2672,
+      "step": 2736
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5821501014198785e-06,
+      "loss": 0.1036,
+      "step": 2737
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5753887762001354e-06,
+      "loss": 0.4565,
+      "step": 2738
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5686274509803923e-06,
+      "loss": 0.1085,
+      "step": 2739
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5618661257606494e-06,
+      "loss": 0.0397,
+      "step": 2740
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5551048005409061e-06,
+      "loss": 0.0554,
+      "step": 2741
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5483434753211632e-06,
+      "loss": 0.1069,
+      "step": 2742
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.54158215010142e-06,
+      "loss": 0.0392,
+      "step": 2743
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.534820824881677e-06,
+      "loss": 0.231,
+      "step": 2744
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5280594996619338e-06,
+      "loss": 0.0477,
+      "step": 2745
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5212981744421909e-06,
+      "loss": 0.3182,
+      "step": 2746
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5145368492224478e-06,
+      "loss": 0.6885,
+      "step": 2747
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5077755240027047e-06,
+      "loss": 0.0683,
+      "step": 2748
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5010141987829616e-06,
+      "loss": 0.0563,
+      "step": 2749
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4942528735632185e-06,
+      "loss": 0.5395,
+      "step": 2750
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4874915483434755e-06,
+      "loss": 0.05,
+      "step": 2751
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4807302231237326e-06,
+      "loss": 0.0625,
+      "step": 2752
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4739688979039893e-06,
+      "loss": 0.1554,
+      "step": 2753
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4672075726842464e-06,
+      "loss": 0.3897,
+      "step": 2754
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.460446247464503e-06,
+      "loss": 0.9572,
+      "step": 2755
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4536849222447602e-06,
+      "loss": 0.0511,
+      "step": 2756
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.446923597025017e-06,
+      "loss": 0.2678,
+      "step": 2757
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.440162271805274e-06,
+      "loss": 0.2848,
+      "step": 2758
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.433400946585531e-06,
+      "loss": 0.7145,
+      "step": 2759
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4266396213657879e-06,
+      "loss": 0.2064,
+      "step": 2760
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4198782961460448e-06,
+      "loss": 0.0608,
+      "step": 2761
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4131169709263017e-06,
+      "loss": 0.0475,
+      "step": 2762
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4063556457065586e-06,
+      "loss": 0.1062,
+      "step": 2763
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3995943204868157e-06,
+      "loss": 0.1985,
+      "step": 2764
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3928329952670724e-06,
+      "loss": 0.1283,
+      "step": 2765
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3860716700473295e-06,
+      "loss": 0.0866,
+      "step": 2766
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3793103448275862e-06,
+      "loss": 0.1703,
+      "step": 2767
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3725490196078434e-06,
+      "loss": 0.184,
+      "step": 2768
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3657876943881e-06,
+      "loss": 0.145,
+      "step": 2769
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.3590263691683572e-06,
+      "loss": 0.131,
+      "step": 2770
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.352265043948614e-06,
+      "loss": 0.4455,
+      "step": 2771
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.345503718728871e-06,
+      "loss": 0.4246,
+      "step": 2772
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.338742393509128e-06,
+      "loss": 0.097,
+      "step": 2773
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.331981068289385e-06,
+      "loss": 0.0471,
+      "step": 2774
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.3252197430696418e-06,
+      "loss": 0.1965,
+      "step": 2775
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.3184584178498989e-06,
+      "loss": 0.0177,
+      "step": 2776
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.3116970926301556e-06,
+      "loss": 0.0676,
+      "step": 2777
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.3049357674104127e-06,
+      "loss": 0.0588,
+      "step": 2778
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2981744421906694e-06,
+      "loss": 0.1566,
+      "step": 2779
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2914131169709265e-06,
+      "loss": 0.0836,
+      "step": 2780
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2846517917511832e-06,
+      "loss": 0.1801,
+      "step": 2781
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2778904665314403e-06,
+      "loss": 0.7818,
+      "step": 2782
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2711291413116973e-06,
+      "loss": 0.0354,
+      "step": 2783
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2643678160919542e-06,
+      "loss": 0.3774,
+      "step": 2784
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.257606490872211e-06,
+      "loss": 0.3304,
+      "step": 2785
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2508451656524682e-06,
+      "loss": 0.2307,
+      "step": 2786
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.244083840432725e-06,
+      "loss": 0.3709,
+      "step": 2787
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2373225152129818e-06,
+      "loss": 0.0754,
+      "step": 2788
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2305611899932387e-06,
+      "loss": 0.4697,
+      "step": 2789
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2237998647734956e-06,
+      "loss": 0.133,
+      "step": 2790
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2170385395537525e-06,
+      "loss": 0.0649,
+      "step": 2791
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2102772143340095e-06,
+      "loss": 0.0985,
+      "step": 2792
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2035158891142666e-06,
+      "loss": 0.0712,
+      "step": 2793
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1967545638945235e-06,
+      "loss": 0.1377,
+      "step": 2794
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1899932386747804e-06,
+      "loss": 0.5879,
+      "step": 2795
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1832319134550373e-06,
+      "loss": 0.1382,
+      "step": 2796
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.0907,
+      "step": 2797
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1697092630155511e-06,
+      "loss": 0.1404,
+      "step": 2798
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.162947937795808e-06,
+      "loss": 0.0843,
+      "step": 2799
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.156186612576065e-06,
+      "loss": 0.2193,
+      "step": 2800
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1494252873563219e-06,
+      "loss": 0.0421,
+      "step": 2801
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1426639621365788e-06,
+      "loss": 0.1244,
+      "step": 2802
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1359026369168357e-06,
+      "loss": 0.131,
+      "step": 2803
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1291413116970926e-06,
+      "loss": 0.0922,
+      "step": 2804
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1223799864773497e-06,
+      "loss": 0.1082,
+      "step": 2805
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1156186612576066e-06,
+      "loss": 0.5936,
+      "step": 2806
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1088573360378636e-06,
+      "loss": 0.5254,
+      "step": 2807
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.1020960108181205e-06,
+      "loss": 0.3425,
+      "step": 2808
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.0953346855983774e-06,
+      "loss": 0.0993,
+      "step": 2809
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.0885733603786343e-06,
+      "loss": 0.0638,
+      "step": 2810
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.0818120351588912e-06,
+      "loss": 0.119,
+      "step": 2811
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.0750507099391481e-06,
+      "loss": 0.4552,
+      "step": 2812
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.068289384719405e-06,
+      "loss": 0.0861,
+      "step": 2813
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.061528059499662e-06,
+      "loss": 0.6719,
+      "step": 2814
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.0547667342799188e-06,
+      "loss": 0.0803,
+      "step": 2815
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.048005409060176e-06,
+      "loss": 0.1072,
+      "step": 2816
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0412440838404329e-06,
+      "loss": 0.2295,
+      "step": 2817
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0344827586206898e-06,
+      "loss": 0.1051,
+      "step": 2818
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0277214334009467e-06,
+      "loss": 0.0449,
+      "step": 2819
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0209601081812036e-06,
+      "loss": 0.0489,
+      "step": 2820
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0141987829614605e-06,
+      "loss": 0.1326,
+      "step": 2821
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0074374577417174e-06,
+      "loss": 0.1212,
+      "step": 2822
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.0006761325219743e-06,
+      "loss": 0.1105,
+      "step": 2823
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.939148073022313e-07,
+      "loss": 0.0597,
+      "step": 2824
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.871534820824882e-07,
+      "loss": 0.0363,
+      "step": 2825
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.80392156862745e-07,
+      "loss": 0.5468,
+      "step": 2826
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.73630831643002e-07,
+      "loss": 0.3014,
+      "step": 2827
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.668695064232591e-07,
+      "loss": 0.1866,
+      "step": 2828
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.60108181203516e-07,
+      "loss": 0.0404,
+      "step": 2829
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.533468559837728e-07,
+      "loss": 0.5578,
+      "step": 2830
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.465855307640299e-07,
+      "loss": 0.1628,
+      "step": 2831
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.398242055442868e-07,
+      "loss": 0.0313,
+      "step": 2832
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.330628803245437e-07,
+      "loss": 0.0178,
+      "step": 2833
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.263015551048006e-07,
+      "loss": 0.0663,
+      "step": 2834
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.195402298850575e-07,
+      "loss": 0.1582,
+      "step": 2835
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.127789046653145e-07,
+      "loss": 0.18,
+      "step": 2836
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.060175794455714e-07,
+      "loss": 0.3728,
+      "step": 2837
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.992562542258283e-07,
+      "loss": 0.5763,
+      "step": 2838
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.924949290060852e-07,
+      "loss": 0.0798,
+      "step": 2839
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.857336037863422e-07,
+      "loss": 0.0944,
+      "step": 2840
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.789722785665991e-07,
+      "loss": 0.065,
+      "step": 2841
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.722109533468561e-07,
+      "loss": 0.2428,
+      "step": 2842
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.65449628127113e-07,
+      "loss": 0.945,
+      "step": 2843
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.586883029073699e-07,
+      "loss": 0.4011,
+      "step": 2844
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.519269776876268e-07,
+      "loss": 0.0652,
+      "step": 2845
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.451656524678837e-07,
+      "loss": 0.0416,
+      "step": 2846
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.384043272481406e-07,
+      "loss": 0.1618,
+      "step": 2847
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.316430020283977e-07,
+      "loss": 0.0951,
+      "step": 2848
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.248816768086546e-07,
+      "loss": 0.5694,
+      "step": 2849
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.181203515889115e-07,
+      "loss": 0.2808,
+      "step": 2850
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.113590263691684e-07,
+      "loss": 0.1895,
+      "step": 2851
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.045977011494253e-07,
+      "loss": 0.1603,
+      "step": 2852
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.978363759296822e-07,
+      "loss": 0.0553,
+      "step": 2853
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.910750507099392e-07,
+      "loss": 0.0514,
+      "step": 2854
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.843137254901962e-07,
+      "loss": 0.4147,
+      "step": 2855
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.775524002704531e-07,
+      "loss": 0.1733,
+      "step": 2856
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.7079107505071e-07,
+      "loss": 0.0311,
+      "step": 2857
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.640297498309669e-07,
+      "loss": 0.0391,
+      "step": 2858
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.572684246112239e-07,
+      "loss": 0.5954,
+      "step": 2859
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.505070993914808e-07,
+      "loss": 0.039,
+      "step": 2860
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.437457741717377e-07,
+      "loss": 0.0468,
+      "step": 2861
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.369844489519946e-07,
+      "loss": 0.167,
+      "step": 2862
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.302231237322515e-07,
+      "loss": 0.0839,
+      "step": 2863
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.234617985125085e-07,
+      "loss": 0.081,
+      "step": 2864
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.167004732927655e-07,
+      "loss": 0.0583,
+      "step": 2865
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.099391480730224e-07,
+      "loss": 0.0491,
+      "step": 2866
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 7.031778228532793e-07,
+      "loss": 0.0729,
+      "step": 2867
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.964164976335362e-07,
+      "loss": 0.1701,
+      "step": 2868
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.896551724137931e-07,
+      "loss": 0.0912,
+      "step": 2869
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.8289384719405e-07,
+      "loss": 0.0133,
+      "step": 2870
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.76132521974307e-07,
+      "loss": 0.2472,
+      "step": 2871
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.69371196754564e-07,
+      "loss": 0.1866,
+      "step": 2872
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.626098715348209e-07,
+      "loss": 0.0566,
+      "step": 2873
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.558485463150778e-07,
+      "loss": 0.0675,
+      "step": 2874
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.490872210953347e-07,
+      "loss": 0.092,
+      "step": 2875
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.423258958755916e-07,
+      "loss": 0.1708,
+      "step": 2876
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.355645706558486e-07,
+      "loss": 0.0407,
+      "step": 2877
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.288032454361055e-07,
+      "loss": 0.1254,
+      "step": 2878
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.220419202163624e-07,
+      "loss": 0.0818,
+      "step": 2879
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.152805949966194e-07,
+      "loss": 0.0384,
+      "step": 2880
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.085192697768763e-07,
+      "loss": 0.7144,
+      "step": 2881
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.017579445571333e-07,
+      "loss": 0.0727,
+      "step": 2882
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.949966193373902e-07,
+      "loss": 0.0618,
+      "step": 2883
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 0.0737,
+      "step": 2884
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.81473968897904e-07,
+      "loss": 0.1035,
+      "step": 2885
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.747126436781609e-07,
+      "loss": 0.6832,
+      "step": 2886
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.679513184584178e-07,
+      "loss": 0.2962,
+      "step": 2887
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.611899932386749e-07,
+      "loss": 0.1293,
+      "step": 2888
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.544286680189318e-07,
+      "loss": 0.1287,
+      "step": 2889
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.476673427991887e-07,
+      "loss": 0.0532,
+      "step": 2890
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.409060175794456e-07,
+      "loss": 0.0197,
+      "step": 2891
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.341446923597025e-07,
+      "loss": 0.1217,
+      "step": 2892
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.273833671399594e-07,
+      "loss": 0.0499,
+      "step": 2893
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.206220419202164e-07,
+      "loss": 0.0718,
+      "step": 2894
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.138607167004734e-07,
+      "loss": 0.1107,
+      "step": 2895
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.070993914807303e-07,
+      "loss": 0.1572,
+      "step": 2896
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.003380662609872e-07,
+      "loss": 0.106,
+      "step": 2897
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.935767410412441e-07,
+      "loss": 0.0466,
+      "step": 2898
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.86815415821501e-07,
+      "loss": 0.1326,
+      "step": 2899
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.80054090601758e-07,
+      "loss": 0.0397,
+      "step": 2900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.732927653820149e-07,
+      "loss": 0.0427,
+      "step": 2901
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.6653144016227184e-07,
+      "loss": 0.0586,
+      "step": 2902
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.5977011494252875e-07,
+      "loss": 0.0615,
+      "step": 2903
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.530087897227857e-07,
+      "loss": 0.0499,
+      "step": 2904
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.462474645030426e-07,
+      "loss": 0.0567,
+      "step": 2905
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.3948613928329954e-07,
+      "loss": 0.077,
+      "step": 2906
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.327248140635565e-07,
+      "loss": 0.206,
+      "step": 2907
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.259634888438134e-07,
+      "loss": 0.0849,
+      "step": 2908
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.192021636240703e-07,
+      "loss": 0.1327,
+      "step": 2909
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.124408384043273e-07,
+      "loss": 0.404,
+      "step": 2910
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.056795131845842e-07,
+      "loss": 0.1235,
+      "step": 2911
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.989181879648411e-07,
+      "loss": 0.0361,
+      "step": 2912
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.921568627450981e-07,
+      "loss": 0.0176,
+      "step": 2913
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.85395537525355e-07,
+      "loss": 0.0784,
+      "step": 2914
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.7863421230561195e-07,
+      "loss": 0.0496,
+      "step": 2915
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.7187288708586886e-07,
+      "loss": 0.0738,
+      "step": 2916
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.651115618661258e-07,
+      "loss": 0.1489,
+      "step": 2917
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.5835023664638274e-07,
+      "loss": 0.1446,
+      "step": 2918
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.5158891142663965e-07,
+      "loss": 0.0718,
+      "step": 2919
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.4482758620689656e-07,
+      "loss": 0.8237,
+      "step": 2920
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.380662609871535e-07,
+      "loss": 0.0737,
+      "step": 2921
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.3130493576741044e-07,
+      "loss": 0.0407,
+      "step": 2922
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2454361054766735e-07,
+      "loss": 0.5541,
+      "step": 2923
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.177822853279243e-07,
+      "loss": 0.0754,
+      "step": 2924
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.110209601081812e-07,
+      "loss": 0.6193,
+      "step": 2925
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0425963488843814e-07,
+      "loss": 0.0798,
+      "step": 2926
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.974983096686951e-07,
+      "loss": 0.1588,
+      "step": 2927
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.90736984448952e-07,
+      "loss": 0.4733,
+      "step": 2928
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.839756592292089e-07,
+      "loss": 0.209,
+      "step": 2929
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.772143340094659e-07,
+      "loss": 0.1059,
+      "step": 2930
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.704530087897228e-07,
+      "loss": 0.1383,
+      "step": 2931
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.636916835699797e-07,
+      "loss": 0.1438,
+      "step": 2932
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.569303583502367e-07,
+      "loss": 0.1083,
+      "step": 2933
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.501690331304936e-07,
+      "loss": 0.0707,
+      "step": 2934
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.434077079107505e-07,
+      "loss": 0.0254,
+      "step": 2935
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.3664638269100746e-07,
+      "loss": 0.2612,
+      "step": 2936
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.2988505747126437e-07,
+      "loss": 0.1135,
+      "step": 2937
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.231237322515213e-07,
+      "loss": 0.9263,
+      "step": 2938
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1636240703177825e-07,
+      "loss": 0.6665,
+      "step": 2939
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0960108181203516e-07,
+      "loss": 0.0457,
+      "step": 2940
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.028397565922921e-07,
+      "loss": 0.1264,
+      "step": 2941
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.9607843137254904e-07,
+      "loss": 0.4868,
+      "step": 2942
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8931710615280598e-07,
+      "loss": 0.1339,
+      "step": 2943
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.825557809330629e-07,
+      "loss": 0.6087,
+      "step": 2944
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7579445571331983e-07,
+      "loss": 0.0775,
+      "step": 2945
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.6903313049357676e-07,
+      "loss": 0.0102,
+      "step": 2946
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.6227180527383367e-07,
+      "loss": 0.4038,
+      "step": 2947
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.555104800540906e-07,
+      "loss": 0.0519,
+      "step": 2948
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.4874915483434755e-07,
+      "loss": 0.0377,
+      "step": 2949
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.4198782961460446e-07,
+      "loss": 0.092,
+      "step": 2950
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.352265043948614e-07,
+      "loss": 0.0783,
+      "step": 2951
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2846517917511834e-07,
+      "loss": 0.1108,
+      "step": 2952
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2170385395537525e-07,
+      "loss": 0.6081,
+      "step": 2953
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.1494252873563219e-07,
+      "loss": 0.0586,
+      "step": 2954
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0818120351588913e-07,
+      "loss": 0.0997,
+      "step": 2955
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0141987829614605e-07,
+      "loss": 0.3301,
+      "step": 2956
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.465855307640299e-08,
+      "loss": 0.0777,
+      "step": 2957
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.789722785665991e-08,
+      "loss": 0.2532,
+      "step": 2958
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.113590263691684e-08,
+      "loss": 0.141,
+      "step": 2959
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.437457741717378e-08,
+      "loss": 0.032,
+      "step": 2960
+    },
+    {
+      "epoch": 4.0,
+      "step": 2960,
+      "total_flos": 6.543141246861312e+16,
+      "train_loss": 0.928271904115477,
+      "train_runtime": 7432.1142,
+      "train_samples_per_second": 0.398,
+      "train_steps_per_second": 0.398
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2960,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 100,
+  "total_flos": 6.543141246861312e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}